Analysis of the pathogenesis and progression of osteoarthritis in canine stifle joints considering three bone healing markers

(1)

Analysis of the pathogenesis and progression of osteoarthritis in canine stifle joints considering three bone healing markers

INAUGURAL-DISSERTATION

To obtain the academic degree - Doctor medicinae veterinariae -

(Dr. med. vet.)

Submitted by

Hanna Ursula Diekmann Haselünne

Hannover 2018

(2)

Academic supervision: 1. Univ.-Prof. Dr. F.-J. Kaup,

German Primate Center, Göttingen

2. External supervisor:

Prof. Dr. S. Neumann,

Small Animal Clinic of the Institute of Veterinary Medicine, Göttingen

1. Referee: Univ.-Prof. Dr. F.-J. Kaup

2. Referee: Univ.-Prof. Dr. M. Fehr

Day of the oral examination: October 25^th, 2018

(3)

Dedicated to Timo and my family

In memory of my grandfather and great-grandfather

(4)

(5)

1. Introduction ... 11

2. Literature ... 13

2.1. Osteoarthritis ... 13

2.2. Bone healing ... 17

2.3. Comparison of fracture healing and osteoarthritis ... 19

2.4. Biochemical markers ... 20

2.4.1. RANKL / sRANKL ... 21

2.4.1.1. Biology of RANKL ... 21

2.4.1.2. RANKL in fracture healing ... 22

2.4.1.3. RANKL in osteoarthritis ... 23

2.4.2. Chordin ... 23

2.4.2.1. Biology of chordin ... 23

2.4.2.2. Chordin in fracture healing ... 24

2.4.2.3. Chordin in osteoarthritis... 25

2.4.3. Osteocalcin ... 26

2.4.3.1. Biology of osteocalcin ... 26

2.4.3.2. Osteocalcin in fracture healing ... 27

2.4.3.3. Osteocalcin in osteoarthritis ... 28

3. Material and methods... 30

3.1. Study design ... 30

3.2. Study population ... 31

3.2.1. Deceased dogs ... 31

3.2.2. Patients in surgery ... 32

(6)

3.3. Clinical examination and further diagnostics ... 33

3.3.2. Joint surgery patients ... 33

3.4. Staging of osteoarthritis severity by radiographs ... 34

3.5. Evaluation of osteoarthritis severity by intraarticular view during surgery.... 40

3.6. Sample collection and preparation ... 41

3.6.2. Patients in surgery ... 42

3.7. Quantification of sRANKL, osteocalcin and chordin by enzyme linked immunosorbent assay (ELISA) ... 42

3.7.1. Assay procedure ... 42

3.7.2. Calculation of sample concentrations ... 43

3.8. Statistical Analysis ... 44

3.9. Equipment and consumables ... 45

3.9.1. Equipment ... 45

3.9.2. Consumables ... 46

3.9.3. Solutions and reagents ... 46

4. Results ... 48

4.1. Analysis of synovial fluid ... 48

4.1.1. sRANKL ... 48

4.1.2. Chordin ... 55

4.1.4. Deceased dogs and surgery patients ... 68

4.2. Analysis of serum ... 70

4.2.1. sRANKL ... 70

4.2.2. Chordin ... 76

(7)

4.3.1. sRANKL ... 88

4.3.2. Chordin ... 90

4.4. Comparing evaluation techniques of OA staging: radiographic evaluation and intra-articular evaluation ... 94

4.5. Results of clinical evaluations and the radiographic grade of osteoarthritis . 95 4.5.1. Degree of lameness ... 95

4.5.2. Duration of lameness or the suspected distance to stifle trauma ... 96

4.6.1. Age ... 97

4.6.2. Weight ... 100

4.6.3. Sex ... 102

4.6.4. Breed ... 103

5. Discussion ... 106

5.1. Concentrations of markers ... 107

5.1.1. sRANKL ... 107

5.1.2. Chordin ... 110

5.1.4. Purebreds and hybrids ... 114

5.2. Comparison of evaluation techniques of OA staging: radiographic evaluation and intra-articular evaluation ... 115

5.3. Clinical investigations ... 115

5.5. Other limitations ... 118

(8)

6. Prospects ... 120

7. Summary ... 121

8. Zusammenfassung... 123

9. References ... 126

10. Appendix ... 144

10.1. Figure index ... 144

10.2. Table index ... 149

10.3. Study population: data... 150

11. Acknowledgements ... 159

(9)

ACLT Anterior cruciate ligament transection

ADAMTS A disintegrin and metalloproteinase with thrombospondin motifs

BGP Bone Gla protein

BMP Bone morphogenetic protein

CHRD Chordin

cf. Confer (compare)

cm Centimetre

CV Coefficient of variation

Cys Cysteine

e.g. Exempli gratia (for example)

ED Elbow dysplasia

ELISA Enzyme linked immunosorbent assay

et al. Et alii

FGF Fibroblast growth factor

FPC Fragmented processus coronoideus Gla γ-carboxyglutamic acid

HD Hip dysplasia

HRP Horseradish peroxidase

IGF Insulin-like growth factor

IL Interleukin

kg Kilogram

mm Millimetre

NC North Carolina

ng Nanogram

OA Osteoarthritis

OC Osteocalcin

OPG Osteoprotegerin

OPGL Osteoprotegerin ligand

(10)

PBS Phosphate buffered saline PCR Polymerase chain reaction

pg Picogram

pH Potential hydrogenii

RANK Receptor activator of nuclear factor-kappaB

RANKL Receptor activator of nuclear factor-kappaB ligand RCCL Rupture of the cranial cruciate ligament

