Multimodal Quantitative Imaging in a Canine Model of Osteoarthritis

Home , Arthrotomy

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Maria Isabel Menendez

Graduate Program in Comparative and Veterinary Medicine

The Ohio State University

2015

Dissertation/Thesis Committee:

Michael V. Knopp, Advisor

Michael F. Tweedle

Thomas J. Rosol

David C. Flanigan

Copyrighted by

Maria Isabel Menendez

2015

Abstract

Osteoarthritis (OA) of the knee is a major public health problem that primarily affects the elderly. Almost 10% of the U. S. population suffers from symptomatic knee OA by the age of 60. There are no approved interventions that ameliorate structural progression of this disorder. The increasing importance of imaging in animal models of osteoarthritis for diagnosis, prognostication, and follow-up is of paramount importance and plays a crucial role in increasing our understanding of the etiology of OA and in the development of new therapies. A primary aim of this study was to provide a comprehensive imaging analysis of the whole knee joint serially in a surgically induced in vivo canine model of OA. We elucidated that quantitative magnetic resonance imaging (MRI) markers demonstrated early changes in the cartilage of the knees that underwent anterior cruciate ligament transection (ACLT) relative to the control knee. This study provided evidence that T2 mapping and delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) are imaging markers relevant to the initiation and progression of OA. Conventional radiography knee assessment, the gold standard in OA diagnosis showed OA signs at a later stage of OA, lacking evidence of premature signs of OA. Serial in vivo imaging utilizing 2-deoxy-2-

[fluorine-18] fluoro- D-glucose (18F-FDG) and sodium 18 F-fluoride (18F-NaF) Positron

Emission Tomography /Computed Tomography (PET/CT) were performed to characterize knee metabolic and remodeling activity. PET was co-registered with MRI to allow us to improve the location of the regions of interest, otherwise unattainable with PET alone. This work demonstrated, providing imaging evidence, that 18F-FDG and 18F-NaF served an

ii important role in detecting early OA metabolic and remodeling changes in the knee prior to the expression of gross changes. These in vivo changes, in addition to ex vivo micro-

PET/CT using 18F-NaF and histomorphometry assessment provided a more valuable understanding of OA. Radiography in combination with clinical imaging technologies, such as, MRI, PET and microcomputed tomography (μCT) produced multimodal imaging techniques that allowed to merge molecular, functional, and anatomical data. These technologies provide a more precise and rigorous methods for exploring OA animal models in greater depth. Collectively, these findings can be interpreted as strong evidence that imaging markers play an important role in post-traumatic OA and that these markers, aimed to detect early signs in OA, may be used clinically to diagnose and follow up therapy treatments in OA.

iii

Dedication

Dedicated to my parents, Justin Scott, Michael James and Henry Edward Williams and

Natalia

Acknowledgments

Dr Menendez was supported by The Wright Center of Innovation in Biomedical Imaging,

Department of Radiology at The Ohio State University Wexner Medical Center. A portion of this work was supported by canine research funds provided by the College of Veterinary

Medicine at The Ohio State University. The authors would like to thank Drs. Bianca

Hettlich, Kristin Lewis, Steven Weisbrode, Lai Wei, Karen Briley, Amir Abduljalil, Daniel

Clark, Katherine Binzel, Jun Zhang, Wenbo Wei and Timothy Vojt for professional assistance. I would like to personally thank the five Beagles and five ponies; which allowed us to complete this research and advance our knowledge in osteoarthritis. Special consideration is given to the members of Dr. Menendez’s graduate committee for their constructive comments and support during the course of this work. Finally, Dr. Michael

Knopp receives highest appreciation for his role as advisor to Dr. Menendez, and for setting the standard of an outstanding mentor.

Vita

2002………………………………………….Doctor of Veterinary Medicine, Leon

University, Spain

2010 to present………………………………Graduate Research Associate, Department

of Veterinary Clinical Sciences, The Ohio

State University

Publications

1. Menendez MI, Ishihara A, Weisbrode S, Bertone A. Radiofrequency energy on Cortical

bone and Soft Tissue: a Pilot Study; Clinical Orthopaedics and Related Research. October

2010 Apr; 137(4):890-7.

2. Menendez MI, Clark DJ, Carlton M, Flanigan DC, Jia G, Sammet S, Knopp MV, Bertone

AL. Direct Human Adenoviral BMP-2 or BMP-6 Gene Therapy for Bone and Cartilage

Regeneration in a Pony Osteochondral Model. Osteoarthritis and Cartilage. 2011

Aug;19(8):1066-75.

3. Jennifer A. Dulin, Wm T. Drost, Mitch A. Phelps, Elizabeth M. Santschi, Maria I.

Menendez, Alicia L. Bertone. Influence of Exercise on the Distribution of a

Radiopharmaceutical (99mTechnetium-Methylene Diphosphonate) Following Intra-

Articular Injection in Horses. Am J Vet Res. 2012 Mar; 73(3):418-25.

4. Menendez MI, Phelps MA, Hothem EA, Bertone AL. Pharmacokinetics of

methylprednisolone acetate after intra-articular administration and subsequent suppression

of endogenous hydrocortisone secretion in exercising horses. Am J Vet Res. 2012 Sep;

73(9):1453-61.

5. Hayam Hussein, Akikazu Ishihara, Maria Menendez, Alicia Bertone. Pharmacokinetics

and bone resorption evaluation of a novel Cathepsin K inhibitor (VEL-0230) in healthy

adult horses. J Vet Pharmacol Ther. 2014 Dec; 37(6):556-64.

6. Menendez MI, Phelps M, Bertone A. Pharmacokinetics of Betamethasone sodium

phosphate and acetate after intra-articular administration and its effect on endogenous

hydrocortisone in exercised horses. J. vet. Pharmacol. Therap. 2015 Apr 3. doi:

10.1111/jvp.12229 [Epub ahead of print]

vii

Fields of Study

Major Field: Comparative and Veterinary Medicine

Minor Field: Imaging and Translational Medicine

viii

Table of Contents Abstract ...... ii Dedication ...... iv Acknowledgments ...... v Vita ...... vi List of Tables...... ix List of Figures ...... x

Chapter 1: The Role of Imaging in Osteoarthritis………………………………………...1

Chapter 2: Non-Invasive Quantitative Imaging Assessment in an In Vivo Canine Model of Osteoarthritis……………………………………………………………………………....7

Chapter 3: Two-deoxy-2-[fluorine-18] fluoro- D-glucose Positron Emission Tomography /Computed Tomography and co-registered Magnetic Resonance Imaging Knee Assessment after Anterior Cruciate Ligament Transection in an In Vivo Canine Model…………………………………………………………………………………….30

Chapter 4: 18F Fluoride Positron Emission Tomography /Computed Tomography and co- registered Magnetic Resonance Imaging Knee Assessment after Anterior Cruciate Ligament Transection in an In Vivo Canine Model…………………………………………………………………………………….42

Chapter 5: 18F-Fluoride Micro Positron Emission Tomography/ Computed Tomography for Ex Vivo Quantification of Bone Metabolism and Morphometry in a Canine Model of Osteoarthritis……………………………………………………………………………..54

References ...... 74

Appendix A: Direct Human Adenoviral BMP-2 or BMP-6 Gene Therapy for Bone and Cartilage Regeneration in a Pony Osteochondral Model………………………………………………………………………………….....89

List of Tables

Table A.1. Histomorphometry and gross photograph osteochondral parameters………105

List of Figures

Figure 2.1. Conventional radiographic scoring showing the overall disease, joint effusion and osteophytes at baseline, 3 and 12 weeks after ACLT. Asterisks (*) showed significant difference (P<0.05). NS: there were no significant differences between groups……..………………………………....…………………………………………...24

Figure 2.2. Mean T2, T1Gd, and MTRasym from ACLT and control articular cartilage in the femoral condyles at baseline, 3, 6 and 12 weeks after ACLT. Asterisks (*) showed significant difference (P<0.05). abc: different letters differ significantly (P<0.05). NS: there were no significant differences between groups…………………………………....25

Figure 2.3. ACLT femoral condyle articular cartilage T2 ROI (A) showing higher T2 than the control contralateral knee (B) at 12 weeks in a T2 color map……………………………………………………………………………………….26

Figure 2.4. dGEMRIC color map ACLT femoral condyle articular cartilage ROIs at baseline (A) 6 weeks (B) and 12 weeks (C) showing decreasing T1Gd over time………………………………………………………………………………………27

Figure 2.5. Representative gross morphology in the control distal femur with an intact ACL (A), articular cartilage lesions in the femoral condyles (B, C and D). Severe synovial pathology with diffuse involvement, severe discoloration and proliferation/fimbriation/thickening with fibrosis and severe hypervascularity on the distal femur (E, F) in addition to osteophyte formation in trochlear ridges and groove (E, F) and in the medial tibia (G). Medial meniscus showing severe tears (H) and complete disruption of structure (I)…………………………………………………………………………………………28

xii

Figure 2.6. Representative H&E histomorphometry showing the control tibia (A) and the medial tibia chondro-osteophyte (B) magnified (D). The medial femoral condyle (C) showing an osteophyte (dashed black line) and pannus (black square) and their magnifications (E, F) respectively. Normal synovium (G) and papillary hyperplasia with cellular infiltrate (H) and their magnifications (I, J) respectively. Normal patella (K), patella with synovial hyperplasia and early osteophyte (L) and patella with papillary hyperplasia and osteophyte (M)………………………………………………………….29

Figure 3.1. Sagittal proton density turbo spin-echo fat saturated MRI (A) and PET-MRI co-registration (B) of a control knee. Showing the ACL and PCL (blue arrows). Figure C and D show the contralateral ACLT knee, with the PCL, synovial effusion and increased FDG uptake (black asterisks)…………………………………………………………….38

Figure 3.2. Representative FDG PET-MRI co-registration showing increased FDG uptake in the lateral femur ROI at baseline and 3 weeks after ACLT…………………………………………………………………………………….39

Figure 3.3. Figure 3.3. 18F-FDG SUVmax were significantly greater in PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus of the ACLT knees relative to the control knees at 3 weeks (P< 0.02), 6 weeks (P< 0.001) and 12 weeks (P<0.002). Data were expressed as mean ± s.e.m. Asterisks (*) showed significant difference between ACLT and control knees. ………………………...………………………………………………………………….40

Figure 3.4. Representative whole body FDG PET at baseline, 3, 6 and 12 weeks. Red arrows show increased FDG uptake overtime in the ACLT knee in comparison with the contralateral control……………………………………………………………………...41

Figure 4.1. 18F-NaF SUV was significantly greater (*P < 0.02) in the lateral and medial femur, lateral and medial tibia, lateral and medial meniscus and PCL in the ACLT group relative to the control group at 12 weeks and 18F-NaF SUVmax was significantly greater at 12 weeks than baseline regardless of treatment. Medial femur had significantly greater 18F- NaF SUVmax in the ACLT than the control at 3 weeks. Data were expressed as mean+s.e.m. abc Different letters differ significantly (P<0.05). NS: there were no significant differences among treatment groups………………………………………………………………….51

xiii

Figure 4.2. Figure 4.2. 18F-NaF SUVmax at 12 weeks, from PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus in the ACLT knees were significantly greater (P< 0.002) than the control knees. At 3 weeks, 18F-NaF SUVmax from medial femoral condyle (P< 0.008), lateral femoral condyle (P< 0.01), and medial meniscus (P< 0.03) in the ACLT knees were significantly greater than the control knees. At baseline, the medial meniscus (P < 0.04) and medial tibia (P< 0.02) 18F-NaF SUVmax were greater in the control than the ACLT knee. Data were expressed as mean ± s.e.m. Asterisks (*) showed significant difference between ACLT and control knees……………………………...... 52

Figure 4.3. Representative whole body 18F-NaF PET at baseline, 3 weeks and 12 weeks. Red arrows show the increased 18F-NaF uptake overtime in the ACLT knee in comparison with the contralateral control…………………………………………………………………………………....53

Figure 5.1. Three (left) and two (right) dimensional images of the femur (A), tibia (B) representing the volumes of interest (VOI) analyzed for the femur and tibia joint and shaft trabecular bone respectively, with the proximal (red) and distal (yellow) landmark slides. Section C represents the VOI for the femoral condyles and the tibia plateau subchondral bone as well as the cortical bone in the femur and tibia shaft………………………………………………………………………………………67

Figure 5.2. Representative 18F-NaF µCT/PET 3D maps showing the control (A) and ACLT (B) femur and tibia with 2D axial, dorsal and sagittal views (from left to right) showing the radiotracer distribution. The ACLT showed greater 18F-NaF uptake than the controls (P=0.001)………………………………..………………………………………68

Figure 5.3. Representative µCT 3D views showing decreased trabecular bone BMD and BV/TV in the distal femur (A) and the distal femoral shaft (C) of the ACLT knee versus the respective control (B, D)……………………………………………………………...... 69

Figure 5.4. Representative µCT 3D views showing the subchondral bone changes and the osteophytes presence in the femoral trochlear ridges and medial tibia plateau in the ACLT

xiv knee (left) versus the contralateral control knee (right)…………………………………...... 70

Figure 5.5. Mean ± S.E.M. 18F-NaF PET (Bq/mL) related to PET findings. Each bar represents the indicated treatment. 18F-NaF uptake in the ACLT treated femur and tibia was significantly greater (P=0.001)……………………………………………………....71

Figure 5.6. A. Trabecular bone in the joint region. Mean ± S.E.M. BS/BV (1/mm), BV/TV (%), F.D, Euler number, Tb. Th (µm), Tb. N. (1/mm), Tb. Sp. (µm), S.M.I., BMD (mg/cm3). B. Subchondral bone in the joint region. Mean ± S.E.M. BMD (mg/cm3). (NS = no significant findings)…………………………………..………………...... 72

Figure 5.7. A. Trabecular bone in the shaft region. Mean ± S.E.M. BS/BV (1/mm), BV/TV (%), F.D, Euler number, Tb. Th (µm), Tb. N. (1/mm), Tb. Sp. (µm), S.M.I., BMD (mg/cm3). B. Cortical bone in the shaft region. Mean ± S.E.M. BMD (mg/cm3). (NS = no significant findings)…………………………………………………..……...... 73

Figure A.1. Images of osteochondral femoral defects in the pony model. Initially at surgery (A) and at 12 weeks in sagittal section in 3D T2 weighted MRI (B), live pony clinical CT (C), micro-CT (D), gross photograph (E) and Safranin-O histochemistry of Ad-GFP (F) and Ad-BMP6 (G). Positive safranin-O staining in the defect on histology of the Ad-BMP- 6 treated defect supported the MRI findings in Figure 2 of greater T1Gd relaxation time (GAG content) in the BMP-6 treated lesion……………………………………………………………………………………106

Figure A.2. Representative quantitative MRI maps of sagittal sections through the osteochondral defect, showing a lesion treated with Ad-BMP-6 across time using three different quantitative techniques (T2 mapping, dGEMRIC and DCE-MRI). (A) 3D T2- weighted shows the anatomy and subchondral bone changes (edema) in the distal femoral condyles, and also displays the traced lesion and adjacent cartilage ROIs. T2 mapping showed a greater T2 relaxation time in the treated lesion than the adjacent cartilage in all time points (p < 0.05). On dGEMRIC, T1Gd is significantly lower in the the treated osteochondral lesions than the adjacent cartilage (p < 0.05). For color map DCE-MRI, Amplitude was greater (p < 0.05) in the lesion at 12 weeks than either 24 or 52 weeks. Amplitude decreased at 24 weeks and was almost negligible at 52 weeks, similar to the adjacent cartilage. As expected, the adjacent cartilage values were almost negligible across xv time. The osteochondral lesion decreased in size over time, and subchondral bone changes were evident from 12 weeks. (B) At 12 weeks, dGEMRIC shows greater T1Gd relaxation time (GAG content) in the BMP-6 treated lesion than GBSS (p < 0.05). (C) Mean T2 relaxation time, T1Gd relaxation time and Amplitude from the osteochondral lesion and the adjacent cartilage at 12, 24, and 52 weeks. Asterisks (*) showed significant difference between Ad-BMP-6 and GBSS. Different letters (ab) differ significantly among time points (p < 0.05) combining all treatment groups. #Represent significant difference between lesion and adjacent cartilage (p < 0.05) combining all treatment groups………………………...... 107

Figure A.3. Bone mineral density (BMD) mean ± SEM in a live clinical CT at 12 and 24 weeks. Ad-BMP-2 treated osteochondral defects had higher BMD than GBSS at 12 weeks (#) in surrounding subchondral bone. Different letters (ab) differ significantly among time points (p < 0.05) combining all treatment groups…………………………………………………………………………………...109

Figure A.4. Mean ± SEM non-mineral area (mm2) in osteochondral lesion and Mean ± SEM bone mineral density (mg/cc) related to micro-CT findings. Each bar represents the indicated treatment. Different letters (ab) differ significantly among treatments (p< 0.05). (NS = no significant findings)………………………………………………………………………………...110

Figure A.5. Post-mortem representative micro-CT and histomorphometry for osteochondral lesions (OL) treated with GBSS, Ad-GFP, Ad-BMP-2 and Ad-BMP-6 at 52 weeks. The micro-CT shows greater area of non-mineral tissue (mm2) (Figure 4) for the Ad-BMP-2 treated lesions (p<0.05) (B) as well as changes in the subchondral bone, interpreted and corroborated by histology as a cyst (D). Three-dimensional images of the OL showing subchondral bone are represented as insets in the right corners. Gross photographs (E-H) demonstrated that lesions treated with BMPs had lower perimeter gap, better integration, and less irregular surface (F, G) than GBSS, which presented higher frequency in central cavitation formation (E). Safranin-O staining at 10X reflected a greater chondrocyte cloning where the lesion integrated with the adjacent un-injured cartilage (zone 1) in lesions treated with BMP-2 and BMP-6 (M, N). Numbers in figure L represent the zones used to evaluate histomorphometry (see Table 6.1)……………………………………………………………………………………...111

xvi

Figure A.6. Detection of active transcription at 12 weeks and adenovirus DNA at 52 weeks. Transcription of GFP was detected via RT-PCR from RNA collected from adjacent cartilage at 12 weeks. The CMV portion of adenovirus vectors were detected in DNA samples taken from surrounding cartilage at 52 weeks in Ad- BMP-2, Ad-BMP-6, and Ad- GFP treated specimens…………………………………………………………………...... 112

xvii

Chapter 1: The Role of Imaging in Osteoarthritis

Knee osteoarthritis (OA) is a major public health problem that primarily affects the elderly.

Approximately 10% of the U. S. population suffers from symptomatic knee OA by the age of 60 [1]. Prevalence of OA is increasing in the aging population and it is a frequent cause of dependency in lower-limb tasks [2]. In total, the health-care expenditures of this condition have been estimated at $189 billion annually [3]. Despite this, there are no approved interventions that ameliorate structural progression of this disorder. The growing importance of imaging in osteoarthritis for diagnosis, prognostication, and follow-up is well recognized by both clinicians and OA researchers. While conventional radiography is the gold standard imaging technique for the evaluation of known or suspected OA in clinical practice and research, it has limitations that have become apparent in the course of large magnetic resonance imaging (MRI) knee OA studies [4]. Pathological changes may be evident in all structures of a joint with OA, although traditionally researchers have viewed articular cartilage as the central feature and as the primary target for intervention and measurement. Among the commonly used imaging techniques, only MRI can assess all structures of the joint, including cartilage, meniscus, ligaments, muscle, subarticular bone marrow, and synovium, and thus can show the knee as a whole organ three- dimensionally. In addition, it can directly help in the assessment of cartilage morphology and composition. This imaging modality, therefore, plays a crucial role in increasing our understanding of the natural history of OA and in the development of new therapies. The

1 advantages and limitations of conventional radiography, MRI, and other techniques, such as nuclear medicine and computed tomography (CT) in the imaging of OA are described in this introduction.

Radiography

Radiography is the simplest, least-expensive, and most widely deployed imaging modality.

It enables detection of OA-associated bony features, such as osteophytes, subchondral sclerosis, and cysts [5]. Radiography can also determine joint space width (JSW).

Osteoarthritis is radiographically defined by the presence of marginal osteophytes, joint space narrowing (JSN) and subchondral sclerosis and cysts [6]. Progression of JSN is the most commonly used criterion for the assessment of structural OA progression [7]. The lack of sensitivity and specificity of radiography for the detection of OA-associated articular tissue damage and its poor sensitivity to change at follow-up imaging, are important limitations of this modality. Despite these limitations, radiography remains the gold standard for establishing an imaging-based diagnosis of OA and assessment of structural modification in clinical trials of knee OA.

MRI

Because of high cost per examination, MRI is not routinely used in clinical initial assessment or during disease follow-up of OA patients. However, MRI has become a key imaging tool for OA research [8, 9] thanks to its ability to visualize pathologies that are not detected on radiographs, including articular cartilage, menisci, ligaments, synovium, capsular structures, fluid collections, and bone marrow lesions (BMLs) [10, 11]. MRI enables the following: joints can be evaluated as a whole organ; multiple tissue changes

2 can be monitored simultaneously over several time points; pathologic changes of pre- radiographic OA can be detected at a much earlier stage of the disease. Physiologic changes within joint tissues can be assessed before morphologic changes become apparent.

Compositional MRI

Compositional MRI allows visualization of the biochemical properties of different joint tissues. It is therefore very sensitive to early, pre-morphologic changes that cannot be seen on conventional MRI. The vast majority of studies applying compositional MRI have focused on cartilage, although the technique can also be used to assess other tissues, such as the meniscus or ligaments. Compositional imaging of cartilage matrix changes can be performed using advanced MRI techniques, such as delayed Gadolinium Enhanced

Magnetic Resonance Imaging of Cartilage (dGEMRIC), T1 rho, and T2 mapping [8, 12,

13].

Nuclear medicine

Use of 99mTc-hydroxymethane diphosphonate (HDP) scintigraphy, 2-18F fluoro-2- deoxy-Dglucose (18-FDG) and 18F-fluoride (18-F) positron emission tomography (PET) for assessing OA have been described in the literature [14]. Bone scintigraphy is a simple examination that can provide a full-body survey that helps to discriminate between soft tissues and bone origin of pain, and to locate the site of pain in patients with complex symptoms [14]. 18-FDG PET can demonstrate the site of synovitis and bone marrow lesions (BMLs) associated with OA [15]. 18-F PET can be used for bone imaging; the amount of tracer uptake depends on the regional blood flow and bone remodeling conditions [16]. An in vivo study by Temmerman and colleagues demonstrated a

3 significant increase in bone metabolism in the proximal femur of patients with symptomatic hip OA [17], showing that 18-F PET is a potentially useful technique for early detection of OA changes. Limitations of radioisotope methods include poor anatomical resolution and the use of ionizing radiation. However, there are ways to overcome these issues. Hybrid technologies such as PET–CT and PET–MRI combine functional imaging with high-resolution anatomical imaging.

In vivo models of OA

Anatomy and joint pathology of the dog knee joint

The focus of the majority of previous research, and of the present work, is the knee (stifle) joint of the dog. However, other joints have also been extensively studied in the dog and have importance especially for spontaneously occurring disease, particularly dysplasias which cause secondary OA in the hip and elbow. However, the present work will focus only on the knee as this joint is the most frequently used as a model of OA. The anatomy of the canine knee is closely matched to that of the human. Both macroscopic and microscopic anatomies are very similar apart from size. The canine knee has medial and lateral femorotibial compartments and a patellofemoral compartment as does man. The anterior (cranial) and posterior (caudal) cruciate ligaments, menisci, meniscal ligaments, fat pad, and patellar ligaments match the human in form and function very closely. The only major gross anatomical differences are that the dog has an intra-articular long digital extensor tendon, which crosses the joint in the anteriorelateral compartment, and that the dog has lateral and medial fabella (sesamoids) in the heads of the gastrocnemius muscle, as well as a popliteal sesamoid. Biomechanically, there are differences with respect to amount of load transmission, relative joint congruency and laxity, range of motion, weight

4 bearing angle, tibial slope, and tibial thrust. Histologically and biochemically, articular cartilage, subchondral bone, synovium, joint capsule, and menisci are very well conserved between these two species. Importantly, spontaneously occurring pathology in terms of anterior cruciate ligament deficiency, meniscal pathology, osteochondrosis, and trauma is comparable in all facets between man and dog. This is one major advantage when using dogs for translational research in OA compared to smaller species where macro and microscopic anatomy, cartilage composition, and matrix turnover may significantly differ from that of humans [18].

Reproducing features of OA in animal models is crucial to gaining a better understanding of disease mechanisms and to assess the response to potential therapies: which is a prerequisite for translating basic findings into therapies for patients. Surgical models of

OA (anterior cruciate ligament transection: ACLT) show that the disease can be rapidly induced and its manifestations are less variable. However, a limitation of these models is that they more closely reflect post-traumatic OA alterations rather than spontaneous changes occurring in human OA. Small animals (mice, rats, rabbits and guinea pigs) are most often used to investigate specific disease mechanisms and for initial drug screenings for reasons of cost-effectiveness, ease of handling and housing, and opportunity for genetic manipulations. Large animal models (dogs, goats, sheep and horses) show more similarity to humans in terms of cartilage morphology, joint anatomy and joint biomechanical function, and thus provide more clinically relevant data. However, these models are relatively expensive, present important ethical concerns, and offer a limited possibility of genetic manipulations. Nevertheless, they are a crucial preclinical system to validate potential therapeutic strategies. The canine model is a valuable model for imaging markers’

5 studies in OA. Dogs are prone to develop naturally occurring OA with overuse or age and receive similar treatments as humans [18]. Synovial fluid from OA-affected dogs contains various MMPs, degradation products and cytokines that are found in human samples, and these factors have been correlated to cartilage breakdown and inflammation [19, 20] The most frequently used OA model in dogs is the surgical ACLT (also known as Pond-Nuki model) [21]. This model shows classical signs of OA, in particular, progressive damage to the articular cartilage.

There is enough evidence to support the role of quantitative imaging markers in the ACLT canine model of OA to detect early changes in knee structures. Moreover, to serially assess the OA progression in order to facilitate future therapies. Continued work is warranted to identify imaging markers implicated in disease initiation.

Chapter 2: Non-Invasive Quantitative Imaging Assessment in an In Vivo Canine Model of Osteoarthritis

INTRODUCTION

Early biochemical changes in osteoarthritis (OA) occur at the cartilage matrix molecular level, without obvious morphologic changes. Reproducing features of OA in animal models is crucial to gaining a better understanding of disease mechanisms and to assess the response to potential therapies; which is a prerequisite for translating basic findings toward therapy for patients. In surgical models of OA, such as anterior cruciate ligament transection (ACLT), OA can be rapidly induced and its manifestations are less variable [1].

This model has been extensively used and reflects post-traumatic OA in humans [2]. Large animal models, such as the canine, show more similarity to humans in terms of cartilage morphology, joint anatomy and joint biochemical function, and thus provide more relevant data [1, 3]. Conventional radiography is the primary imaging technique used to non- invasively evaluate the severity and progression of OA [4]. This modality is fundamentally limited by its inability to directly visualize articular cartilage, synovium, menisci, and other non-osseous structures involved in the pathophysiology of OA [5]. The first radiological sign of OA is the development of osteophytes; however, joint space width (JSW) measurement is the most reliable and sensitive method of grading severity. Limitations include poor sensitivity and represents cartilage loss when OA is irreversible [6]. Non- invasive in vivo imaging can be used for longitudinal follow-up in the same animal to

7 monitor the disease progression. Standard clinical Magnetic Resonance Imaging (MRI) at

3T is being used in large animal models to assess OA-associated alterations in all joint structures [7]. MRI can detect cartilage breakdown and changes even when there is normal

JSW on plain radiography [8]. Morphological MRI can measure parameters such as cartilage thickness, volume, and area. Physiological MRI methods assess the biochemical composition of cartilage: including glycosaminoglycan depletion, loss of collagen fiber orientation, and increased in water content, which are the typical characteristic changes of early OA [9]. MRI sequences include: T2 mapping, delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) and chemical exchange saturation transfer (gagCEST).

T2 mapping

It is sensitive to the mobility of protons within the water content of the extracellular matrix, which is determined by the surrounding architecture of collagen fibers and glycosaminoglycan. The signal is greater in the cartilage of patients with radiographic OA than in healthy controls; however, there is no correlation with the severity of radiographic

OA [10]. T2 mapping may be less sensitive than dGEMRIC at detecting early degenerative change [11]. dGEMRIC dGEMRIC is a validated technique for indirectly measuring cartilage glycosaminoglycan content [12]. Negatively charged ionic gadolinium contrast is injected intravenously, which diffuses into cartilage. Because glycosaminoglycans (GAGs) are also negatively charged, charge repulsion dictates that the concentration of GAG is inversely proportional to that of

8 gadolinium. Exceptional correlation exists between in vivo dGEMRIC values and the histological grade of early OA in the knee [12]. gagCEST

This technique has many applications in MRI and can be used to selectively measure signal arising from protons bound to different molecules. The hydroxyl and amide protons of

GAGs are ideally suited for CEST experiments and can be used as a biomarker for cartilage glycosaminoglycan content [13]. Unlike dGEMRIC, CEST does not require contrast.

However, it does require field strengths greater than 3T, which renders a high spatial resolution [14].

Histologic assessment of OA is currently considered the gold standard for determining presence, extent and severity of OA. The dog is the most studied species with respect to models of OA and, importantly, spontaneously occurring OA is common in dogs in multiple joints from various etiologies. The vast majority of laboratory and clinical studies have used the Mankin or modified Mankin scoring systems to determine the presence or absence of OA and assess the extent and severity of OA based on subjective identification of histologic criteria of articular cartilage pathology. While this methodology has produced a vast amount of useful and relatively repeatable data, a comprehensive literature search in conjunction with review by recognized experts in the field (Cook et al) suggest that important limitations exist including: 1. This methodology only assesses articular cartilage pathology and does not consider other tissues including synovium, subchondral bone, or menisci, which are known to be important components of initiation and progression of OA.

2. This methodology does not provide a global assessment of the joint, only those areas evaluated histologically. 3. The scoring systems arbitrarily assign values to different

9 degrees of pathology and overall values to various categories evaluated (e.g., structural changes, changes in matrix composition, and cellular changes) without making attempts to

“weight” the scores for each category based on relative importance in OA. 4. There is no standardized methodology for number of sections scored, location of sections for scoring, or ensuring normalization of staining for comparison between batches, studies, or institutions. 5. This methodology has not been extensively analyzed for statistical validity nor has it been truly validated to clinical or functional outcome measures. Therefore, Cook et al developed an OARSI histopathology initiative that address the majority of these limitations by developing, and then subsequently validating, a comprehensive histologic assessment system that is standardized, repeatable, and comparable among studies and institutions [15].

The Oseoarthritis Research Society International (OARSI) histopathology initiative filled the gaps of previous scoring systems and provided additional parameters to assess alterations in different joint structures including synovium, subchondral bone, menisci, tendons and ligaments [15].

The aim of this study was to investigate the potential of serial non-invasive biochemical quantitative imaging to quantify early disease progression in a canine model of surgically induced OA; and to correlate it with the gold standards of OA diagnosis and assessment, such as radiography and histology. We hypothesized that the knees that underwent ACLT would present greater early OA changes, including greater T2 values and lower gagCEST

(MTRasym) and dGEMRIC (T1post values) than the sham knees.

MATERIALS AND METHODS

Study design

Procedures were approved by the university’s Institutional Laboratory Animal Care and

Use Committee. Five (n=5) healthy, skeletally mature male Beagles (age 5 years; weighing

10 to 13 kg) were used. All dogs were without any clinical and radiological signs of orthopedic disorders.

Induction of OA

They underwent general anesthesia induced by acepromazine (Vedco; Saint Josep, MO,

U.S.A.; intravenously (i.v.), 0.2 mg kg), ketamine (Ketaset; Fort Dodge Animal Health,

Overland Park, KS, USA; i.v., 6 mg kg) and diazepam (Valium; Roche, Madison, WI,

USA; i.v., 0.35 mg kg) and maintained by Isoflurane (IsoFlo; Abbott, Parsippany, NJ,

USA; infusion, 2-4%). Bilateral knee arthroscopy was performed. One knee had the anterior cruciate ligament transected and the contralateral knee was explored via arthroscopy and the ACL was left intact. For the ACLT group: three standard arthroscopy portals for the canine stifle joint were established (one camera, one instrument and one egress portal). The infrapatellar fat pad was removed with a shaver as needed to allow visualization of the intra articular structures. The joint was evaluated for possible cartilage abnormalities, and the presence of osteophytes or evidence of synovitis. The anterior cruciate was transected using a hook knife; the posterior cruciate ligament remained intact.

Portals were closed subcutaneously using monofilament absorbable suture material.

Sham group: arthroscopy portals were placed in the same manner as described for the

ACLT group. The infrapatellar fat pad was removed with a shaver as needed to allow

11 visualization of the intra articular structures. After joint assessment, instruments were removed and portals closed in a routine fashion.

The dogs were individually housed in indoor pens and were fed a standard diet with water ad libitum.

Radiographs

Anteroposterior and lateral radiographs were taken at baseline, 3, and 12 weeks after ACLT under sedation with Dexdomitor (Zoetis) 0.5 mg/ml IV. Antisedan (Pfizer) 5 mg/ml IM was administered after the radiographs were taken. A radiographic scoring system described by Innes et al. for features of OA of the canine knee joint was used.

Anteroposterior and lateral radiographs were assessed by one investigator blinded to treatment. Each knee was scored for several parameters using discontinuous ordinal scales.

The outcome parameters included: Overall disease (0-3), joint effusion (0-2) and osteophytosis (0-3).

Quantitative MRI (qMRI)

Before, 3, 6, and 12 weeks after ACLT, the dogs underwent qMRI, under general anesthesia, in a 3 Tesla MRI (Achieva, Philips, Cleveland, OH) using an 8 channel SENSE knee coil (Invivo Corp, Gainesville, FL). High resolution anatomical imaging was performed in the sagittal plane with a 3D-water-selective single-shot turbo steady-state- free-precession pulse sequence in 2 stacks each covering one knee with parameters:

TR/TE=9.8/4.9 ms, flip angle=35°, acquired matrix = 132x150x31, voxel size=0.6x0.6x1.5 mm, SENSE 2, NSA 6. Each sequence was acquired as four independent stacks of one slice

12 each in the sagittal plane in order to image a sample of each of the four femoral condyles and tibia plateaus.

T2-Mapping

A multi-echo TSE sequence with 10 echoes; TE=12 to 120 ms; TR=3000ms; acquired matrix =156x174; pixel size = 0.50x0.50 mm; slice thickness = 3mm; SENSE factor 3;

NSA 2. . T2 values were calculated via linear least-squares fit [16].

Glycosaminoglycan Chemical Exchange Saturation Transfer (gagCEST) MRI

A multi-shot sequence was used with TSE factor 12; TR/TE = 1000/8ms; 148x140;

0.61x0.64 mm; slice thickness = 3 mm; SENSE factor 2; NSA 2; pre-saturation train of 16 block pulses, each 29 ms and 630°, 33 offset frequencies from -4 to +4 ppm and one S0 image at 100 kHz. B0 mapping was performed using a dual-echo SSFP sequence

(TR/TE=13/4.1, 10.1, flip angle = 10°, acq. matrix=148x142, voxel size=0.6x0.6 mm, slice thickness=3 mm). The z-spectrum was B0-corrected and the asymmetry was measured at

1 ppm.

Delayed Gadolinium-Enhanced MRI of Cartilage (dGEMRIC)

A multi-flip angle SPGR sequence was used. Flip angles = 4, 8, 12, 16, 20°, TR/TE=6.3/3.2 ms; acquired matrix= 148x148; pixel size = 0.61x0.61 mm; slice thickness = 3mm; NSA

20. Post-Gd-DTPA imaging was taken after 20 min of passive exercise following injection.

This protocol was adapted from human clinical patients [17]. T1 values were calculated by performing Levenberge-Marquardt least-squares fit.

Data analysis

Quantitative parameter mapping was performed on a voxel-by-voxel basis using in-house software written in the IDL environment (Exelis, Boulder, CO). Regions of interest (ROIs) outlining the femoral condyle cartilage in a sagittal cross section were manually traced by one investigator blinded to treatment. Outcome measurements included T2 relaxation time for T2 mapping, MTRasym for gagCEST, and T1Gd relaxation time for dGEMRIC.

Gross evaluation

Standardized digitized color photographs were taken for gross assessment of the knee. The

OARSI histological score system in dog OA was used by one investigator blinded to treatment [15]. Outcome parameters included:

Macroscopic scoring of cartilage. Score from 0-4. (Smooth surface (0) to large areas of severe damage (4).

Synovial pathology. Score form 0-5. (Normal: opal white, semitranslucent, smooth, with sparse well defined blood vessels (0) to severe: diffuse involvement, severe discoloration, consistent and severe proliferation/fimbriation/thickening, thickening to the point of fibrosis, and severe hypervascularity (5).

Meniscal changes. Score from 0-4. (None (0). to complete disruption of structure

(maceration of tissue) (4).

Histological evaluation

Sagittal cross-sectioned blocks of femoral condyle, tibia plateau, patella and meniscus, synovial membrane, and posterior cruciate ligament were fixed with 10% neutral-buffered formalin and decalcified in formic/hydrochloric acid (8% solution of each). Specimens

14 were paraffin-embedded, sectioned at 5 mm, and stained with H&E. Femoral condyle and tibia plateau were additionally stained Toluidine Blue and Safranin-O Fast Green. Sections were examined by three investigators blinded to treatment. OARSI histological score system in dog OA was used [15]. Outcome parameters included microscopic grading of cartilage, synovial and meniscal changes. The severity of cartilage, synovial and meniscal pathology was assessed according to the area of section affected. From 0 (none) to global

(>2/3).

1. Grading of cartilage structure. (A Normal volume, smooth surface with all zones

intact to E Full thickness loss of cartilage)

2. Grading of chondrocyte pathology. (A Normal to E Cell loss (necrosis/apoptosis)

predominates).

3. Grading of proteoglycan staining. (A Normal to E Full depth decrease in

proteoglycan content).

4. Grading of tidemark. (A Intact and distinct to C Loss of tidemark which is crossed

by blood vessels).

5. Grading of subchondral bone plate. (A Intact with normal thickness (≤ 300 µm) to

D Marked increased in thickness (> 750 µm) subchondral pseudocysts, and/or

marrow fibrosis).

6. Microscopic grading of synovial changes

Lining cells characteristics. (A 1-2 layers of cells to C >6 layers of cells)

Lining characteristics. (A No villous hyperplasia to C Finger-like hyperplasia).

Cell infiltration characteristics. (A No cellular infiltration. C Marked, diffuse

inflammatory cell infiltrates including large lymphoid follicles).

7. Microscopic grading of meniscal changes

Tissue Architecture-Tissue Loss. (Normal (0) to complete loss of tissue architecture,

>50% loss (3).

Cell and Matrix (PG and Collagen) Content and Morphology. (Normal (0) to severe

loss/disruption of cells, PG, and collagen (3).

Proliferative Response. None (0). Minimal proliferation of cells at synovial-meniscal

junction (1) to marked proliferation of cells involving majority of remaining tissue (3).

Statistical analysis

Student’s t-test were performed to compare ACLT versus control for continuous data.

Mann-Whitney was used for ordinal data. Repeated-measure analysis of variance

(ANOVA) (IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp) evaluated the effects of ACLT for continuous dependent variables (i.e., MRI). The post- test multiple comparisons were made among time points. Significance level was set at P <

0.05 for all analyses.

RESULTS

Outcomes were completed before, 3, 6, and 12 weeks after ACLT for all dogs.

Radiographs

Knees that underwent ACLT had significantly higher joint effusion at 3 weeks (P= 0.008) and 12 weeks (P=0.03), higher overall disease at 12 weeks (P= 0.03) and higher osteophytosis at 12 weeks (P= 0.03) than the control knees. (Figure 2.1) Overall disease and osteophytosis were significantly greater at 12 weeks than baseline and 3 weeks

(P=0.04). Joint effusion was significantly greater at 3 weeks than baseline and 12 weeks

(P= 0.03)

MRI

T2-Mapping

Femur

At 3 (P =0.009), 6, (P =0.02) and 12 (P =0.001) weeks, T2 was significantly greater in the

ACLT femoral condyles cartilage relative to the control. The ACLT cartilage group had greater T2 at 6 (P = 0.04) and 12 (P = 0.03) weeks than the baseline. No differences were shown among timelines regardless of treatment. At the baseline (P = 0.02) and 6 weeks (P=

0.04), T2 was significantly greater in the lateral femoral condyle than the medial femoral condyle (Figure 2.2, 2.3) dGEMRIC

Cartilage in ACLT at 12 weeks had significantly lower T1Gd than the control cartilage (P

= 0.04). The cartilage in the control and ACLT group showed significantly lower T1Gd at

6 (P = 0.001; p = 0.002) and 12 weeks (P = 0.02; P = 0.001) respectively, than the baseline.

At 3 (P = 0.006), 6 (P = 0.0001) and 12 weeks (P = 0.0001), T1Gd was significantly lower than the baseline regardless of treatment (Figure 2.2, 2.4)

2 A moderate negative correlation between T2 and T1Gd was found in the femur (r =-0.386;

P = 0.0001). gagCEST

No significant differences among groups or time were observed (Figure 2.2)

Gross evaluation

Macroscopic score of cartilage was significantly higher in lateral femur (P = 0.008), medial femur (P = 0.03), medial tibia (P = 0.03) and patella (P = 0.008) of ACLT knees than the controls. Synovial pathology score was significantly higher in the ACLT knees than the controls (P = 0.005). Meniscal scoring was significantly higher in medial meniscus (P =

0.02) in the ACLT knees than the controls. (Figure 2.5). Osteophytosis frequency was significantly greater in the ACLT femur, tibia, and patella (15/15) than the control ones

(0/15) (Figure 2.5).

Histological evaluation

Histomorphometry at 12 weeks showed significantly greater grading in cartilage structure in the lateral tibia than the control (P = 0.008). The subchondral bone plate in the ACLT medial tibia had significantly greater thickness than the control (P = 0.008). The meniscus in the control showed significantly greater proliferative response in the total cell and matrix score than ACLT. The cell infiltration in the synovium was significantly greater in the

ACLT than the control (P = 0.03). Osteophyte formation frequency in lateral and medial femur, tibia and patella was significantly greater in ACLT (17/35) than the control (0/35).

(Figure 2.6).

DISCUSSION

Quantitative MRI cartilage assessment in a surgically-induced canine model of OA demonstrated biochemical cartilage degeneration in the early stages of post-traumatic OA.

To the best of our knowledge, this is the first study that quantifies early biochemical and morphometric cartilage changes in an OA canine model.

The T2 values were increased in the ACLT relative to the control in all time points after surgery (3, 6, and 12 weeks). These results correlate with reported studies in humans and a mini pig model [9, 18-22]. The ACLT group had greater T2 values at 6 and 12 weeks than the baseline. The resulted increased in T2 mapping is due to the articular cartilage water content and mobility, which increased due to reduce pressure and GAG content, respectively [21]. For dGEMRIC, T1Gd was lower at 12 weeks in the ACLT group relative to the control group. The results demonstrated a reduction in T1Gd over time; including the

ACLT group and the control group, regardless of treatment. The results of T1Gd decrease over time is consistent with previous work [9, 18-21]. The dGEMRIC-T1 mapping technique has been validated in many studies to allow the evaluation of the PG degeneration of human cartilage [23-26]. Two separate imaging sessions before and after contrast agent administration with an intervention of 90-minute waiting [26] make this quantitation method time-consuming and non-feasible for a large animal model that has to be scanned under general anesthesia. We performed a twenty minute manual exercise protocol (knee flexion and extension) already described in a pony model [3]. The dGEMRIC technique is reliable in evaluating early OA because it can provide the valid information on the distribution and content of GAG in cartilage [27]. A moderate negative correlation between T2 and T1Gd in the femur correlates with previous studies [18, 19, 28].

New techniques like gagCEST are promising but not established in clinical routine. We tried to assess gagCEST and compared it with T2 and dGEMRIC. Unfortunately, we could not find any significant difference between ACLT and control group or among time. GAG loss is reported to produce lower gagCEST values [13, 14, 22, 29, 30]. Singh et al [30] concluded that CEST effect will be reduced at 3T relative to 7T and that CEST might not

19 be valuable at 3T. 3T will demand more sophisticated measures allowing for high image signal intensities, as well as reliable B0 and motion correction. Kim et al showed that the

CEST technique is feasible in a 3T setting, as they assessed GAG content in intervertebral disks [31].

The biochemical imaging assessment used in this study, is a novel approach that aims to detect early cartilage changes after ACLT. Previous MRI studies using the canine model focus on a morphological assessment of cartilage, measuring outcomes such as cartilage volume, area and qualitative morphology [4, 32-36]. These values are heavily influenced by the inter- and intra-observer variability; moreover, the morphological changes assessed indicated that OA is already present.

Conventional radiography confirmed greater overall disease and osteophytosis at 12 weeks relative to baseline and 3 weeks. These results correlate with previous work [34, 35, 37].

Joint effusion was greater after surgery (3 weeks) than the baseline, as described previously

[35]. In canine radiography, much emphasis is placed on the presence and degree of osteophytosis [37], a consequence of the inability to obtain weight-wearing radiographs, along with the low incidence of intra-articular mineralization and subchondral sclerosis.

Whereas in humans, OA is also radiographically defined by the presence of osteophytes

[38]. Progression of joint space narrowing is the most commonly used criterion for OA assessment [8]. The lack of sensitivity and specificity of radiography for the detection of

OA-associated articular tissue damage and its poor sensitivity to change at follow-up imaging are important limitations of this modality. Despite these limitations, radiography remains the gold standard for establishing an imaging-based diagnosis of OA and assessment of structural modification in clinical trials of knee OA [8].

Gross morphology and histomorphometry were assessed according to Cook et al. in the

OARSI histopathology initiative in the dog [15]. The patella, the posterior cruciate ligament (PCL) assessment, and the frequency of osteophytosis were added to the scoring protocol. The macroscopic scoring of cartilage in the lateral and medial femoral condyle, tibial plateau and the patella, along with the synovial pathology and medial meniscus was significantly greater (greater damage) in the ACLT group relative to the control. This result is due to the instability created by ACLT as described previously [21]. Osteophytes were present in all the ACLT knees, as well as in the femur, tibia, and patella. In the tibia, the osteophytes start below the rim of the medial tibia plateau and extend to more distant regions. This location coincides with osteophyte locations in a rabbit and dog ACLT model

[39, 40]. In human OA, osteophytes are formed close to the joint surface; it may be that the load bearing area increases to compensate for instability [41]. The difference in location in comparison to humans may be that the ligaments in dogs are attached to the bone at a different location than in humans, causing high stresses on the bone in a different location

[42]. In the femur, the osteophytes formed around the femoral trochlear ridges and lateral and medial femoral condyles.

The hystomorphometry at 12 weeks showed significantly greater grading in the cartilage structure (increase severity of cartilage pathology) in the lateral tibia of the ACLT group than the controls. Along with greater thickness in the subchondral bone plate of the ACLT group in the medial tibia relative to the control. The cell infiltration in the synovium was greater in the ACLT group than the control. These changes may be related with gait alterations due to the joint instability created, and the dog’s weight bearing management between the ACLT and control knees after surgery and during the following 12 weeks.

Unlike with gross morphology assessment, histology failed to identify all the osteophytes formed.

Arthroscopic surgery was chosen to deliver joint insults so as to minimize profound effects of arthrotomy on the joint including substantial synovitis, hemorrhage, joint capsular fibrosis, and the associated pain and dysfunction. The dog is one of the most studied species with respect to models of OA along with rabbits and rodents. Most importantly, clinical

OA occurs in dogs from similar causes and results in similar signs and symptoms as is seen in humans [1, 43]. Among various canine OA models, ACLT is the most frequently studied model [1]. The ACLT model is also one of the more commonly used models of OA in multiple species. Limitations of the study design should be considered. We used the contralateral knee joints as a control group instead of using control, non-operated dogs.

This minimized inter-animal variation.

The inclusion of control non-operated dogs could provide additional insights into bone morphometric changes due to possibly altered limb loading patterns after ACLT also in the contralateral joint. The study analyses were performed at 12 weeks after ACLT, not allowing longitudinal change assessment. Further studies need to be conducted with a greater dog sample to elucidate the different bone morphological changes in the ACLT model that we observed in the weight bearing femoral condyles.

This study provides novel information about the capability of each imaging modality to detect early changes in cartilage after ACLT and to manifest sensitivity to local compositional changes. These results could lead to the use of potential clinically feasible cartilage imaging biomarkers in the early stages after joint injury, facilitating OA diagnosis

22 and treatment. Further research is warranted to optimize the use of imaging biomarkers in translational research to diagnose post-traumatic OA in this canine model.

Figure 2.1: Conventional radiographic scoring showing the overall disease, joint effusion and osteophytes at baseline, 3 and 12 weeks after ACLT. Asterisks (*) showed significant difference (P<0.05). NS: there were no significant differences between groups.

Figure 2.2: Mean T2, T1Gd, and MTRasym from ACLT and control articular cartilage in the femoral condyles at baseline, 3, 6 and 12 weeks after ACLT. Asterisks (*) showed significant difference (P<0.05) between control and ACLT. abc: different letters differ significantly (P<0.05). NS: there were no significant differences between groups.

Figure 2.3: Representative T2 color map showing the ACLT femoral condyle articular cartilage T2 mapping ROI (A) showing higher T2 than the control contralateral knee (B) at 12 weeks.

Figure 2.4: Representative dGEMRIC color map ACLT femoral condyle articular cartilage ROIs at baseline (A) 6 weeks (B) and 12 weeks (C) showing decreasing T1Gd over time.

Figure 2.5: Representative gross morphology in the control distal femur with an intact ACL (A), articular cartilage lesions in the femoral condyles (B, C and D). Severe synovial pathology with diffuse involvement, severe discoloration and proliferation/fimbriation/thickening with fibrosis and severe hypervascularity on the distal femur (E, F) in addition to osteophyte formation in trochlear ridges and groove (E, F) and in the medial tibia (G). Medial meniscus showing severe tears (H) and complete disruption of structure (I).

Figure 2.6: Representative H&E histomorphometry showing the control tibia (A) and the medial tibia chondro-osteophyte (B) magnified (D). The medial femoral condyle (C) showing an osteophyte (dashed black line) and pannus (black square) and their magnifications (E, F) respectively. Normal synovium (G) and papillary hyperplasia with cellular infiltrate (H) and their magnifications (I, J) respectively. Normal patella (K), patella with synovial hyperplasia and early osteophyte (L) and patella with papillary hyperplasia and osteophyte (M) in an ACLT knee.

Chapter 3: Two-deoxy-2-[fluorine-18]fluoro- D-glucose Positron Emission Tomography

/Computed Tomography and co-registered Magnetic Resonance Imaging Knee

Assessment after Anterior Cruciate Ligament Transection in an In Vivo Canine Model

INTRODUCTION

Osteoarthritis (OA) is characterized by the degradation and loss of articular cartilage and remodeling of underlying bone. Currently, conventional radiography is the standard method for diagnosis and evaluation of severity of OA [1]. The current gold standard for the assessment of inflammation in animal models of arthritis is histopathological evaluation of joint sections. 2-deoxy-2-[18F]-fluoro-D-glucose (18F-FDG) as a metabolic tracer is a functional in vivo imaging technique that is clinically used mainly for tumor diagnosis, therapy monitoring, and experimental cancer research [2]. 18F-FDG PET can demonstrate the site of synovitis and bone marrow lesions associated with OA [3]. 18F-FDG is a glucose analog in which the 2’-OH has been replaced by 18F. Consequently, 18F-FDG cannot be further metabolized after phosphorylation and is trapped and enriched within the cell. This offers the opportunity of a quantifiable 18F-FDG PET signal from sites of pathological increased glucose metabolism in the tissue, such as in inflammation. In addition, glucose metabolism is affected by proinflammatory tumor necrosis factor-alpha (TNF-a) and increases in inflamed tissue [4], making PET a potentially interesting technique for the detection and quantification of inflammation. The accumulation of 18F-FDG in cells

30 contributing to synovial inflammation [5, 6] could provide a sensitive and non-invasive tool for visualization and quantification of joint inflammation in vivo [7]. In a mouse study using 18F-FDG, Irmler et al showed a correlation of 18F-FDG PET/CT quantification and histopathological evaluation of inflammatory experimental arthritis [7]. Paquet et al work showed that 18F-FDG accumulation increased with the progression of arthritis in a rat arthritis model [8].

18F-FDG PET has also proven its usefulness in diagnosing inflammation and infection in patients with painful hip and knee arthroplasties [9, 10]. In infection and inflammation, the increased glycolytic activity in neutrophils and activated macrophages leads to 18F-FDG uptake [9]. The spatial resolution of 18F-FDG PET is superior to conventional nuclear medicine techniques and the procedure is prompt to complete. Although 18F-FDG has a high sensitivity, it may accumulate in processes other than infection [10].

As far as we know, there is no published literature concerning serial imaging by both 18F-

FDG PET and MRI co-registered techniques in an in vivo canine model of OA.

The aim of this study was to serially assess the metabolic activity in the knee using a combination of two imaging modalities such as 18F-FDG PET co-registered with MRI in an in vivo anterior cruciate ligament transection (ACLT) canine model of OA. We hypothesized than the knees that underwent ACLT would show greater 18F-FDG uptake than the control ones.

MATERIALS AND METHODS

Study design

Procedures were approved by the university’s Institutional Laboratory Animal Care and

Use Committee. Five (n=5) healthy, skeletally mature male Beagles (age 5 years; weighing

10 to 13 kg) were used. All dogs were without any clinical and radiological signs of orthopedic disorders. The dogs were individually housed in indoor pens and were fed a standard diet with water ad libitum.

Induction of OA

They underwent general anesthesia induced by acepromazine (Vedco; Saint Josep, MO,

U.S.A.; intravenously (i.v.), 0.2 mg kg), ketamine (Ketaset; Fort Dodge Animal Health,

Overland Park, KS, USA; i.v., 6 mg kg) and diazepam (Valium; Roche, Madison, WI,

USA; i.v., 0.35 mg kg) and maintained by Isoflurane (IsoFlo; Abbott, Parsippany, NJ,

Portals were closed subcutaneously using monofilament absorbable suture material.

Sham group: arthroscopy portals were placed in the same manner as described for the

ACLT group. The infrapatellar fat pad was removed with a shaver as needed to allow visualization of the intra articular structures. After joint assessment, instruments were removed and portals closed in a routine fashion.

The dogs were individually housed in indoor pens and were fed a standard diet with water ad libitum.

Magnetic Resonance Imaging (MRI)

Before, 3, 6 and 12 weeks after ACLT, under general anesthesia, the dogs underwent MRI.

MRI was performed using a 3 Tesla MRI body system (Achieva, Philips Healthcare,

Cleveland, Ohio) with an 8-channel coil. Dogs were placed in supine position, with both knees extended in the knee coil. A custom made table was used to provide the same position in the MRI and PET imaging, in order to co-register both images. An axial proton density turbo spin-echo SPIR (TE=15 ms TR= 2.1s, flip angle 90 slice thickness= 2mm FOV

115mm) and a sagittal proton density turbo spin-echo fat saturated (TE=45ms TR= 2.2s

Flip angle 90, slice thickness 2mm FOV= 88 mm) were acquired.

PET- CT Imaging

Before, 3, 6 and 12 weeks after ACLT, under general anesthesia, the dogs underwent 18F-

FDG PET-CT. Dogs fasted six hours previous to scan. There were kept in transport cages four hours prior to scan to avoid muscle uptake. They were placed in supine position, with both knees extended in a custom made foam knee coil in order to mimic the MRI knee coil and facilitate MRI co-registration. Glucose levels were measured before 111 MBq (3 mCi) of 18F-FDG were injected and PET/CT was performed using the Gemini TF 64 PET-CT system (Philips, Cleveland, Ohio). CT was acquired with 120KVp, 163mAs and 4 mm slice thickness, 90 seconds per bed position. Whole body PET was acquired 30 minutes after 18F-FDG administration for a duration of 20 minutes.

PET- MRI Analysis

The MRI was co-registered with the PET using the Philips IntelliSpace Portal and three dimensional (3D) regions of interest (ROIs) were traced manually to assess standard uptake value (SUVs) means. Six millimeter (mm) 3D sphere were traced for the lateral femoral condyle, medial femoral condyle, lateral tibia plateau, and medial tibia plateau. 4 millimeter (mm) 3D sphere were traced for medial and lateral menisci and 3 mm 3D sphere was traced for posterior cruciate ligament (PCL). (Figure 3.1)

Statistical analyses

Linear mixed effect models were used to study the association between the type of treatment (ACLT and control) and the SUVmax at each ROI and time point, and at each ROI across all the time points, and among the time points at each ROI regardless of treatment, respectively. Linea mixed effect models were used to capture the correlations within the dogs and between dogs. P values <0.05 were considered statistically significant. Statistical analysis was performed using SAS v. 9.3 (SAS Institute, Cary, North Carolina).

RESULTS

MRI and 18F-FDG PET/CT were completed before, 3, 6, and 12 weeks after ACLT for all dogs and all the images were satisfactory co-registered.

In all structures evaluated: PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus, 18F-FDG SUVmax were greater in the ACLT knees than the control knees at 3 weeks (P< 0.02), 6 weeks (P< 0.001) and 12 weeks (P<0.002) (Figure 3.3, 3.4).

At baseline, the medial meniscus 18F-FDG SUVmax was greater in the control knee than the ACLT knee (P< 0.008).

18F-FDG SUVmax ROIs (PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus) were compared between ACLT and control knees across all time points. Each ROI had a significantly higher SUVmax in the ACLT knees than the control knees. PCL (P< 0.0005), medial femoral condyle (P<

0.008), lateral femoral condyle (P< 0.01), medial tibia (P< 0.04), lateral tibia (P< 0.02), lateral meniscus (P< 0.0001) and medial meniscus (P< 0.01).

DISCUSSION

Our findings demonstrate the feasibility of 18F-FDG to serially quantify metabolic activity and inflammation in an ACLT induced canine model of OA.

At all-time points, each ROI: PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus, had greater 18F-FDG SUVmax in the ACLT knees than the control knees.

18F-FDG PET/CT and MRI imaging modalities allow repeated investigations of the same subjects. This not only drastically reduces the number of animals required for a study but also offers the appealing possibility of longitudinal analyses within the same animal.

Moreover, in vivo imaging increases statistical quality of data in treatment studies because the status of an animal before and after disease-modulating intervention can be compared directly. Therefore, combined PET/CT using 18F-FDG as a tracer is a new and powerful tool to quantify experimental joint metabolic activity and inflammation accurately and non- invasively in vivo [7].

Histopathological assessment gives information about the number of cells present at the site of inflammation, whereas 18F-FDG PET/CT gives information about metabolic

35 activation of these cells serially in an in vivo animal model. Adding the MRI co-registration enhances the spatial resolution of PET, especially when assessing soft tissues that may be hard to localize with the CT.

The higher 18F-FDG uptake exhibited in all knee structures assessed from the knees at 3 weeks, 6 weeks, and 12 weeks, are indicative of acute metabolic alterations occurring after the ACLT. Greater 18F-FDG uptake at 3, 6, and 12 weeks for all ACLT knee structures assessed versus the control knees reflected greater inflammation and metabolic changes in the injured knees over time relative to the controls. These results are consistent with a reported rat study where18F-FDG accumulation in arthritis reflected proliferating pannus and inflammatory activity enhanced by inflammatory cytokines; suggesting that¹⁸F-FDG

μPET was effective for quantifying the inflammatory activity of arthritis and/or its therapeutic response [8].

18F-FDG uptake of the knee in an in vivo ACLT canine model using PET- MRI co- registration demonstrated to be highly sensitive in the detection of metabolic alterations in different structures comprising the knee joint. Thus, 18F-FDG uptake appeared to be a potential imaging biomarker for an early OA diagnosis prior to the expression of gross changes: as well as a diagnostic method to assess OA over time.

This study combined innovative multimodal imaging techniques to provide novel metabolic and inflammation information in the knee in an ACLT canine model of OA. To the best of our knowledge, this is the first study that utilizes different imaging modalities such as 18F-FDG PET co-registered with MRI to serially quantify early OA metabolic changes and inflammation in the whole joint in an in vivo canine model of OA. Thus, this study increases the knowledge of OA mechanisms at very early stages. The present

36 findings highlight the importance of considering both PET and MRI to assess knee structure metabolic changes in the initiation of OA. Further work is warranted to optimize the use of radiotracers in combination with MRI in translational research to diagnose post- traumatic OA in this canine model.

Figure 3.1. Sagittal proton density turbo spin-echo fat saturated MRI (A) and FDG PET- MRI co-registration (B) of a control knee. Showing the ACL and PCL (blue arrows). Figure C and D show the contralateral ACLT knee, with the PCL, synovial effusion and increased FDG uptake (black asterisks).

Figure 3.2. Representative FDG PET-MRI co-registration showing increased FDG uptake in the lateral femoral condyle ROI at baseline and 3 weeks after ACLT.

Figure 3.3. 18F-FDG SUVmax were significantly greater in PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus of the ACLT knees relative to the control knees at 3 weeks (P< 0.02), 6 weeks (P< 0.001) and 12 weeks (P<0.002). Data were expressed as mean ± s.e.m. Asterisks (*) showed significant difference between ACLT and control knees.

Figure 3.4. Representative whole body FDG PET at baseline, 3, 6 and 12 weeks. Red arrows show increased FDG uptake overtime in the ACLT knee in comparison with the contralateral control.

INTRODUCTION

Osteoarthritis (OA) is a complex degenerative disease affecting not only articular cartilage, but also the entire joint including synovium, menisci, ligaments and subchondral bone [1].

In addition, it has been suggested that these subchondral bone changes are related to the severity of the cartilage lesions [2]. Surgically-induced animal models of OA, such as an anterior cruciate ligament transection (ACLT) mainly involve inducing a mechanical instability within the joint, leading to pathological changes analogous to those observed in post-traumatic human OA [3, 4]. Sodium 18F-fluoride (18F-NaF) as a radiotracer per se in noninvasive in vivo imaging has been used to investigate musculoskeletal diseases [5-10].

The use of 18F-NaF as a bone imaging probe was established by Blau et al. in the early

1960s [11, 12], but it was subsequently replaced by 99mTc-labeled tracers due to their availability, lower costs, and the lower energy of 140-keV photons: allowing the use of γ- cameras [13]. Compared to γ-cameras, molecular imaging using PET provides the advantages of higher spatial resolution, higher sensitivity, and three dimensional tomographic image reconstructions. Furthermore, the combination of PET with CT enables attenuation correction of radiotracer signaling; allowing quantitative measurements using

18F-NaF PET/CT. Applied 18F-NaF dissociated into Na+ and 18F− is rapidly cleared from

42 the blood and accumulates in the bone where on the hydroxyapatite surface, an OH− ion is replaced by an 18F− ion to form fluorapatite. The incorporation of 18F-NaF in the bone is determined by vascular perfusion and bone surface accessibility for ion exchange; indirectly reflecting bone formation and bone resorption [14]. In clinical oncology, primary bone tumors, skeletal metastasis, benign bone diseases, and patellofemoral pain can reliably be detected by 18F-NaF PET [5-7]. In mice, pathological osteoblastic activity can be detected even earlier by 18F-NaF PET/CT imaging than by radiography, and corresponds to histological evaluation of increased bone formation [8]. As with bone tumor pathogenesis, a pathologically increased bone metabolism is a central feature of chronic arthritis: resulting in functional disorders of the joints [6]. The PET-MRI co-registration allows greater accuracy when assessing the radiotracer uptake regionally, and also provides a better insight for the new field in PET-MRI research. To the best of our knowledge, this is the first study that serially characterized bone metabolic activity in the whole knee joint using 18F-NaF PET co-registered with MRI in an in vivo ACLT canine model of OA.

The aim of this study was to examine the use of 18F-NaF PET and to co-register with MRI for the quantitative serially in vivo assessment of pathophysiological bone metabolism in early stages of induced OA in a canine model. We hypothesized than the knees that underwent ACLT would show greater 18F-NaF uptake than the control ones.

MATERIAL AND METHODS

Study design

Procedures were approved by the university’s Institutional Laboratory Animal Care and

Use Committee. Five (n=5) healthy, skeletally mature male Beagles (age 5 years; weighing

Induction of OA

They underwent general anesthesia induced by acepromazine (Vedco; Saint Josep, MO,

U.S.A.; intravenously (i.v.), 0.2 mg kg), ketamine (Ketaset; Fort Dodge Animal Health,

Overland Park, KS, USA; i.v., 6 mg kg) and diazepam (Valium; Roche, Madison, WI,

USA; i.v., 0.35 mg kg) and maintained by Isoflurane (IsoFlo; Abbott, Parsippany, NJ,

Portals were closed subcutaneously using monofilament absorbable suture material.

Sham group: arthroscopy portals were placed in the same manner as described for the

The dogs were individually housed in indoor pens and were fed a standard diet with water ad libitum.

Magnetic Resonance Imaging (MRI)

Before, 3, and 12 weeks after ACLT under general anesthesia, the dogs underwent MRI.

MRI was performed using a 3 Tesla MRI body system (Achieva, Philips Healthcare,

115mm) and a sagittal proton density turbo spin-echo fat saturated (TE=45ms TR= 2.2s

Flip angle 90, slice thickness 2mm FOV= 88 mm) were acquired.

PET- CT Imaging

Before, 3, and 12 weeks after ACLT, under general anesthesia, the dogs underwent 18 F-

NaF PET-CT. Dogs were placed in supine position, with both knees extended in a custom made foam knee coil in order to mimic the MRI knee coil and facilitate MRI co- registration. 111MBq (3 mCi) of 18 F-NaF were injected and PET/CT was performed using the Gemini TF 64 PET-CT system (Philips, Cleveland, Ohio). CT was acquired with

120KVp, 163mAs and 4 mm slice thickness, 90 seconds per bed position. Body PET was acquired for 20 minutes 30 minutes after the 18 F-NaF injection.

PET- MRI Analysis

The MRI was co-registered with the PET using the Philips IntelliSpace Portal and three dimensional (3D) regions of interest (ROIs) were traced manually to assess standard uptake

45 value (SUVs) means. 6 millimeter (mm) 3D sphere were traced for the lateral femoral condyle, medial femoral condyle, lateral tibia plateau, and medial tibia plateau. 4 millimeter (mm) 3D sphere were traced for medial and lateral menisci and 3 mm 3D sphere was traced for posterior cruciate ligament (PCL). (Figure 4.1).

Statistical analyses

RESULTS

MRI and 18F-NaF PET/CT were completed before, 3, and 12 weeks after ACLT for all dogs and all the images were satisfactory co-registered.

At 12 weeks, 18F-NaF SUVmax from PCL (P< 0.001), medial femoral condyle (P< 0.001), lateral femoral condyle (P< 0.001), medial tibia (P< 0.001), lateral tibia (P< 0.002), lateral meniscus (P< 0.001) and medial meniscus (P< 0.001) in the ACLT knees were significantly greater than the control knees (Figure 4.2 and 4.3).

At 3 weeks, 18F-NaF SUVmax from medial femoral condyle (P< 0.008), lateral femoral condyle (P< 0.01), and medial meniscus (P< 0.03) in the ACLT knees were significantly greater than the control knees (Figure 4.2 and 4.3).

At baseline, the medial meniscus (P < 0.04) and medial tibia (P< 0.02) 18 F-NaF SUVmax were greater in the control than the ACLT knee (Figure 4.2).

When 18F-NaF SUVmax in each structure (PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus) were compared between ACLT and control knees across all time points. Lateral femoral condyle (P<

0.001), lateral meniscus (P< 0.001) and medial meniscus (P< 0.001) had significantly higher SUVmax in the ACLT knees than the control knees. (Figure 4.2 and 4.3)

DISCUSSION

The usefulness of the animal PET/CT imaging modality using 18F-NaF as a radiotracer is not restricted to the detection of primary bone tumors and skeletal metastasis in cancer research. 18F-NaF PET/CT imaging is also a valid technique for use in the assessment of disease severity according to pathological bone turnover in the field of preclinical arthritis research [8]. 18F-NaF uptake in an in vivo ACLT canine model of OA using PET-CT and

MRI co-registration demonstrated to be highly sensitive in the detection of perfusion and metabolic alterations in different structures comprising the knee joint.

Femoral condyles, tibia plateau, meniscus and PCL had greater 18F-NaF uptake in ACLT knees relative to the controls at 12 weeks; indicating increased bone perfusion and highly significant bone formation (osteophytes). Interestingly, at 3 weeks, the medial femoral condyle, lateral femoral condyle and medial meniscus had significantly greater 18F-NaF uptake in the ACLT knee versus the control due to the instability created by ACLT. Similar results were found in an ACLT rat model [9] assessing 2 rats at 2, 4, and 8 weeks. At 12 weeks, greater 18F-NaF uptake was detected relative to baseline and 3 weeks, regardless

47 of treatment. This finding may be indicative of late bone turnover and abnormal bone metabolism detected by 18F-NaF, in comparison with other radiotracers such as FDG.

Nakamura et al. performed FDG PET in human knee OA, and reported that uptake was upregulated, commonly accumulating in periarticular regions [15]. Wandler et al. [16] reported that diffuse uptake in human shoulder joints was associated with OA or bursitis.

There are several reports of 18F-NaF PET in rheumatoid arthritis (RA). Kubota et al. [17] performed whole-body 18F-NaF PET/CT on RA patients and reported its usefulness for the evaluation of inflammatory activity in large joints. Matsui et al. [18] showed that 18F-

NaF uptake was correlated with active pannus in animal RA model joints.

In addition to joint inflammation, which can easily be quantified by 18F-NaF PET/CT, bone damage is the second major parameter used for assessment of arthritis severity [8]. In contrast to 18F-FDG, which is trapped in cells at sites of inflammation due to pathologically increased glucose metabolism, 18F-NaF represents specific radiotracer accumulation in the bone. Erosive processes degrading bone and cartilage in OA are associated with an increased bone surface. Therefore, the increased mineral-binding capacity results in site-specific 18F-NaF uptake in arthritic joints, which can easily be used for visualization; more importantly, it provides a measurement method for the quantification of pathological bone metabolism in preclinical arthritis models [8]. Our results are consistent with Irmler et al, and showed 18F-NaF PET/CT quantification significantly correlated with pathophysiologic bone surface alterations in a murine rheumatoid arthritis model. In a guinea pig OA model they showed that 18F-NaF uptake had decreased in areas of severe OA, contrary to our results [19]. Iagaru et al compared

99mTc-MDP bone scanning, 18F-NaF PET/CT, 18F-FDG PET/CT, and whole-body MRI

(WBMRI) for detection of known osseous metastases; and demonstrated superior image quality and evaluation of skeletal disease extent with 18F-NaF PET/CT [10]. Compared to other radiopharmaceuticals used for bone imaging, such as 99mTc, 18F-NaF has some beneficial attributes. Firstly, 18F-NaF has a high affinity to bone, resulting in favorable skeletal kinetics. Within 60 minutes after intravenous injection, only 10% of the injected dose is still located in the bloodstream [8]. Thus, the concurrence of rapid bone uptake and fast blood-pool clearance yields a favorable bone-to-background ratio. Additionally, 18F-

NaF does not accumulate in inflamed soft tissue and only minimally binds to serum proteins [20]. One limiting factor in 18F-NaF PET imaging may be vascularization of the tissue restricting tracer delivery. In contrast to epithelial tissue, the circulation in well vascularized bone tissue is less affected by exogenous factors, allowing reproducible data acquisition. In experimental arthritis, the increased vascularization and blood flow in inflamed tissue may influence tracer delivery; and therefore PET signaling at stages of acute inflammation [8, 21]. This may be the reason why we found elevated 18F-NaF uptake in meniscus and PCL of ACLT knees.

Due to the high cost per examination, MRI is not routinely used in clinical initial assessment or during disease follow-up of OA patients. However, MRI has become a key imaging tool for OA research [22-25] thanks to its ability to visualize pathologies that are not detected on radiographs, such as articular cartilage, menisci, ligaments, synovium, capsular structures, fluid collections, and bone marrow lesions (BMLs) [26-29]. MRI enables the following: the joint can be evaluated as a whole organ; multiple tissue changes can be monitored simultaneously over several time points; pathologic changes of pre- radiographic OA can be detected at a much earlier stage of the disease; and physiologic

49 changes within joint tissues (e.g., cartilage and menisci) can be assessed before morphologic changes become apparent [30]. The MRI co-registration added to the PET boosts the PET anatomical resolution, especially to trace and assess the joint ROIs. This is specifically in soft tissues, and otherwise hard to visualize with PET-CT. Overall, 18 F-

NaF uptake appeared to be a potential marker for bone turnover in early OA diagnosis, as well as in OA assessment over time.

This study combined innovative multimodal imaging techniques such as 18 F-NaF PET and MRI to provide novel information about bone remodeling changes in a canine model of early OA. To the best of our knowledge, this is the first study that serially quantifies early OA bone changes within the knee joint in an in vivo canine model of OA. Thus, this study suggested that 18 F-NaF PET is a useful tracer for assessing early changes of OA in

ACLT dogs. Furthermore, our results suggest that using 18 F-NaF PET co-registered with

MRI have the potential for early detection OA changes. Further work is warranted to optimize the use of radiotracers in combination with MRI in translational research to diagnose post-traumatic OA in this canine model.

Figure 4.1. Representative 18F-NaF PET-MRI co-registered showing increased 18F- NaF uptake in the lateral femoral condyle ROI at baseline and 12 weeks after ACLT.

Figure 4.2. 18F-NaF SUVmax at 12 weeks, from PCL, medial femoral condyle, lateral femoral condyle, medial tibia, lateral tibia, lateral meniscus and medial meniscus in the ACLT knees were significantly greater (P< 0.002) than the control knees. At 3 weeks, 18F-NaF SUVmax from medial femoral condyle (P< 0.008), lateral femoral condyle (P< 0.01), and medial meniscus (P< 0.03) in the ACLT knees were significantly greater than the control knees. At baseline, the medial meniscus (P < 0.04) and medial tibia (P< 0.02) 18F-NaF SUVmax were greater in the control than the ACLT knee. Data were expressed as mean ± s.e.m. Asterisks (*) showed significant difference between ACLT and control knees.

Chapter 5: 18F-Fluoride Micro Positron Emission Tomography/ Computed

Tomography for Ex Vivo Quantification of Bone Metabolism and Morphometry in

a Canine Model of Osteoarthritis

ABSTRACT

Objective: To quantitatively determine the changes in the properties of subchondral bone plate, trabecular bone in the joint and cortical bone and trabecular bone in the shaft. In addition, to assess the levels of bone metabolic activity in the femur and tibia in a canine model of osteoarthritis (OA).

Methods: In Five (n=5) Beagles, OA was induced via arthroscopy, one knee had the anterior cruciate ligament transected (ACLT) and the contralateral knee served as a sham- arthroscopy internal control. Twelve weeks after surgery, the knee joints were assessed by

Micro Positron Emission Tomography/ Computed Tomography (μPET/CT). Results: The mean 18F-NaF uptake (Bq/mL) was greater in the ACLT femur (p = 0.001) and tibia (p =

0.001) groups relative to the respective control groups. Morphometric analysis of the μCT data showed increased bone mineral density (BMD, p=0.007), Euler number (p=0.04) and relative percent bone volume (BV/TV, p=0.07) for the trabecular bone in the control femur joint relative to the ACLT group. A moderate inverse correlation between the trabecular and subchondral BMD and 18F-NaF uptake in the distal femur (r2= -0. 629; p = 0.051 and r2= -0. 428; p = 0.217 respectively) was shown.

Conclusions: This work confirmed that the knees that underwent ACLT had greater bone morphological changes as well as an increased bone metabolic activity. Thus, evidence is provided that μPET/CT using 18F-NaF serves as a quantitative imaging technique to detect early OA and might be beneficial to an early diagnose and treatment of OA.

INTRODUCTION

Osteoarthritis (OA) is a complex degenerative disease affecting not only articular cartilage, but also the entire joint including synovium, menisci, ligaments and subchondral bone [1].

It is important to differentiate between the subchondral bone plate and the trabecular bone contribution in OA pathogenesis, since their morphology and mechanical properties are distinctive. They also have different responses in the progression of OA [2]. Subchondral bone plate thickening as well as increased trabecular bone volume fraction have been reported in patients with late stage of OA. In addition, it has been suggested that these subchondral bone changes are related to the severity of the cartilage lesions [2, 3].

However, whether changes in the subchondral bone plate or trabecular bone occur, precede or follow cartilage degeneration at early stages of OA are still uncertain. Surgically- induced animal models of OA, such as an anterior cruciate ligament transection (ACLT)

[4] mainly involve inducing a mechanical instability within the joint, leading to pathological changes analogous to those observed in post-traumatic human OA. [5].

Structural, compositional and mechanical changes have been shown to occur in cartilage using various ACLT animal models [6]. In canine ACLT models, subchondral bone loss has been stated to occur in early OA [7-9], followed by bone sclerosis at later stages [10].

Feline models of ACLT, however, show long term thinning of the subchondral plate [11].

The differences observed in these surgically-induced OA models cannot solely be

55 explained by the use of different animals, but also by the different post-surgical time points used for monitoring the changes [2].

Methods to directly image bone metabolic activity may provide insight into the role of bone pathology in early OA. Technetium-99m hydroxymethylene diphosphonate (Tc-99m

MDP) bone scintigraphy previously revealed increased bone turnover at the knee in patients with knee osteoarthritis [12]; however, the poor spatial resolution of this technique makes it difficult to localize the specific regions of tracer uptake. 18F-sodium fluoride

(18F-NaF) Positron emission tomography/ Computed tomography (PET/CT) is an alternative to Tc-99m bone scintigraphy that makes possible to specifically and accurately localize regions of elevated bone metabolic activity. 18F fluoride ions, which have a high affinity for bone mineral, are injected intravenously. The 18F fluoride ions exchange with hydroxl ions in bone crystal to become naturally incorporated into cortical bone.

Incorporation of the 18F fluoride ion in bone is due to the activity of osteoblasts and osteoclasts during bone remodeling; therefore, processes that increase bone remodeling or bone metabolic activity will result in increased uptake of the 18F tracer [13]. The advantages of 18F-NaF PET/CT over bone scintigraphy are improved spatial resolution, greater accurate anatomical localization of tracer uptake using co-registered CT, larger ratio of bone uptake to soft-tissue uptake, and faster study times [14, 15]. To our knowledge, none of the previous studies characterized early osteoarthritic changes in the subchondral bone plate and trabecular bone in the joint, as well as cortical bone and trabecular bone in the shaft of the distal femur and proximal tibia, in addition to characterize bone metabolic activity using 18F-NaF PET/micro CT in an ACLT canine model.

The aims of the present study were to quantitatively determine the changes in the properties of subchondral bone plate, trabecular bone in the joint and cortical bone and trabecular bone in the shaft and to assess the levels of bone metabolic activity in the femur and tibia.

We hypothesized that changes in bone morphometry and metabolism would be significantly different between the ACLT and the untreated knees, and that ACLT knees would have greater bone metabolic activity than the untreated.

MATERIALS AND METHODS

Study design

Procedures were approved by the university’s Institutional Laboratory Animal Care and

Use Committee. Five (n=5) healthy, skeletally mature male Beagles (age 5 years; weighing

Induction of OA

They underwent general anesthesia induced by acepromazine (Vedco; Saint Josep, MO,

U.S.A.; intravenously (i.v.), 0.2 mg kg), ketamine (Ketaset; Fort Dodge Animal Health,

Overland Park, KS, USA; i.v., 6 mg kg) and diazepam (Valium; Roche, Madison, WI,

USA; i.v., 0.35 mg kg) and maintained by Isoflurane (IsoFlo; Abbott, Parsippany, NJ,

57 egress portal). The infrapatellar fat pad was removed with a shaver as needed to allow visualization of the intra articular structures. The joint was evaluated for possible cartilage abnormalities, and the presence of osteophytes or evidence of synovitis. The anterior cruciate was transected using a hook knife; the posterior cruciate ligament remained intact.

Portals were closed subcutaneously using monofilament absorbable suture material.

Sham group: arthroscopy portals were placed in the same manner as described for the

The dogs were individually housed in indoor pens and were fed a standard diet with water ad libitum.

PET-Micro CT Imaging

At 12 weeks the dogs received 111 MBq (3 mCi) of 18 F-NaF intravenously, underwent

PET/CT and were euthanized. Three hours after the dogs were euthanized, the distal femur and proximal tibia were harvested and immediately scanned in the µPET/CT. All the soft tissue was removed from the bones. All imaging was performed using an Inveon µPET/CT

(Siemens Preclinical, Knoxville, TN) system. Micro CT was performed using the following scan parameters: 50 um resolution, 80kV, 500uA, 360 projection, and full rotation cone beam. The PET images were acquired using the following parameters: axial FOV: 12 cm, matrix: 128 x 128 x 159, pixel size in x and y dimensions of 0.776 mm, slice thickness:

0.796 mm, 10 min acquisition, one bed position, and two dimensional filtered back projection reconstruction (2DFBP) for counting accuracy. The energy resolution of PET

58 images from this type of scanner was 1.5–2.1 mm (full width at half maximum) [16].

Morphometric analyses were performed using IAW 2.0 analysis software. The bone mineral densities (BMD) were obtained using the RATOC TRIBon software, Tokyo,

Japan. Standards containing seven disks of polymerized hydroxyapatite (HA) resin ranging from 200 mg/cc to 700 mg/cc were used to calibrate the attenuation co-efficient relative to the known BMD. All images were reconstructed using Inveon Acquisition Workplace 2.0 from Siemens Preclinical, Erlangen, Germany. A proprietary modified Feldkamp was used for reconstruction.

PET- Micro CT Analysis

Morphometric analysis was performed on the proximal tibia and distal femurs (covering the knee joint) including some of the shaft extending from the joint. All of the shaft bones were measured and normalized to the shortest bone length available (11 mm). Two dimensional (2D) and three dimensional (3D) landmarks for the volumes of interest (VOIs) in are shown in Figure 5.1. For the distal femur joint, the VOI was defined as starting from one slice distal 49.4 mm from the break in the growth plate and extending 25 mm distally until the subchondral bone was no longer continuous in the femoral condyles. For the proximal tibia joint, the VOI was defined as starting 49.4 mm, proximal from the break in connection between the growth plate and the proximal extent of the fibula. The tibia joint continues proximally until the subchondral bone on the tibia plateau was no longer continuous. The femur shaft VOI was defined as starting at the proximal limit of the growth plate and extending 11 mm proximally toward the femur shaft. The VOI of the tibia shaft started distal to apex of the fibula head and extended 11 mm distally toward the shaft.

(Figure 5.1).

The femur and tibia trabecular analysis was performed for both the joint and shaft regions of each leg. Regions of Interest (ROIs) were manually traced to within the VOI cortex to determine separate tissues to be discriminated and analyzed. To optimize the SNR across the cohort, binarization was kept at a threshold of 33,333 bit units for all samples. This level corresponds to 73.51 mg/cc BMD. Three dimensional morphometric analysis was then performed using IAW 2.0 analysis software on all trabecular bone (Tb) ROIs. The 3D outcome measurements for Tb included: bone mineral density (BMD; mg/cm3), Tb bone volume (BV/TV; %), bone surface volume ratio (BS/BV; 1/mm), Tb thickness (Tb Th; mm), trabecular number (Tb N), trabecular separation (Tb Sp; mm), fractal dimension

(FD), Euler number (EN) and structure model index (SMI). 2D morphometric analysis was performed on all subchondral bone (Sb) and cortical bone (Cb) ROIs. The 2D and 3D views were used to assess the presence or absence of osteophytes. The PET data was analyzed with the Inveon Research Workplace 3.0 from Siemens Medical Solutions USA.

Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 22.0.

Armonk, NY: IBM Corp. Student t-test comparisons of the ACLT knee and the control knee morphometric outcomes were performed. A Pearson’s correlation test was used to evaluate the correlation between 18F-NaF uptake and the µCT outcomes. In all tests, an effect was considered to be significant if the P-value was less than 0.05.

RESULTS

18F-NaF Micro PET

The mean 18F-NaF uptake (Bq/mL) was significantly greater in the ACLT femur (P =

0.001) and tibia (P = 0.001) from the ACLT knees relative to the respective control knees

(Figures 5.2, 5.5).

Quantitative Micro CT

FEMUR

Trabecular bone

In the shaft, BMD (P = 0.013), BV/TV (P = 0.04), SMI (P= 0.04) and Tb Th. (P = 0.03) were significantly greater in the control knees relative to the ACLT knees (Figures 5.3, 5.7)

In the joint, BMD (P = 0.007) and Euler number (P= 0.04) were significantly greater in the control knees relative to the ACLT knees. No significant difference in the BV/TV (P=0.07) was, however, observed. A moderate inverse correlation was observed between the BMD and 18F-NaF uptake (r2= -0. 629; P = 0.051) in the femur. (Figures 5.3, 5.6)

Subchondral bone in the femoral condyles

An inverse correlation was observed between the BMD and 18F-NaF uptake (r2= -0. 428;

P = 0.217) in the femur. (Figures 2, 4)

TIBIA

Trabecular bone

In the shaft, fractal dimension was significantly higher in the control knees versus the

ACLT (P= 0.04) knees. SMI was significantly greater in the ACLT knees versus the control knees (P=0.03) (Figures 5.4, 5.7).

In the joint, Tb. N. was significantly higher in the ACLT knees versus the control knees

(P=0.009). An inverse correlation was observed between the BMD and 18F-NaF uptake

(r2= -0. 385; P = 0.272) in the tibia (Figures 5.2, 5.6).

OSTEOPHYTES

Osteophytes were clearly seen in the µCT images as well as the PET images of the ACLT femurs and tibias in all five animals. The osteophytes were located around the trochlear ridges of the femurs and on the medial site of the tibia plateau. In none of the control joints osteophytes were observed. (Figures 5.2, 5.4)

DISCUSSION

18F-NaF μPET/CT imaging in a surgically-induced canine model of OA demonstrated morphometric and metabolic bone changes in the ACLT group compared to untreated group. To the best of our knowledge, this is the first study that quantifies morphometric and metabolic bone changes in this experimental OA animal model. In our study, 12 weeks after ACLT, we found that the distal femur and proximal tibia in the ACLT group had greater 18F-NaF uptake relative to the untreated control group. Additionally, increased bone metabolic activity was primarily localized in the medial trochlear ridges of the femur and medially in the tibia plateau, also in the intercondylar eminence and femoral condyles

(Figure 5.2). These results correlate with reported studies that show that the uptake of radioactivity was significantly greater in ACLT knees relative to sham-operated knees in the femur and tibia of a rat model of OA [17]. A similar rat ACLT model of OA also showed time-dependent pathophysiologic changes, increased bone turnover, and subchondral sclerosis [18]. Contrary results were reported in a Guinea Pig model of OA,

62 were they showed decreased 18F-NaF uptake in areas of severe OA [19]. Recent human studies have demonstrated increased subchondral bone turnover accompanied by specific architectural changes in the subchondral trabecular bone in OA joints [20, 21], which correlates with our strong inversely correlation between 18F-NaF uptake and BMD in the trabecular bone in the femur joint. Elevated bone metabolic activity (18F-NaF) was correlated with patellofemoral pain, as a result of excessive stress in the subchondral bone of the patella in a human study [15].

In our study, µCT morphometric assessment revealed that the BMD and Euler number decreased in the trabecular bone of the distal femur for the ACLT group. Interestingly, no significant difference in the BV/TV or BS/BV was observed, although BS/BV was greater

(P= 0.09) and BV/TV lower (P=0.07) in the ACLT group relative to the control. In addition structural changes were observed in the ACTL distal femur trabecular bone, as reflect by the EN (P=0.04) and SMI (P=0.07) values. Additionally, a decrease in the BMD was also observed in the femoral subchondral bone of the ACLT knees relative to the control knees, which inversely correlates with an increase in 18F-NaF uptake, indicating bone remodeling

(Figure 5.2).

MicroCT also showed that all the outcomes were not significantly different in the trabecular bone in the proximal tibia joint except for the Tb. N that was increased relative to the untreated group (Figure 5.6). Since Tb. N. must correlate inversely to 1/(Tb. Sp +

Tb. Th), the observation cannot be considered accurate. As a result, we conclude that no morphological changes were observed.

We observed a similar trend in the morphological changes in the femur and tibia Tb in the shaft region of the ACLT knees in comparison with the control knees. Showing lower BMD

(P=0.013 in femur), BV/TV (P=0.04; P=0.09), Tb. Th. (P= 0.03 in femur) and Tb. N

(P=0.13; P=0.06), and greater BS/BV (P=0.06; P=0.1) and Tb Sp. (P=0.08; P=0.08). In addition structural changes as reflected in the in F.D. (P=0.06; P=0.04) and SMI (P=0.04;

P=0.03) values were observed. These morphological and structural changes reflect bone changes that affect a part of the bones that is usually not analyzed and may provide an insight into OA early development and progression.

In the femur shaft, the trabecular bone had lower BMD, BV/TV, Tb. Th. and higher SMI in the ACLT knees compared to the control knees (Figure 5.3). None of the ACLT canine model studies looked at the femoral shaft, making these results difficult to compare to previous studies. They are, however, consistent with the loss of trabecular bone that was reported in the distal femur in two different canine models of OA and in one rabbit ACLT study [2, 8, 9]. The architectural changes in trabecular and subchondral bone in the femoral shaft may be due to the gait alterations and joint unloading previously mentioned.

In the tibia shaft, the trabecular bone had greater SMI and lower fractal dimension in the

ACLT knees compared to the control knees, thereby indicating structural change. In a study of a canine model of OA [22], they looked at the diaphyseal cortical bone and found no differences between ACLT and control knees. There was no trabecular bone assessment reported. We did not find as many trabecular changes in the tibia trabecular bone in the shaft compared to the femur. These results correlate well with previous studies that show the femur is more extensively affected, relative to the tibia, following ACLT [23, 24].

These alterations in trabecular bone and subchondral bone loss, are consistent with earlier canine ACLT model studies, which reported loss of trabecular bone after 10-12 weeks post- surgery [8, 9]. These authors concluded that the trabecular bone loss is likely a result of

64 altered joint load due to the ACLT. The µCT changes described are visually in concordance with the 18F-NaF uptake in the PET scans.

Osteophytes were present in all the ACLT knees, both in the femur and tibia. In the tibia, the osteophytes start below the rim of the medial tibia plateau and extend to more distant regions. This location coincide with osteophyte location in a rabbit ACLT model [26]. In human OA, osteophytes are formed close to the joint surface; it may be that the load bearing area increases as to compensate for instability [27]. The difference in location in comparison to humans may be that the ligaments in dogs are attached to the bone at a different location than in humans, causing high stresses on the bone in a different location

[9]. In the femur, the osteophytes formed around the femoral trochlear ridges and lateral and medial femoral condyles. (Figure 5.4). 18F-NaF uptake was extremely helpful to confirm the size and extent of the osteophytes shown by µCT.

We chose arthroscopic surgery to deliver joint insults so as to minimize profound effects of arthrotomy on the joint including substantial synovitis, hemorrhage, joint capsular fibrosis and the associated pain and dysfunction. The dog is one of the most studied species with respect to models of OA along with rabbits and rodents, and importantly, clinical OA occurs in dogs from similar causes and results in similar signs and symptoms as is seen in humans. Among various canine OA models, ACLT is the most frequently studied model.

The ACLT model is also one of the more commonly used models of OA in multiple species. Limitations of the study design should be considered. We used the contralateral knee joints as a control group instead of using control, non-operated dogs. This minimized inter-animal variation.

The inclusion of control non-operated dogs could provide additional insights into bone morphometric changes due to possibly altered limb loading patterns after ACLT also in the contralateral joint. The study analyses were performed at 12 weeks after ACLT, not allowing longitudinal changes assessment. Further studies need to be done in the future with a greater dog sample to elucidate the different bone morphological changes in the

ACLT model that we observed in the weight bearing femoral condyles.

This study provides novel information about the microstructural changes and bone metabolic activity of subchondral bone plate and trabecular bone in the joint as well as trabecular bone and cortical bone in the shaft of the femur and tibia, in a dog model of very early OA. Thus, this study increases the knowledge of the mechanisms at very early stages of OA. The present findings highlight the importance of considering bone changes and metabolism in an altered joint loading due to the ACLT model in the initiation of OA. This study shows the results of 18F-fluoride as a useful tracer for assessing early changes of OA in ACLT dogs. Our results suggest that 18F-Na has potential for the early detection of osteoarthritic changes.

Figure 5.4. Representative µCT 3D views showing the subchondral bone changes and the osteophytes presence in the femoral trochlear ridges and medial tibia plateau in the ACLT knee (left) versus the contralateral control knee (right).

References

Chapter 1:

1. Losina, E., et al., Lifetime risk and age at diagnosis of symptomatic knee osteoarthritis in the US. Arthritis Care Res (Hoboken), 2013. 65(5): p. 703-11.

2. Felson, D.T., et al., The incidence and natural history of knee osteoarthritis in the elderly. The Framingham Osteoarthritis Study. Arthritis Rheum, 1995. 38(10): p. 1500-5.

3. Kotlarz, H., et al., Insurer and out-of-pocket costs of osteoarthritis in the US: evidence from national survey data. Arthritis Rheum, 2009. 60(12): p. 3546-53.

4. Amin, S., et al., The relationship between cartilage loss on magnetic resonance imaging and radiographic progression in men and women with knee osteoarthritis. Arthritis Rheum, 2005. 52(10): p. 3152-9.

5. Altman, R.D. and G.E. Gold, Atlas of individual radiographic features in osteoarthritis, revised. Osteoarthritis Cartilage, 2007. 15 Suppl A: p. A1-56.

6. Kellgren, J.H. and J.S. Lawrence, Radiological assessment of osteo-arthrosis. Ann Rheum Dis, 1957. 16(4): p. 494-502.

7. Carr, A.J., et al., Knee replacement. Lancet, 2012. 379(9823): p. 1331-40.

8. Roemer, F.W., et al., Advances in imaging of osteoarthritis and cartilage. Radiology, 2011. 260(2): p. 332-54.

9. Eckstein, F., W. Wirth, and M.C. Nevitt, Recent advances in osteoarthritis imaging- -the osteoarthritis initiative. Nat Rev Rheumatol, 2012. 8(10): p. 622-30.

10. Guermazi, A., et al., Prevalence of abnormalities in knees detected by MRI in adults without knee osteoarthritis: population based observational study (Framingham Osteoarthritis Study). Bmj, 2012. 345: p. e5339.

11. Roemer, F.W., et al., Co-localisation of non-cartilaginous articular pathology increases risk of cartilage loss in the tibiofemoral joint--the MOST study. Ann Rheum Dis, 2013. 72(6): p. 942-8.

12. Crema, M.D., et al., Articular cartilage in the knee: current MR imaging techniques and applications in clinical practice and research. Radiographics, 2011. 31(1): p. 37-61.

13. Trattnig, S., et al., MR imaging of cartilage and its repair in the knee--a review. Eur Radiol, 2009. 19(7): p. 1582-94.

14. Omoumi, P., et al., CT arthrography, MR arthrography, PET, and scintigraphy in osteoarthritis. Radiol Clin North Am, 2009. 47(4): p. 595-615.

15. Nakamura, H., et al., Positron emission tomography with 18F-FDG in osteoarthritic knee. Osteoarthritis Cartilage, 2007. 15(6): p. 673-81.

16. Umemoto, Y., et al., Imaging of a rat osteoarthritis model using (18)F-fluoride positron emission tomography. Ann Nucl Med, 2010. 24(9): p. 663-9.

17. Temmerman, O.P., et al., In vivo measurements of blood flow and bone metabolism in osteoarthritis. Rheumatol Int, 2013. 33(4): p. 959-63.

18. Cook, J.L., et al., The OARSI histopathology initiative - recommendations for histological assessments of osteoarthritis in the dog. Osteoarthritis Cartilage, 2010. 18 Suppl 3: p. S66-79.

19. Pelletier, J.P., et al., The protective effect of licofelone on experimental osteoarthritis is correlated with the downregulation of gene expression and protein synthesis of several major cartilage catabolic factors: MMP-13, cathepsin K and aggrecanases. Arthritis Res Ther, 2005. 7(5): p. R1091-102.

20. Pelletier, J.P., et al., The inhibition of subchondral bone resorption in the early phase of experimental dog osteoarthritis by licofelone is associated with a reduction in the synthesis of MMP-13 and cathepsin K. Bone, 2004. 34(3): p. 527-38.

21. Kuroki, K., C.R. Cook, and J.L. Cook, Subchondral bone changes in three different canine models of osteoarthritis. Osteoarthritis Cartilage, 2011. 19(9): p. 1142-9.

Chapter 2:

Thysen, S., F.P. Luyten, and R.J. Lories, Targets, models and challenges in osteoarthritis research. Dis Model Mech, 2015. 8(1): p. 17-30.

2. Kuroki, K., C.R. Cook, and J.L. Cook, Subchondral bone changes in three different canine models of osteoarthritis. Osteoarthritis Cartilage, 2011. 19(9): p. 1142-9.

3. Menendez, M.I., et al., Direct delayed human adenoviral BMP-2 or BMP-6 gene therapy for bone and cartilage regeneration in a pony osteochondral model. Osteoarthritis Cartilage, 2011. 19(8): p. 1066-75.

4. Batiste, D.L., et al., High-resolution MRI and micro-CT in an ex vivo rabbit anterior cruciate ligament transection model of osteoarthritis. Osteoarthritis Cartilage, 2004. 12(8): p. 614-26.

5. Fife, R.S., et al., Relationship between arthroscopic evidence of cartilage damage and radiographic evidence of joint space narrowing in early osteoarthritis of the knee. Arthritis Rheum, 1991. 34(4): p. 377-82.

6. Palmer, A.J., et al., Non-invasive imaging of cartilage in early osteoarthritis. Bone Joint J, 2013. 95-b(6): p. 738-46.

7. Galindo-Zamora, V., et al., Diagnostic accuracy of a short-duration 3 Tesla magnetic resonance protocol for diagnosing stifle joint lesions in dogs with non-traumatic cranial cruciate ligament rupture. BMC Vet Res, 2013. 9: p. 40.

8. Guermazi, A., et al., Prevalence of abnormalities in knees detected by MRI in adults without knee osteoarthritis: population based observational study (Framingham Osteoarthritis Study). Bmj, 2012. 345: p. e5339.

9. Wang, L. and R.R. Regatte, Quantitative mapping of human cartilage at 3.0T: parallel changes in T(2), T(1)rho, and dGEMRIC. Acad Radiol, 2014. 21(4): p. 463-71.

10. Dunn, T.C., et al., T2 relaxation time of cartilage at MR imaging: comparison with severity of knee osteoarthritis. Radiology, 2004. 232(2): p. 592-8.

11. Kumahashi, N., et al., A longitudinal study of the quantitative evaluation of patella cartilage after total knee replacement by delayed gadolinium-enhanced magnetic resonance imaging of cartilage (dGEMRIC) and T2 mapping at 3.0 T: preliminary results. Osteoarthritis Cartilage, 2013. 21(1): p. 126-35.

12. Trattnig, S., et al., MRI visualization of proteoglycan depletion in articular cartilage via intravenous administration of Gd-DTPA. Magn Reson Imaging, 1999. 17(4): p. 577-83.

13. Ling, W., et al., Assessment of glycosaminoglycan concentration in vivo by chemical exchange-dependent saturation transfer (gagCEST). Proc Natl Acad Sci U S A, 2008. 105(7): p. 2266-70.

14. Singh, A., et al., Chemical exchange saturation transfer magnetic resonance imaging of human knee cartilage at 3 T and 7 T. Magn Reson Med, 2012. 68(2): p. 588- 94.

15. Cook, J.L., et al., The OARSI histopathology initiative - recommendations for histological assessments of osteoarthritis in the dog. Osteoarthritis Cartilage, 2010. 18 Suppl 3: p. S66-79.

16. White, L.M., et al., Cartilage T2 assessment: differentiation of normal hyaline cartilage and reparative tissue after arthroscopic cartilage repair in equine subjects. Radiology, 2006. 241(2): p. 407-14.

17. Domayer, S.E., et al., Delayed gadolinium-enhanced MRI of cartilage in the ankle at 3 T: feasibility and preliminary results after matrix-associated autologous chondrocyte implantation. J Magn Reson Imaging, 2010. 31(3): p. 732-9.

18. Klocke, N.F., et al., Comparison of T1rho, dGEMRIC, and quantitative T2 MRI in preoperative ACL rupture patients. Acad Radiol, 2013. 20(1): p. 99-107.

19. Rehnitz, C., et al., Comparison of biochemical cartilage imaging techniques at 3 T MRI. Osteoarthritis Cartilage, 2014. 22(10): p. 1732-42.

20. Baum, T., et al., Cartilage and meniscal T2 relaxation time as non-invasive biomarker for knee osteoarthritis and cartilage repair procedures. Osteoarthritis Cartilage, 2013. 21(10): p. 1474-84.

21. Wei, B., et al., Use of quantitative MRI for the detection of progressive cartilage degeneration in a mini-pig model of osteoarthritis caused by anterior cruciate ligament transection. J Magn Reson Imaging, 2015.

22. Apprich, S., et al., Detection of degenerative cartilage disease: comparison of high- resolution morphological MR and quantitative T2 mapping at 3.0 Tesla. Osteoarthritis Cartilage, 2010. 18(9): p. 1211-7.

23. Williams, A., et al., Glycosaminoglycan distribution in cartilage as determined by delayed gadolinium-enhanced MRI of cartilage (dGEMRIC): potential clinical applications. AJR Am J Roentgenol, 2004. 182(1): p. 167-72.

24. McKenzie, C.A., et al., Three-dimensional delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) at 1.5T and 3.0T. J Magn Reson Imaging, 2006. 24(4): p. 928-33.

25. Tiderius, C.J., et al., Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) in early knee osteoarthritis. Magn Reson Med, 2003. 49(3): p. 488-92.

26. Burstein, D., et al., Protocol issues for delayed Gd(DTPA)(2-)-enhanced MRI (dGEMRIC) for clinical evaluation of articular cartilage. Magn Reson Med, 2001. 45(1): p. 36-41.

27. Blumenkrantz, G. and S. Majumdar, Quantitative magnetic resonance imaging of articular cartilage in osteoarthritis. Eur Cell Mater, 2007. 13: p. 76-86.

28. Neuman, P., et al., Longitudinal assessment of femoral knee cartilage quality using contrast enhanced MRI (dGEMRIC) in patients with anterior cruciate ligament injury-- comparison with asymptomatic volunteers. Osteoarthritis Cartilage, 2011. 19(8): p. 977- 83.

29. Schmitt, B., et al., Cartilage quality assessment by using glycosaminoglycan chemical exchange saturation transfer and (23)Na MR imaging at 7 T. Radiology, 2011. 260(1): p. 257-64.

30. Krusche-Mandl, I., et al., Long-term results 8 years after autologous osteochondral transplantation: 7 T gagCEST and sodium magnetic resonance imaging with morphological and clinical correlation. Osteoarthritis Cartilage, 2012. 20(5): p. 357-63.

31. Kim, M., et al., Assessment of glycosaminoglycan distribution in human lumbar intervertebral discs using chemical exchange saturation transfer at 3 T: feasibility and initial experience. NMR Biomed, 2011. 24(9): p. 1137-44.

32. Pujol, E., et al., Anatomic study of the canine stifle using low-field magnetic resonance imaging (MRI) and MRI arthrography. Vet Surg, 2011. 40(4): p. 395-401.

33. Pepin, S.R., et al., A comparative analysis of 7.0-Tesla magnetic resonance imaging and histology measurements of knee articular cartilage in a canine posterolateral knee injury model: a preliminary analysis. Am J Sports Med, 2009. 37 Suppl 1: p. 119s-24s.

34. Libicher, M., et al., Early changes in experimental osteoarthritis using the Pond- Nuki dog model: technical procedure and initial results of in vivo MR imaging. Eur Radiol, 2005. 15(2): p. 390-4.

35. D'Anjou, M.A., et al., Osteophytosis, subchondral bone sclerosis, joint effusion and soft tissue thickening in canine experimental stifle osteoarthritis: comparison between 1.5 T magnetic resonance imaging and computed radiography. Vet Surg, 2008. 37(2): p. 166- 77.

36. Endo, J., et al., Utility of T2 mapping and dGEMRIC for evaluation of cartilage repair after allograft chondrocyte implantation in a rabbit model. Osteoarthritis Cartilage, 2015. 23(2): p. 280-8.

37. Innes, J.F., et al., Radiographic progression of osteoarthritis of the canine stifle joint: a prospective study. Vet Radiol Ultrasound, 2004. 45(2): p. 143-8.

38. Roemer, F.W., et al., The role of imaging in osteoarthritis. Best Pract Res Clin Rheumatol, 2014. 28(1): p. 31-60.

39. Batiste, D.L., et al., Ex vivo characterization of articular cartilage and bone lesions in a rabbit ACL transection model of osteoarthritis using MRI and micro-CT. Osteoarthritis Cartilage, 2004. 12(12): p. 986-96.

40. Boileau, C., et al., Magnetic resonance imaging can accurately assess the long- term progression of knee structural changes in experimental dog osteoarthritis. Ann Rheum Dis, 2008. 67(7): p. 926-32.

41. Dayal, N., et al., The natural history of anteroposterior laxity and its role in knee osteoarthritis progression. Arthritis Rheum, 2005. 52(8): p. 2343-9.

42. Sniekers, Y.H., A role for subchondral bone changes in the process of osteoarthritis; a micro-CT study of two canine models. BME Musculoskelet. Disord., 2008. 9: p. 20.

43. Lampropoulou-Adamidou, K., et al., Useful animal models for the research of osteoarthritis. Eur J Orthop Surg Traumatol, 2014. 24(3): p. 263-71.

Chapter 3:

1. Palmer, A.J., et al., Non-invasive imaging of cartilage in early osteoarthritis. Bone Joint J, 2013. 95-b(6): p. 738-46.

2. Lawrence, J., E. Rohren, and J. Provenzale, PET/CT today and tomorrow in veterinary cancer diagnosis and monitoring: fundamentals, early results and future perspectives. Vet Comp Oncol, 2010. 8(3): p. 163-87.

3. Ju, J.H., et al., Visualization and localization of rheumatoid knee synovitis with FDG-PET/CT images. Clin Rheumatol, 2008. 27 Suppl 2: p. S39-41.

4. Yamada, S., et al., High accumulation of fluorine-18-fluorodeoxyglucose in turpentine-induced inflammatory tissue. J Nucl Med, 1995. 36(7): p. 1301-6.

5. Jones, H.A., et al., Dissociation between respiratory burst activity and deoxyglucose uptake in human neutrophil granulocytes: implications for interpretation of (18)F-FDG PET images. J Nucl Med, 2002. 43(5): p. 652-7.

6. Deichen, J.T., et al., Uptake of [18F]fluorodeoxyglucose in human monocyte- macrophages in vitro. Eur J Nucl Med Mol Imaging, 2003. 30(2): p. 267-73.

7. Irmler, I.M., et al., In vivo molecular imaging of experimental joint inflammation by combined (18)F-FDG positron emission tomography and computed tomography. Arthritis Res Ther, 2010. 12(6): p. R203.

8. Paquet, J., et al., Evaluation of a rat knee mono-arthritis using microPET. Biomed Mater Eng, 2010. 20(3): p. 195-202.

9. Reinartz, P., FDG-PET in patients with painful hip and knee arthroplasty: technical breakthrough or just more of the same. Q J Nucl Med Mol Imaging, 2009. 53(1): p. 41-50.

10. Zajonz, D., et al., Diagnostic work-up strategy for periprosthetic joint infections after total hip and knee arthroplasty: a 12-year experience on 320 consecutive cases. Patient Saf Surg, 2015. 9: p. 20.

Chapter 4:

1. Buckwalter, J.A., C. Saltzman, and T. Brown, The impact of osteoarthritis: implications for research. Clin Orthop Relat Res, 2004(427 Suppl): p. S6-15.

2. Intema, F., et al., In early OA, thinning of the subchondral plate is directly related to cartilage damage: results from a canine ACLT-meniscectomy model. Osteoarthritis Cartilage, 2010. 18(5): p. 691-8.

3. Pond MJ., N.G., Experimentally-induced osteoarthritis in the dog. Ann Rheum Dis, 1973(32): p. 387-8.

4. Florea, C., et al., Alterations in subchondral bone plate, trabecular bone and articular cartilage properties of rabbit femoral condyles at 4 weeks after anterior cruciate ligament transection. Osteoarthritis Cartilage, 2014.

5. Jadvar, H., B. Desai, and P.S. Conti, Sodium 18F-fluoride PET/CT of bone, joint, and other disorders. Semin Nucl Med, 2015. 45(1): p. 58-65.

6. Li, Y., et al., Clinical utility of (18)F-fluoride PET/CT in benign and malignant bone diseases. Bone, 2012. 50(1): p. 128-39.

7. Draper, C.E., et al., Patients with patellofemoral pain exhibit elevated bone metabolic activity at the patellofemoral joint. J Orthop Res, 2012. 30(2): p. 209-13.

8. Irmler, I.M., et al., 18 F-Fluoride positron emission tomography/computed tomography for noninvasive in vivo quantification of pathophysiological bone metabolism in experimental murine arthritis. Arthritis Res Ther, 2014. 16(4): p. R155.

9. Umemoto, Y., et al., Imaging of a rat osteoarthritis model using (18)F-fluoride positron emission tomography. Ann Nucl Med, 2010. 24(9): p. 663-9.

10. Iagaru, A., et al., Prospective evaluation of (99m)Tc MDP scintigraphy, (18)F NaF PET/CT, and (18)F FDG PET/CT for detection of skeletal metastases. Mol Imaging Biol, 2012. 14(2): p. 252-9.

11. Blau, M., W. Nagler, and M.A. Bender, Fluorine-18: a new isotope for bone scanning. J Nucl Med, 1962. 3: p. 332-4.

12. Blau, M., R. Ganatra, and M.A. Bender, 18 F-fluoride for bone imaging. Semin Nucl Med, 1972. 2(1): p. 31-7.

13. Grant, F.D., et al., Skeletal PET with 18F-fluoride: applying new technology to an old tracer. J Nucl Med, 2008. 49(1): p. 68-78.

14. Tomlinson, R.E., M.J. Silva, and K.I. Shoghi, Quantification of skeletal blood flow and fluoride metabolism in rats using PET in a pre-clinical stress fracture model. Mol Imaging Biol, 2012. 14(3): p. 348-54.

15. Nakamura, H., et al., Positron emission tomography with 18F-FDG in osteoarthritic knee. Osteoarthritis Cartilage, 2007. 15(6): p. 673-81.

16. Wandler, E., et al., Diffuse FDG shoulder uptake on PET is associated with clinical findings of osteoarthritis. AJR Am J Roentgenol, 2005. 185(3): p. 797-803.

17. Kubota, K., et al., Whole-body FDG-PET/CT on rheumatoid arthritis of large joints. Ann Nucl Med, 2009. 23(9): p. 783-91.

18. Matsui, T., et al., Inflammatory cytokines and hypoxia contribute to 18F-FDG uptake by cells involved in pannus formation in rheumatoid arthritis. J Nucl Med, 2009. 50(6): p. 920-6.

19. Dyke, J.P., et al., Characterization of bone perfusion by dynamic contrast-enhanced magnetic resonance imaging and positron emission tomography in the Dunkin-Hartley guinea pig model of advanced osteoarthritis. J Orthop Res, 2015. 33(3): p. 366-72.

20. Blake, G.M., et al., Quantitative studies of bone with the use of 18F-fluoride and 99mTc-methylene diphosphonate. Semin Nucl Med, 2001. 31(1): p. 28-49.

21. Frost, M.L., et al., Long-term precision of 18F-fluoride PET skeletal kinetic studies in the assessment of bone metabolism. J Nucl Med, 2008. 49(5): p. 700-7.

22. Roemer, F.W., et al., Advances in imaging of osteoarthritis and cartilage. Radiology, 2011. 260(2): p. 332-54.

23. Conaghan, P.G., et al., Summary and recommendations of the OARSI FDA osteoarthritis Assessment of Structural Change Working Group. Osteoarthritis Cartilage, 2011. 19(5): p. 606-10.

24. Eckstein, F., W. Wirth, and M.C. Nevitt, Recent advances in osteoarthritis imaging- -the osteoarthritis initiative. Nat Rev Rheumatol, 2012. 8(10): p. 622-30.

25. Hayashi, D., A. Guermazi, and F.W. Roemer, MRI of osteoarthritis: the challenges of definition and quantification. Semin Musculoskelet Radiol, 2012. 16(5): p. 419-30.

26. Guermazi, A., et al., Prevalence of abnormalities in knees detected by MRI in adults without knee osteoarthritis: population based observational study (Framingham Osteoarthritis Study). Bmj, 2012. 345: p. e5339.

27. Crema, M.D., et al., Articular cartilage in the knee: current MR imaging techniques and applications in clinical practice and research. Radiographics, 2011. 31(1): p. 37-61.

28. Felson, D.T., et al., Bone marrow lesions in knee osteoarthritis change in 6-12 weeks. Osteoarthritis Cartilage, 2012. 20(12): p. 1514-8.

29. Roemer, F.W., et al., Prevalence of magnetic resonance imaging-defined atrophic and hypertrophic phenotypes of knee osteoarthritis in a population-based cohort. Arthritis Rheum, 2012. 64(2): p. 429-37.

30. Roemer, F.W., et al., The role of imaging in osteoarthritis. Best Pract Res Clin Rheumatol, 2014. 28(1): p. 31-60.

Chapter 5:

1. Buckwalter, J.A., C. Saltzman, and T. Brown, The impact of osteoarthritis: implications for research. Clin Orthop Relat Res, 2004(427 Suppl): p. S6-15.

2. Florea, C., et al., Alterations in subchondral bone plate, trabecular bone and articular cartilage properties of rabbit femoral condyles at 4 weeks after anterior cruciate ligament transection. Osteoarthritis Cartilage, 2014.

3. Bobinac, D., et al., Microstructural alterations of femoral head articular cartilage and subchondral bone in osteoarthritis and osteoporosis. Osteoarthritis Cartilage, 2013. 21(11): p. 1724-30.

4. Pond MJ., N.G., Experimentally-induced osteoarthritis in the dog. Ann Rheum Dis, 1973(32): p. 387-8.

5. Bendele, A.M., Animal models of osteoarthritis. J Musculoskelet Neuronal Interact, 2001. 1(4): p. 363-76.

6. Intema, F., et al., In early OA, thinning of the subchondral plate is directly related to cartilage damage: results from a canine ACLT-meniscectomy model. Osteoarthritis Cartilage, 2010. 18(5): p. 691-8.

7. Intema, F., Similarities and discrepancies in subchondral bone structure in two differently induced canine models of osteoarthritis. J. Bone Miner. Res., 2010. 25: p. 1650- 1657.

8. Kuroki, K., C.R. Cook, and J.L. Cook, Subchondral bone changes in three different canine models of osteoarthritis. Osteoarthritis Cartilage, 2011. 19(9): p. 1142-9.

9. Sniekers, Y.H., A role for subchondral bone changes in the process of osteoarthritis; a micro-CT study of two canine models. BME Musculoskelet. Disord., 2008. 9: p. 20.

10. Hayami, T., et al., Characterization of articular cartilage and subchondral bone changes in the rat anterior cruciate ligament transection and meniscectomized models of osteoarthritis. Bone, 2006. 38(2): p. 234-43.

11. Boyd, S.K., et al., Long-term periarticular bone adaptation in a feline knee injury model for post-traumatic experimental osteoarthritis. Osteoarthritis Cartilage, 2005. 13(3): p. 235-42.

12. Dieppe, P., et al., Prediction of the progression of joint space narrowing in osteoarthritis of the knee by bone scintigraphy. Ann. Rheum. Dis., 1993. 52: p. 557-563.

13. Blau, M., R. Ganatra, and M.A. Bender, 18 F-fluoride for bone imaging. Semin Nucl Med, 1972. 2(1): p. 31-7.

14. Grant, F.D., et al., Skeletal PET with 18F-fluoride: applying new technology to an old tracer. J Nucl Med, 2008. 49(1): p. 68-78.

15. Draper, C.E., et al., Patients with patellofemoral pain exhibit elevated bone metabolic activity at the patellofemoral joint. J Orthop Res, 2012. 30(2): p. 209-13.

16. Constantinescu, C.C. and J. Mukherjee, Performance evaluation of an Inveon PET preclinical scanner. Phys Med Biol, 2009. 54(9): p. 2885-99.

17. Umemoto, Y., et al., Imaging of a rat osteoarthritis model using (18)F-fluoride positron emission tomography. Ann Nucl Med, 2010. 24(9): p. 663-9.

18. Hayami, T., The role of subchondral bone remodeling in osteoarthritis: reduction of cartilage degeneration and prevention of osteophyte formation by alendronate in the rat anterior cruciate ligament transaction model. Arthritis Rheum., 2004. 50: p. 1193-1206.

20. Burr, D.B., The importance of subchondral bone in osteoarthrosis. Curr Opin Rheumatol, 1998. 10(3): p. 256-62.

21. Goker, B., et al., Bone mineral density varies as a function of the rate of joint space narrowing in the hip. J Rheumatol, 2000. 27(3): p. 735-8.

22. Sniekers, Y.H., et al., A role for subchondral bone changes in the process of osteoarthritis; a micro-CT study of two canine models. BMC Musculoskelet Disord, 2008. 9: p. 20.

23. Wei, X., T. Rasanen, and K. Messner, Maturation-related compressive properties of rabbit knee articular cartilage and volume fraction of subchondral tissue. Osteoarthritis Cartilage, 1998. 6(6): p. 400-9.

24. Mohan, G., et al., Pre-emptive, early, and delayed alendronate treatment in a rat model of knee osteoarthritis: effect on subchondral trabecular bone microarchitecture and cartilage degradation of the tibia, bone/cartilage turnover, and joint discomfort. Osteoarthritis Cartilage, 2013. 21(10): p. 1595-604.

25. O'Connor, B.L., et al., Gait alterations in dogs after transection of the anterior cruciate ligament. Arthritis Rheum, 1989. 32(9): p. 1142-7.

26. Batiste, D.L., et al., Ex vivo characterization of articular cartilage and bone lesions in a rabbit ACL transection model of osteoarthritis using MRI and micro-CT. Osteoarthritis Cartilage, 2004. 12(12): p. 986-96.

27. Dayal, N., et al., The natural history of anteroposterior laxity and its role in knee osteoarthritis progression. Arthritis Rheum, 2005. 52(8): p. 2343-9.

Appendix A:

1. Buckwalter JA, Mankin HJ. Articular cartilage: degeneration and osteoarthritis, repair, regeneration, and transplantation. Instr Course Lect 1998; 47: 487-504.

2. Hunziker EB. Articular cartilage repair: basic science and clinical progress. A review of the current status and prospects. Osteoarthritis Cartilage 2002; 10: 432-463.

3. Bobic V. Arthroscopic osteochondral autograft transplantation in anterior cruciate ligament reconstruction: a preliminary clinical study. Knee Surg Sports Traumatol Arthrosc 1996; 3: 262-264.

4. Haddo O, Mahroof S, Higgs D, David L, Pringle J, Bayliss M, et al. The use of chondrogide membrane in autologous chondrocyte implantation. Knee 2004; 11: 51-55.

5. Cherubino P, Grassi FA, Bulgheroni P, Ronga M. Autologous chondrocyte implantation using a bilayer collagen membrane: a preliminary report. J Orthop Surg (Hong Kong) 2003; 11: 10-15.

6. Adkisson HDt, Martin JA, Amendola RL, Milliman C, Mauch KA, Katwal AB, et al. The potential of human allogeneic juvenile chondrocytes for restoration of articular cartilage. Am J Sports Med 2010; 38: 1324-1333.

7. Mankin HJ. The response of articular cartilage to mechanical injury. J Bone Joint Surg Am 1982; 64: 460-466.

8. Coletti JM, Jr., Akeson WH, Woo SL. A comparison of the physical behavior of normal articular cartilage and the arthroplasty surface. J Bone Joint Surg Am 1972; 54: 147-160.

9. Radin EL, Abernethy PJ, Townsend PM, Rose RM. The role of bone changes in the degeneration of articular cartilage in osteoarthrosis. Acta Orthop Belg 1978; 44: 55-63.

10. Goodrich LR, Hidaka C, Robbins PD, Evans CH, Nixon AJ. Genetic modification of chondrocytes with insulin-like growth factor-1 enhances cartilage healing in an equine model. J Bone Joint Surg Br 2007; 89: 672-685.

11. Cook SD, Patron LP, Salkeld SL, Rueger DC. Repair of articular cartilage defects with osteogenic protein-1 (BMP-7) in dogs. J Bone Joint Surg Am 2003; 85-A Suppl 3: 116-123.

12. Frisbie DD, Cross MW, McIlwraith CW. A comparative study of articular cartilage thickness in the stifle of animal species used in human pre-clinical studies compared to articular cartilage thickness in the human knee. Vet Comp Orthop Traumatol 2006; 19: 142-146.

13. Fortier LA, Potter HG, Rickey EJ, Schnabel LV, Foo LF, Chong LR, et al. Concentrated bone marrow aspirate improves full-thickness cartilage repair compared with microfracture in the equine model. J Bone Joint Surg Am 2010; 92: 1927-1937.

14. Nixon AJ, Fortier LA, Williams J, Mohammed H. Enhanced repair of extensive articular defects by insulin-like growth factor-I-laden fibrin composites. J Orthop Res 1999; 17: 475-487.

15. Fortier LA, Nixon AJ, Lust G. Phenotypic expression of equine articular chondrocytes grown in three-dimensional cultures supplemented with supraphysiologic concentrations of insulin-like growth factor-1. Am J Vet Res 2002; 63: 301-305.

16. Strauss EJ, Goodrich LR, Chen CT, Hidaka C, Nixon AJ. Biochemical and biomechanical properties of lesion and adjacent articular cartilage after chondral defect repair in an equine model. Am J Sports Med 2005; 33: 1647-1653.

17. Sams AE, Nixon AJ. Chondrocyte-laden collagen scaffolds for resurfacing extensive articular cartilage defects. Osteoarthritis Cartilage 1995; 3: 47-59.

18. Hendrickson DA, Nixon AJ, Erb HN, Lust G. Phenotype and biological activity of neonatal equine chondrocytes cultured in a three-dimensional fibrin matrix. Am J Vet Res 1994; 55: 410-414.

19. Gratz KR, Wong VW, Chen AC, Fortier LA, Nixon AJ, Sah RL. Biomechanical assessment of tissue retrieved after in vivo cartilage defect repair: tensile modulus of repair tissue and integration with host cartilage. J Biomech 2006; 39: 138-146.

20. Watts AE, Nixon AJ. Comparison of arthroscopic approaches and accessible anatomic structures during arthroscopy of the caudal pouches of equine femorotibial joints. Vet Surg 2006; 35: 219-226.

21. Cucchiarini M, Madry H. Gene therapy for cartilage defects. J Gene Med 2005; 7: 1495-1509.

22. Sellers RS, Zhang R, Glasson SS, Kim HD, Peluso D, D'Augusta DA, et al. Repair of articular cartilage defects one year after treatment with recombinant human bone morphogenetic protein-2 (rhBMP-2). J Bone Joint Surg Am 2000; 82: 151-160.

23. Zachos T, Diggs A, Weisbrode S, Bartlett J, Bertone A. Mesenchymal stem cell- mediated gene delivery of bone morphogenetic protein-2 in an articular fracture model. Mol Ther 2007; 15: 1543-1550.

24. Steinert AF, Ghivizzani SC, Rethwilm A, Tuan RS, Evans CH, Noth U. Major biological obstacles for persistent cell-based regeneration of articular cartilage. Arthritis Res Ther 2007; 9: 213.

25. Ishihara A, Shields KM, Litsky AS, Mattoon JS, Weisbrode SE, Bartlett JS, et al. Osteogenic gene regulation and relative acceleration of healing by adenoviral-mediated transfer of human BMP-2 or -6 in equine osteotomy and ostectomy models. J Orthop Res 2008; 26: 764-771.

26. Ishihara A, Zekas LJ, Litsky AS, Weisbrode SE, Bertone AL. Dermal fibroblast- mediated BMP2 therapy to accelerate bone healing in an equine osteotomy model. J Orthop Res 2010; 28: 403-411.

27. Ishihara A, Zekas LJ, Weisbrode SE, Bertone AL. Comparative efficacy of dermal fibroblast-mediated and direct adenoviral bone morphogenetic protein-2 gene therapy for bone regeneration in an equine rib model. Gene Ther 2010; 17: 733-744.

28. Santangelo KS, Baker SA, Nuovo G, Dyce J, Bartlett JS, Bertone AL. Detectable reporter gene expression following transduction of adenovirus and adeno-associated virus serotype 2 vectors within full-thickness osteoarthritic and unaffected canine cartilage in vitro and unaffected guinea pig cartilage in vivo. J Orthop Res 2010; 28: 149-155.

29. Lee JH, Dyke JP, Ballon D, Ciombor DM, Rosenwasser MP, Aaron RK. Subchondral fluid dynamics in a model of osteoarthritis: use of dynamic contrast-enhanced magnetic resonance imaging. Osteoarthritis Cartilage 2009; 17: 1350-1355.

30. Burstein D, Gray M, Mosher T, Dardzinski B. Measures of molecular composition and structure in osteoarthritis. Radiol Clin North Am 2009; 47: 675-686.

31. Domayer SE, Trattnig S, Stelzeneder D, Hirschfeld C, Quirbach S, Dorotka R, et al. Delayed gadolinium-enhanced MRI of cartilage in the ankle at 3 T: feasibility and preliminary results after matrix-associated autologous chondrocyte implantation. J Magn Reson Imaging 2010; 31: 732-739.

32. White LM, Sussman MS, Hurtig M, Probyn L, Tomlinson G, Kandel R. Cartilage T2 assessment: differentiation of normal hyaline cartilage and reparative tissue after arthroscopic cartilage repair in equine subjects. Radiology 2006; 241: 407-414.

33. Evans CH. Gene therapy for bone healing. Expert Rev Mol Med 2010; 12: e18.

34. Bertone AL, Pittman DD, Bouxsein ML, Li J, Clancy B, Seeherman HJ. Adenoviral-mediated transfer of human BMP-6 gene accelerates healing in a rabbit ulnar osteotomy model. J Orthop Res 2004; 22: 1261-1270.

35. Watanabe A, Wada Y, Obata T, Ueda T, Tamura M, Ikehira H, et al. Delayed gadolinium-enhanced MR to determine glycosaminoglycan concentration in reparative cartilage after autologous chondrocyte implantation: preliminary results. Radiology 2006; 239: 201-208.

36. Blaney Davidson EN, Vitters EL, van Lent PL, van de Loo FA, van den Berg WB, van der Kraan PM. Elevated extracellular matrix production and degradation upon bone morphogenetic protein-2 (BMP-2) stimulation point toward a role for BMP-2 in cartilage repair and remodeling. Arthritis Res Ther 2007; 9: R102.

37. Potter K, Butler JJ, Horton WE, Spencer RG. Response of engineered cartilage tissue to biochemical agents as studied by proton magnetic resonance microscopy. Arthritis Rheum 2000; 43: 1580-1590.

38. Recht M, Bobic V, Burstein D, Disler D, Gold G, Gray M, et al. Magnetic resonance imaging of articular cartilage. Clin Orthop Relat Res 2001: S379-396.

39. Glansbeek HL, van Beuningen HM, Vitters EL, Morris EA, van der Kraan PM, van den Berg WB. Bone morphogenetic protein 2 stimulates articular cartilage proteoglycan synthesis in vivo but does not counteract interleukin-1alpha effects on proteoglycan synthesis and content. Arthritis Rheum 1997; 40: 1020-1028.

40. Grunder T, Gaissmaier C, Fritz J, Stoop R, Hortschansky P, Mollenhauer J, et al. Bone morphogenetic protein (BMP)-2 enhances the expression of type II collagen and aggrecan in chondrocytes embedded in alginate beads. Osteoarthritis Cartilage 2004; 12: 559-567.

41. Kim DJ, Moon SH, Kim H, Kwon UH, Park MS, Han KJ, et al. Bone morphogenetic protein-2 facilitates expression of chondrogenic, not osteogenic, phenotype of human intervertebral disc cells. Spine (Phila Pa 1976) 2003; 28: 2679-2684.

42. Li J, Yoon ST, Hutton WC. Effect of bone morphogenetic protein-2 (BMP-2) on matrix production, other BMPs, and BMP receptors in rat intervertebral disc cells. J Spinal Disord Tech 2004; 17: 423-428.

43. Summer-Smith G. Bone in Clinical Orthopedics. Volume 1. second Edition. Dubendorf, AO Publishing 2002.

44. van der Kraan PM, Davidson EN, van den Berg WB. Bone morphogenetic proteins and articular cartilage: To serve and protect or a wolf in sheep clothing's? Osteoarthritis Cartilage 2010; 18: 735-741.

45. Trippel SB, Ghivizzani SC, Nixon AJ. Gene-based approaches for the repair of articular cartilage. Gene Ther 2004; 11: 351-359.

46. Hidaka C, Goodrich LR, Chen CT, Warren RF, Crystal RG, Nixon AJ. Acceleration of cartilage repair by genetically modified chondrocytes over expressing bone morphogenetic protein-7. J Orthop Res 2003; 21: 573-583.

47. Steinert AF, Noth U, Tuan RS. Concepts in gene therapy for cartilage repair. Injury 2008; 39 Suppl 1: S97-113.

48. Evans CH, Ghivizzani SC, Robbins PD. Orthopedic gene therapy in 2008. Mol Ther 2009; 17: 231-244.

Appendix A: Direct Human Adenoviral BMP-2 or BMP-6 Gene Therapy for Bone and Cartilage Regeneration in a Pony Osteochondral Model

ABSTRACT

Objective: To evaluate healing of surgically created large osteochondral defects in a weight-bearing femoral condyle in response to percutaneous direct injection of adenoviral

(Ad) vectors containing coding regions for either human bone morphogenetic proteins 2

(BMP-2) or -6.

Methods: Four 13 mm diameter and 7 mm depth circular osteochondral defects were drilled, 1 per femoral condyle (n=20 defects). Direct injection of Ad-BMP-2, Ad-BMP-6,

Ad- green fluorescence protein (GFP), or saline into the defect was performed 14 days after surgery. Quantitative magnetic resonance imaging (qMRI) and computed tomography

(CT) were serially performed at 12, 24, and 52 weeks. At 52 weeks, histomorphometry and microtomographic analyses were performed to assess final subchondral bone and cartilage repair tissue quality.

Results: Direct delivery of Ad-BMP-6 into healing large femoral condyle lesions demonstrated dGEMRIC qMRI and histologic evidence of greater GAG-content in repair tissue at 12 weeks (p<0.05), while Ad-BMP-2 had greater nonmineral cartilage at the surface at 52 weeks. Ad-BMP-2 demonstrated greater CT subchondral bone mineral density (BMD) by 12 weeks and both Ad-BMP-2 and -6 had greater subchondral BMD at

52 weeks (p<0.05). Despite these observations of earlier (Ad-BMP-6) and persistent (Ad-

BMP-2) cartilage repair and greater subchondral bone regeneration (Ad-BMP-2 and -6),

89 the tissue within the large weight-bearing defects at 52 weeks was suboptimal in all groups due to poor quality repair cartilage, central fibrocartilage retention, and central bone cavitation. Delivery of either BMP by this method had greater frequencies of subchondral bone cystic formation (p<0.05).

Conclusions: Delivery of Ad-BMP-2 or Ad-BMP-6 to large weight-bearing osteochondral defects via direct injection provided evidence of support to cartilage and subchondral bone regeneration but was insufficient to provide long-term quality osteochondral repair in this femoral condyle pony model.

INTRODUCTION

Weight-bearing full thickness articular cartilage lesions in the femoral condyle occur frequently, are difficult to heal, negatively affect the surrounding cartilage and, often lead to degenerative arthritis and pain [1]. Currently, these lesions are managed by several surgical and cell-based approaches such as subchondral drilling, abrasion, micro-fracture

[2], osteochondral autograft transfer (OAT) [3], autologous chondrocyte implantation

(ACI) [4], matrix-induced autologous chondrocyte implantation (MACI) [5], and juvenile allogeneic chondrocyte implantation [6]. The long-term regeneration of hyaline cartilage achieved with these methods, however, is less than clinically acceptable in large weight- bearing osteochondral defects [7-9]. Current pre-clinical experimental procedures evaluating use of genes and cells, such as IGF and BMP-7, and cells, such as chondrocytes and stem cells, have shown promise to enhance cartilage healing [10, 11], but are limited in scope and have not been followed serially in a live animal model.

The equine cartilage defect repair model is particularly relevant to cartilage healing in humans. The thickness of equine cartilage, including the thickness of the calcified

90 cartilage and the subchondral bone plate, is closer to humans than any other animal used for experimental studies of cartilage repair [12, 13]. The biochemical, biomechanical and histological characteristics of equine repair and cartilage tissue are particularly well defined

[14-19]. The limitations of the equine model include body weight and size, which limit imaging modalities and can produce excessive load stress on joint surfaces, and are thus particularly important for articular cartilage weight-bearing defects. Our goal was to evaluate an equine pony model that more closely mimics the joint size and loading of humans and counteracts the traditional equine model size limitations. The pony model provided the opportunity for serial evaluations using available imaging modalities such as clinical magnetic resonance imaging (MRI) and computed tomography (CT). In the equine species, the medial and lateral femoro-tibial joints are non-communicating compartments

[20].

Due to the fact that there is currently no regimen, either pharmacological or surgical, that is capable of restoring damaged cartilage to its normal phenotype, there is a good incentive for pursuing alternatives, such as gene transfer. Among the list of potentially useful cDNAs for cartilage repair are anabolic growth factors of the transforming growth factor (TGF)-b superfamily, including several bone morphogenetic proteins (BMPs) [21, 22]. There is evidence that BMP-2 and BMP-6 can be chondrogenic both in vitro and in vivo in animal models [23]. It is well known that both BMP-2 and

BMP-6 are potently osteogenic and can enhance and accelerate bone regeneration in vivo

[24-27]

Our hypothesis was that BMPs may provide improved subchondral bone formation in healing large full-thickness weight-bearing articular cartilage defects, as well as support

91 chondrogenesis and restoration of the articular cartilage surface. We selected the use of several quantitative serial imaging modalities to assess bone and cartilage restoration over time and to evaluate intra-lesional delivery of BMP-2 and BMP-6 gene therapy.

MATERIALS AND METHODS

Study Design

Procedures performed for this study were reviewed and approved by the Institutional

Animal Care and Use Committee at The Ohio State University. Five healthy, skeletally mature ponies (weighting 121-161 kg) underwent general anesthesia induced by xylazine

(Rompun; Bayer, Pittsburgh, PA, USA; intravenously (i.v.), 1.1 mg kg�1), ketamine

(Ketaset; Fort Dodge Animal Health, Overland Park, KS, USA; i.v., 2.2 mg kg�1) and diazepam (Valium; Roche, Madison, WI, USA; i.v., 0.11 mg kg�1) and maintained by

Isoﬂurane (IsoFlo; Abbott, Parsippany, NJ, USA; infusion, 2–5%). The ponies were positioned in dorsal recumbency and via arthrotomy, four 13 mm diameter and 7 mm depth circular osteochondral defects were drilled, 1 per femoral condyle (n= 20). Additionally, a

2.5 mm in diameter and 10 mm depth was drilled within each lesion (Fig 1). Two weeks after the surgery, Ad containing green fluorescent protein (GFP), Ad-BMP2, and Ad-

BMP6 vectors were diluted into a 500 µl total volume with GBSS containing 4 x109 IFU

(Ad-BMP-2) and 2.4 x 109 (Ad-BMP-6). Under general anesthesia, each pony had the osteochondral femoral lesions treated with percutaneous direct fluoroscopic guided injections of Ad-BMP2 (n=5), Ad-BMP6 (n=5), Ad-GFP (n=5) or Gey’s balanced salt solution (n=5; GBSS -Sigma-Aldrich, St Louis, MO, USA). The four defects in each pony were assigned in a block design to rotate the treatments. At 12 or 52 weeks, after undergoing qMRI, the ponies were euthanized. Efficacy was assessed by qMRI, clinical

CT, ex vivo micro-tomography (CT), histology and gene expression analysis. Evidence of systemic biodistribution of vector was assessed by physical examination and quantitative

PCR for the CMV promoter region of the Ad in multiple organ tissues.

Ad vector production and treatments

Recombinant, replication-deﬁcient, serotype-5 Ad vectors containing either a 1547 base- pair open reading frame segment of human BMP-2 (Ad-BMP-2), BMP-6 (Ad-BMP-6) or

GFP (Ad-GFP) under the control of the cytomegalovirus (CMV) promoter were generated as previously reported [23, 25-27]. The expression of transgenes were veriﬁed in cell culture [23]. The infectious units per milliliter (IFU/ml) of Ad-BMP-2, Ad-BMP-6 and Ad-

GFP were calculated according to manufacturer’s instructions using a commercially available kit (Clontech Laboratories, Inc., Mountain View, CA)[28].

Quantitative Magnetic Resonance Imaging

At 12, 24, and 52 weeks after the osteochondral lesions were treated, under general anesthesia, in a 3 Tesla MRI (Achieva, Philips, Cleveland, Ohio, U.S.A.), quatitative MRI

(qMRI) was performed using a transmit quadrature body coil in combination with a 4- channel array of 10 cm loop coils per knee.

Dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI) was performed by administering a bolus injection of double dose (0.2mmol/kg) post-gadopentate dimeglumine (Gd-DTPA; Magnevist ®, Wayne, NJ)) while acquiring a 3D T1 weighted turbo field echo (T1-TFE) sequence (TR/TE=3.15/1.60 ms; flip angle=12⁰; TFE factor=50;

FOV = 64x180x180mm3; matrix=32x120x120; slice thickness=4mm; 30 dynamic scans,

15.14 s per scan). Dynamic parameters were calculated by fitting to a pharmacokinetic

93 modified Brix model with arterial input function sampled from the popliteal artery using

Levenberg-Marquardt [29, 30]

Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC): A sagittal slice through each defect was imaged via a multi-inversion recovery turbo spin echo (IR-TSE) sequence

(TR/TE = 3740/28 ms; TSE factor=10; FOV= 165 x 165 mm2; matrix = 332x328; slice thickness=3mm). Six acquisitions were taken of each slice with varied inversion times (0,

60, 150, 350, 1100, 1680 ms). Post-Gd-DTPA imaging was taken after 30 minutes of passive exercise following injection. This protocol was adapted from human clinical patients [31]. T1 values were calculated by performing Levenberg-Marquardt least-squares fit.

T2 mapping: A multi-echo TSE sequence (TR/TE = 3000 /10, 20, 30, 40, 50, 60, 70, 80 ms; FOV=165x165 mm2; matrix = 164x165; slice thickness=3mm) was performed on each defect. T2 values were calculated via linear least-squares fit [32]. All calculations were performed using in-house software written in the IDL environment (Interactive Data

Language, ITT Visual Information Systems, Boulder, CO). Regions of Interest (ROIs) outlining the osteochondral lesion and adjacent non-injured cartilage located 1cm distant from the lesion in a sagittal cross section were manually traced by two authors blinded to treatments (MIM, DJC). Outcome measurements included: T1Gd (after gadopentate dimeglumine administration) for dGEMRIC, T2 relaxation time for T2 mapping and

Amplitude (relative intensity of contrast enhancement after injection) for DCE-MRI.

Live Pony Computer Tomography

CT was performed at 12 and 24 weeks after treatment under general anesthesia with a

Lightspeed 3X (GE Healthcare, Wisconsin, USA). The images were acquired in bone

94 algorithm and reconstructed in detail (soft tissue) algorithm (170 mA, 140 kVp, 40 cm

FOV) using a transverse scan; 0.625 mm contiguous images were scanned. The image was comprised from tissue just proximal to the patella to 2 cm distal to the tibial plateau. The ponies were positioned in left lateral recumbency with pelvic limbs in extension (feet first through gantry). A dipotassium phosphate phantom was placed on top of right stifle. Data was registered and analyzed within the Inveon Research workplace (Siemens, Knoxville,

TN). ROI’s were traced for the osteochondral lesions and surrounding subchondral bone without knowledge of the treatment groups (MIM, MC). Outcomes quantified included:

Bone mineral density (BMD) in osteochondral lesions and BMD in surrounding subchondral bone.

Quantitative Micro-Computed Tomography

Fifty-two weeks after vector injection, and following MRI, ponies were sedated with xylazine and euthanized by lethal intravenous overdose of pentobarbital (Beuthanasia; i.v.,

2.2 mg/kg). Immediately after the euthanasia, tissues were harvested and the femoral condyles were photographed and dissected. Quantitative (q) CT (Inveon, Siemens,

Knoxville, TN) was analyzed using a cropped Siemens format, registered to TIFF format in RATOC (Japan) to provide correct alignment of the lesions. Data was converted to a

RAW volume and imported to Inveon Research Workplace as DICOM images to generate the ROIs. The threshold level for cartilage was established at a RAW value of 32000 units which corresponds to 874 mg/cc absolute density and -277 mg of Ca/cc collagen. Outcome measurements included lesion area (mm2) and BMD for the lesion, drill and adjacent subchondral bone ROIs (Figures 5 and 6).

Gross Photograph Evaluation

Standardized digitized color photographs were taken for gross assessment of the surface.

The repair tissue surface was semiquantitatively scored by two authors (MIM, ALB) blinded to treatment from 0 (not present) to 4 (present in > 75% of the area) for perimeter gap, surface irregularity and cartilage recess [11]. The location and presence of central cavitations was noted and expressed as a frequency by group.

Histological Evaluation

Sagittal cross-sectioned blocks of femoral condyle osteochondral defects were fixed with

10% neutral-buffered formalin and decalcified in formic/hydrochloric acid (8% solution of each). Specimens were paraffin-embedded, sectioned at 5 µm, and stained with H&E,

Toluidine Blue and Safranin-O Fast Green. Sections were examined by three investigators blinded to lesion treatments (SEW, ALB, MIM). Outcome parameters included chondrocyte cloning, hypocellularity, cartilage repair thickness, surface irregularity and safranin-O staining [11](Table 1, Figure 5)

Gene expression analyses via reverse transcription PCR (RT-PCR)

Two-step RT-PCR was performed using RNA extracted from intra-lesional tissue, surrounding cartilage, synovial membrane, and liver. Total RNA was extracted by TRIzol

(Invitrogen Life Technologies, Carlsbad, CA) from fresh homogenized tissues using established protocol [25] and cDNA was constructed using Taqman® Reverse

Transcription Reagents (Applied Biosystems, Foster City, CA). Primers for the hbmp2

(forward: AAAACGTCAAGCCAAACACAAA; reverse:

GTCACTGAAGTCCACGTACAAAGG), hbmp6 (forward:

CAACAGAGTCGTAATCGCTCTAC: reverse: TTAGTGGCATCCACAAGCTCT) and

GFP gene (forward: CATGATATAGACGTTGTGGCTGTTG-3; reverse:

AAGCTGACCCTGAAGTTCATCTGC) were used in Ad-GFP, Ad-BMP-6, and Ad-

BMP-2 treated tissue to detect presence of active transcription.

Biolocalization of Adenoviral Vectors

Genomic DNA was extracted from tissue samples, including cartilage surrounding the lesion, using the QIAamp Mini Kit (Qiagen ®, Valencia, CA). PCR was performed using primers (forward: CTGGCTGACCGCCCAACGAC; reverse:

CACCGTACACGCCTACCGCC) that amplify CMV sequences.

Statistical Analysis

Repeated-measure analysis of variance (ANOVA) (SAS Institute Inc., Cary, NC) was used to evaluate the effects of Ad-BMPs gene therapy using Proc Mixed models for continuous outcomes (i.e., MRI, micro-CT) and Genmod models for categorical outcomes (i.e., histologic data). Repeated variables were considered to be nested within horses, and the distribution of data was assessed by the use of a subset of normality tests (such as the

Shapiro-Wilk, Kolmogorov-Smirnov, Cramer-von Mises and Anderson-Darling tests).

The post-test multiple comparisons were made between the treatment groups, among time points, or between the lesion and control sites (i.e., MRI). The frequency outcomes in histology (i.e., cystic formation, cartilage contiguity) were compared between the Ad-

BMP-2/6 and non-Ad-BMP-2/6 groups using Chi-square test. Significance level was set at

P<0.05 for all analyses.

RESULTS

Outcomes were completed on femurs of one pony at 12 weeks and remaining ponies at 12,

24, and 52 weeks.

Magnetic Resonance Imaging

Dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI): Amplitude in the lesion was significantly greater at 12 weeks than 24 and 52 weeks, decreased at 24 weeks, and was negligible by 52 weeks (P<0.05). Adjacent un-injured cartilage had negligible values at 12, 24 and 52 weeks (Figure 2).

Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC): Lesions treated with Ad-

BMP-6 had a greater T1Gd relaxation time (GAG content) at 12 weeks (P<0.05) than GBSS treatment. The adjacent non-injured cartilage had a significant lower T1Gd value than lesion repair tissue for all time points, 12, 24 and 52 weeks. At 52 weeks T1Gd values (GAG content) were lower in both the lesion and the adjacent cartilage than at 24 weeks (P<0.05)

(Figure 2).

T2 mapping: T2 relaxation time was significantly greater in the lesion than adjacent cartilage (P<0.05) without differences among groups or time (Figure 2).

Live Pony Computed Tomography

The subchondral bone surrounding the lesions treated with Ad-BMP-2 at 12 weeks had greater bone mineral density than the GBSS group (p=0.038). Additionally, the lesion, as well as the surrounding subchondral bone, had an increase in bone mineral density from

12 to 24 weeks (p<0.05) (Figure 3).

Quantitative micro-Computed Tomography

The non-mineral ROI (mm2) for the BMP-2 group was significantly greater than the GBSS group (p<0.05) (Figure 5, 6). The BMD within the drill ROI was significantly greater in the BMP-6 treated lesion than the GBSS group (p<0.05) (Figure 4, 5).

Gross photograph Evaluation

Perimeter gap score was lower in lesions treated with Ad-BMP-2 than control groups

(GBSS and Ad-GFP) (p<0.05). BMP-treated lesions had a greater frequency of central cavitation formation than non-BMP-treated lesions (Table 1, Figure 5) (P<0.05).

Histological Evaluation

Histomorphometry at 52 weeks showed greater chondrocyte cloning and chondrone formation at the integration of the lesion and the adjacent un-injured cartilage in the lesions treated with Ad-BMP-2 and Ad-BMP-6 (p<0.05) (Figure 5). The frequency of subchondral bone cystic formation was greater in the lesions treated with BMPs (p<0.05) than in lesions treated without BMPs. Irregular surface score was lower for Ad-BMP6 treated group than

GBSS group (p<0.05) (Figure 5, Table 1)

Histology at 12 weeks demonstrated positive safranin-O staining within the surface repair tissue of the Ad-BMP-6 group and not present in other groups. (Figure 1)

Gene Expression Analysis

RT-PCR at 12 weeks detected gene expression of BMP-2 and GFP within the lesions and

GFP in adjacent cartilage (Figure 6). RT-PCR at 52 weeks detected gene expression of

GFP in one sample of adjacent cartilage.

Biolocalization of Adenoviral Vectors

DNA from the CMV promoter portion of the Ad vector was detected in surrounding cartilage at 52 weeks in BMP-2, BMP-6 and GFP specimens in nine out of twelve possible positive samples (Figure 6).

DISCUSSION

Single, delayed injection of adenovirus encoding BMP-2 or BMP-6 genes was beneficial in the short term to accelerate GAG containing repair cartilage and/or subchondral bone

99 density in large, weight bearing surgically created osteochondral lesions. Delayed direct vector injection (2 weeks after surgery) into solid granulation tissue has been shown to be advantageous to contain the Ad-BMP in an osteotomy healing site [33, 34]. Presumably, transcribed and secreted BMP functions in the local region and targets host cells [25-27].

To our knowledge, our study is the first to show this effect in osteochondral healing sites.

Our central drill hole provided a site within the lesion for localized delivery of gene transfer vectors to provide sustained and concentrated protein synthesis at the site of the created lesion. Limited spillover of vector to adjacent cartilage was expected and confirmed. Our study detected the CMV promoter portion of the Ad vector at 52 weeks in adjacent cartilage documenting the injection of the vector near the osteochondral lesion and persistence of adjacent cartilage cells for over one year. It has been demonstrated that Ad vectors successfully transduced chondrocytes within the surface of canine and guinea pig cartilage explants and, in vivo, Ad-GFP expression persisted for 12 weeks in 5 of 24 specimens [28], although is known to have transient expression [28].

Our study demonstrated greater T1Gd values in the Ad-BMP-6 group than in the GBSS control (p<0.05) early in lesion healing (12 weeks). This imaging finding was confirmed by histology. Because GAG is composed of abundant carboxyl and sulfate groups, it is negatively charged within the cartilage matrix. Anionic gadopentate dimeglumine

(Magnevist®, Wayne, NJ), given a sufficient time after its injection to penetrate the cartilage, will distribute inversely to the concentration of negatively charged cartilaginous

GAG [35]. The protocol for administration of Gd in the ponies was similar as used in human clinical patients with suspected cartilage injury [31]. BMPs may have an early effect in chondrogenesis. It has been shown that over-expression of BMP-2 by Ad vector gave

100 rise to increased proteoglycan synthesis by more than 300% compared with normal turnover proteoglycan synthesis [36]. The limitations of small size and site morbidity of using biopsy specimens for GAG assessment can be avoided by use of MRI which can provide information on the whole cartilage surface. In addition, MRI is noninvasive and serial scans can be performed, allowing longitudinal monitoring at different time points.

MR images have been shown to correlate with biochemical composition in other tissues, in cartilage in vivo, and in engineered cartilage [37, 38].

Quantitative CT confirmed greater bone mineral density in the subchondral bone of BMP-

6 and BMP-2 lesions. Additionally, at 52 weeks, BMP-2 lesions had a greater non-mineral lesion area, indicating, and supported by histology, a greater amount of cartilage repair tissue centrally at the surface of the lesion (Figure 5). In the literature, BMP-2 is proposed as a stimulant for cartilage regeneration. BMP-2 is able to stimulate proteoglycan synthesis in murine cartilage and enhances collagen type II expression in chondrocytes seeded in alginate [39, 40]. Also in rats and humans, BMP-2 is able to stimulate the chondrogenic phenotype on the mRNA level and to stimulate cartilage extracellular matrix proteoglycan production [41, 42]. Furthermore, BMP-2 boosts matrix turnover in intact and interleukin- damaged cartilage, contributing to the intrinsic repair capacity of damaged cartilage [36].

Our imaging data supported that both BMP-2 and -6 may enhance chondrogenesis as well as osteogenesis when injected directly into the defect using an Ad vector. Concomitantly, both BMP-2 and BMP-6 had greater integration of the repair tissue with surrounding cartilage both grossly and histologically. (Figure 5) Interestingly, a dramatic and significantly greater density of chondrocyte cloning occurred in this repair cartilage tissue/adjacent cartilage interface suggesting a trophic effect of the BMPs and resultant

101 chondrocyte proliferation (Figure 5). The result was a solid cartilage tissue within the lesion/adjacent cartilage interface likely translating to a more stable weight-bearing osteochondral repair. However, regardless of these composite findings that Ad-BMP-2 and/or Ad-BMP-6 supported cartilage repair and denser subchondral bone in healing osteochondral defects, this did not translate into effective, clinically acceptable healing of the defect. Long-term assessment of Ad-BMP-2/6 delivery by this method resulted in greater central cavitation of the repair tissue and subchondral bone cystic formation. BMPs may promote cystic formation due to acceleration of bone remodeling, which includes bone absorption prior to bone deposition [43]. Bone resorption on a weight-bearing surface area may predispose to cavitation due to pressure on loading. Further investigation of alternate strategies to deliver BMPs to osteochondral defects, such as use of other vectors, cell delivery or recombinant protein delivery in scaffolds, with or without cells, would be warranted.

In MRI, as expected, the lesion, compared to adjacent articular cartilage, had lower T1Gd values, thus, lower GAG content than the adjacent un-injured cartilage. Interestingly, at 52 weeks the un-injured adjacent cartilage was significantly lower than 12 and 24 weeks suggesting loss of GAG throughout the cartilage in knees with chronic poor healing defects. This may represent early signs of arthritic changes and OA degeneration in the knee as a whole. Other considerations for this effect include the observation by Van der

Kraan et al that elevated BMP levels in damaged cartilage can on one side contribute to tissue repair by boosting matrix synthesis but on the other side stimulate cartilage degeneration by altering chondrocyte behavior and stimulating MMP-13 expression. These effects could be protective for articular cartilage but can also have a harmful role [44].

102

Currently used cartilage repair techniques, both experimental and clinical, are still unable to generate a repair tissue that is comparable to the native cartilage tissue quality and stability, particularly in large defects on the weight-bearing surface of the knee. Although short-term success in generating hyaline cartilage repair tissue after six weeks in vivo has been demonstrated, long-term in vivo analyses indicate only unsatisfactory results [45-47].

Hidaka et al conducted a cartilage repair study in horses using chondrocytes transduced with Ad-BMP-7. Compared to defects receiving unmodified cells, healing was accelerated at 4 week. However, by 8 months this difference had disappeared and few donor cells could be detected in the repair tissue [46]. Contrarily, Cook et al. showed that a collagen/BMP-

7 construct was able to induce a hyaline-like repair of full thickness defects in a canine osteochondral model and to maintain repair over the course of one year. In addition, acceleration of subchondral bone filling was noted in collagen/BMP-7 treated defects at early-time periods, which may have aided repair [11]. In cell delivery of BMPs, donor cell survival may be an issue for cartilage repair, unlike bone where cell turnover is part of the natural physiology of the tissue [48]. In collagen constructs, control groups are collagen alone and comparison to untreated defects was not made.

Healing of full-thickness large osteochondral defects on the femoral condyle of larger species, including human, are notoriously difficult to obtain a long-term clinically effective repair tissue. Our study demonstrated some evidence for BMP-2 or -6 to provide potential benefit to cartilage and subchondral bone healing, but the method of gene delivery, as well as location and size of the defects may have contributed to the lack effective healing. For a successful gene therapy approach for cartilage repair, the mode of delivery, level and duration of transgene expression, as well as the type and dosage of vectors used have to be

103 well considered [45, 47] Alternatives such as use of adeno-associated viral vectors, use of cells or scaffolds as delivery vehicles, and various targeting strategies may enhance the biologic response to BMPs in this location.

Our study is the first to report the use of ponies as a more relevant equine model to successfully evaluate serial healing osteochondral defects in living subjects. Using serial, non-invasive in vivo imaging modalities (MRI and CT) offer the advantage to follow osteochondral regeneration within individual subjects across time, avoiding invasive techniques such as biopsies. The pony model had greater relevance to human medicine due to similarities in cartilage thickness and knee size between the species.

Although delivery of BMP-2 or BMP-6 to large weight-bearing osteochondral defects via adenoviral vector and direct injection provided evidence of support to cartilage and subchondral bone regeneration, it was insufficient to provide long-term quality osteochondral repair in the pony model.

104

Table A.1. Histomorphometry and gross photograph osteochondral parameters A) Histomorphometry

Parameter Treatment Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Chondrocyte cloning score [median (range)] GBSS 0 (0-1) 0 (0-2) 0 (0) 0 (0) 0 (0) 0 (0) Ad-GFP 1.5 (0.5-2.5) 1.5 (0-2) 0 (0) 0 (0-0.5) 0 (0) 0 (0) Ad-BMP2 2.0 (1-3.5)* 0.5 (0-0.5) 0 (0) 0 (0-0.5) 0 (0) 0 (0) Ad-BMP6 1.0 (1-3)* 0.3 (0-2) 0 (0) 0 (0-2) 0 (0) 0 (0)

Hypocellularity score [median (range)] GBSS 3.0 (1.5-4) 0 (0-2) 0 (0-0.5) 0 (0-0.5) 0 (0) 0 (0) Ad-GFP 2.0 (1.5-3.5) 0 (0) 0 (0-0.5) 0 (0-1) 0 (0) 0 (0) Ad-BMP2 3.5 (2-4) 0 (0-1) 0.3 (0-1) 0.5 (0-1) 0 (0) 0 (0) Ad-BMP6 3.3 (3-3.5) 0.8 (0-1.5) 0 (0-0.5) 0 (0) 0 (0) 0 (0)

Thickness repair (%) [mean±SEM] GBSS 58.3 ± 9.3 100.0 ± 19.0 51.3 ± 19.6 0 ± 0 0 ± 0 100.0 ± 0 Ad-GFP 71.7 ± 15.1 81.0 ± 8.3 78.3 ± 12.8 0 ± 0 0 ± 0 100.0 ± 0 Ad-BMP2 67.5 ± 10.1 114.5 ± 28.9 80.8 ± 10.6 0 ± 0 0 ± 0 100.0 ± 0 Ad-BMP6 63.8 ± 5.2 103.3 ± 15.0 80.0 ± 31.0 0 ± 0 0 ± 0 100.0 ± 0

Surface irregularity (index) [median (range)] GBSS 0.1 (0-0.3) 0.8 (0-1.4) 0 (0) N/A N/A 0 (0) Ad-GFP 0 (0-0.3) 0 (0-1.1) 0 (0-0.2) N/A N/A 0 (0) Ad-BMP2 0.3 (0-9) 0.9 (0-2.8) 0.3 (0-2) N/A N/A 0 (0) Ad-BMP6 1.6 (0.3-3.3) 0.5 (0-1.5) 0.2 (0-0.5) N/A N/A 0 (0)

Safranin-O staining score [median (range)] GBSS 0.6 (0-1.6) 1.2 (0.6-2.2) 1.2 (0.3-1.6) 1.7 (0.5-2.6) N/A 3.5 (2-4) Ad-GFP 1.5 (0.2-3) 2.4 (0.2-3) 0.6 (0.2-1.4) 1.5 (0.4-3.3) N/A 4.0 (3.5-4) Ad-BMP2 0.6 (0.3-3) 1.0 (0.9-2.2) 1.2 (0.1-1.8) 2.6 (0.9-2.8) 0.7 (0.2-1.2) 2.3 (0.8-2.9) Ad-BMP6 0.5 (0.2-1.5) 2.2 (0.9-2.7) 1.0 (0-2.1) 0.3 (0-3.4) 0.1 (0.1) 3.9 (1.5-4)

B) Gross photograph parameters

Parameter GBSS Ad-GFP Ad-BMP2 Ad-BMP6 Central cavitation frequency 1 / 4 0 / 4 1 / 4 2 / 4 Perimeter gap score [median (range)] 4 (3-4) 2 (2) 1 (0-1)* 2 (1-4) Irregular surface score [median (range)] 3 (3-4) 2.5 (1-4) 1.5 (1-4) 1.5 (0-3)* Estimate of recessed score [median 1.5 (1-2) 1 (1-4) 0.5 (0-2) 1.5 (0-3) (range)] *Differs from GBSS (p< 0.05) SEM: Standard Error of the Mean See Figure 6 (L) for zone number reference.

105

Figure A.1. Images of osteochondral femoral defects in the pony model. Initally at surgery (A) and at 12 weeks in sagittal section in 3D T2 weighted MRI (B), live pony clinical CT (C), micro-CT (D), gross photograph (E) and Safranin-O histochemistry of Ad-GFP (F) and Ad-BMP6 (G). Positive safranin-O staining in the defect on histology of the Ad-BMP-6 treated defect supported the MRI findings in Figure 2 of greater T1Gd relaxation time (GAG content) in the BMP-6 treated lesion.

106

Figure A.2. Representative quantitative MRI maps of sagittal sections through the osteochondral defect, showing a lesion treated with Ad-BMP-6 across time using three different quantitative techniques (T2 mapping, dGEMRIC and DCE-MRI). (A) 3D T2- weighted shows the anatomy and subchondral bone changes (edema) in the distal femoral condyles, and also displays the traced lesion and adjacent cartilage ROIs. T2 mapping showed a greater T2 relaxation time in the treated lesion than the adjacent cartilage in all time points (p < 0.05). On dGEMRIC, T1Gd is significantly lower in the the treated osteochondral lesions than the adjacent cartilage (p < 0.05). For color map DCE-MRI, Amplitude was greater (p < 0.05) in the lesion at 12 weeks than either 24 and 52 weeks. Amplitude decreased at 24 weeks and was almost negligible at 52 weeks, similar to the adjacent cartilage. As expected, the adjacent cartilage values were almost negligible across time. The osteochondral lesion decreased in size over time, a;nd subchondral bone changes were evident from 12 weeks. (B) At 12 weeks, dGEMRIC shows greater T1Gd relaxation time (GAG content) in the BMP-6 treated lesion than GBSS (p < 0.05). (C) Mean T2 relaxation time, T1Gd relaxation time and Amplitude from the osteochondral lesion and the adjacent cartilage at 12, 24, and 52 weeks. Asterisks (*) showed significant difference between Ad-BMP-6 and GBSS. Different letters (ab) differ significantly among time points (p < 0.05) combining all treatment groups. #Represent significant difference between lesion and adjacent cartilage (p < 0.05) combining all treatment groups.

107

Figure A.2.

108

109

110

Figure A.5. Post-mortem representative micro-CT and histomorphometry for osteochondral lesions (OL) treated with GBSS, Ad-GFP, Ad-BMP-2 and Ad-BMP-6 at 52 weeks. The micro-CT shows greater area of non-mineral tissue (mm2) (Figure 4) for the Ad-BMP-2 treated lesions (p<0.05) (B) as well as changes in the subchondral bone, interpreted and corroborated by histology as a cyst (D). Three-dimensional images of the OL showing subchondral bone are represented as insets in the right corners. Gross photographs (E-H) demonstrated that lesions treated with BMPs had lower perimeter gap, better integration, and less irregular surface (F, G) than GBSS, which presented higher frequency in central cavitation formation (E). Safranin-O staining at 10X reflected a greater chondrocyte cloning where the lesion integrated with the adjacent un-injured cartilage (zone 1) in lesions treated with BMP-2 and BMP-6 (M,N). Numbers in figure L represent the zones used to evaluate histomorphometry (see Table 1).

111

112