SDS Sodium lauryl sulfate

sRANKL Soluble receptor activator of nuclear factor-kappaB ligand TGF Transforming growth factor

TMB 3,3’,5,5’-Tetramethylbenzidine

TNF Tumor necrosis factor

TRANCE TNF-related activation-induced cytokine

UAP Ununited anconeal process

USA United States of America

VEGF Vascular endothelial growth factor

x g Times gravity, equivalent to relative centrifugal force (RCF)

(11)

1. Introduction

Osteoarthritis (OA) is a common disease that affects animals as well as humans. It is the most common joint disease in mature humans (RABENBERG 2013) and dogs (MELE 2007) worldwide. Although there are various therapeutic strategies to reduce osteoarthritic symptoms, no curative therapy is available. OA has been the subject of extensive research over the last years, but triggers of this disease still remain unknown. Therefore, no satisfying treatment has been found so far to stop OA in the early stages.

In order to help OA patients many medical and physical therapies have been applied to reduce pain and to slow down disease progression. Available therapeutic strategies to reduce symptoms are, for example, non-steroidal antiphlogistics, chondroprotective agents or hyaluronic acid. Physiotherapy and mild but continuous exercise can help to keep the joint flexible and to strengthen muscles and tendons.

These methods can be supportive, but in some cases additional surgery is necessary. This means higher risks for patients and increased expenses. Altogether, even if one or more kinds of treatment can reduce symptoms, these therapies can hardly replace the damaged cartilage or stop osteophyte growth, except for total joint replacement which is strongly invasive. More important, the existing treatments listed here are predominantly therapies for advanced osteoarthritis, as there is no satisfying treatment to heal osteoarthritis in early stages before damage is caused. But of crucial importance for the cognition of OA in early stages are authentic diagnostic parameters which have not been established yet. Symptoms like pain and lameness typically manifest when osteophytic building and cartilage degradation already have taken place. At this point, osteoarthritis can be diagnosed by imaging techniques like radiographs, computer tomography, magnetic resonance imaging or by arthroscopy.

Thus, there is still a lack of early diagnosing techniques or parameters and, consequently, of OA stopping treatment in the early stages.

Considering the mechanisms that characterize fracture repair, which is another type of a bone affecting process, it seems that disease progressions of bone healing and OA closely resemble each other: both start with an inflammation and lead to a more

(12)

12 2.1. Osteoarthritis

or less stable bony connection. Moreover, if the patient continues to move an OA joint, it will stay rather flexible. This is similar to a fracture that is not fixated in place and might develop into a pseudarthrosis if it is moved constantly.

These are the reasons why it seems possible that OA is a type of bone healing in a misguided context. More precisely, OA might be initiated by the same biochemical reactions as fracture healing because both begin with an inflammation and continue by bone remodeling and formation.

To proof this theory, the biochemical markers sRANKL (soluble receptor activator of nuclear factor-kappaB ligand), chordin and osteocalcin were analyzed in synovial fluid and serum of dogs by enzyme linked immunosorbent assay (ELISA). Their function during fracture repair and bone remodeling is rather clear and they all have been found in healing bone tissue (cf. 2.4), but most of them have not been examined in canine osteoarthritic tissues yet and their roles in osteoarthritis development still have to be elucidated. The marker concentrations were compared to the individual radiographic grades of osteoarthritis. For this purpose, a dog-specific radiographic scoring system for OA in the stifle joint was developed. Other clinical findings like the degree of lameness and duration of orthopedic problems were taken into consideration as well.

The aim of this study was to find out if the markers sRANKL, chordin and osteocalcin can be detected locally in the synovial fluid, how they are concentrated systemically in the blood and how concentrations in both fluids vary among different stages of osteoarthritis. Also, clinical aspects of OA course in dogs should be explored and the diagnostics should be improved.

(13)

2. Literature

Arthrosis is defined as a destructive chronic disease of articular cartilage with a primary or secondary inflammation. In case of involvement of bone, it is called osteoarthrosis, but more common is the term osteoarthritis. Histologically, degeneration of cartilage is visible as well as proliferation of synovial membrane and infiltration of inflammatory cells (GRÖNE and BAUMGÄRTNER 2007).

The term osteoarthritis implies that this disease is mainly inflammatory. Still, it is the most common description used for this process in the English language. In German, the word osteoarthrosis is preferred which focuses on the degenerative principle of this disease. Therefore, the German language distinguishs precisely between the degenerative process arthrosis and the pure joint inflammation (arthritis). Many authors believe that the English term osteoarthritis is misleading and wrong as it

“incorrectly implies an inflammatory origin” (CRAIG et al. 2007). But as inflammation plays a pivotal role in OA initiation and progression (see below), the English term is not incorrect. However, in this dissertation the word osteoarthritis is used as it is most common.

The disease OA can be categorized as primary or secondary OA. Primary OA is diagnosed if pathogenetic factors are unknown. In contrast, secondary OA is caused either by congenital disorders or acquired prepositions (like trauma, high weight or aseptic necrosis (GHARBI et al. 2013; MELE 2007)).

For example, congenital disorders like limb malposition and osteochondrosis might lead to a fragmented processus coronoideus (FPC), ununited anconeal process (UAP), joint incongruity and cartilage erosion, all grouped under the term elbow dysplasia (ED). In young dogs (4–8 months) this often results in lameness and can cause elbow OA (KIRBERGER and FOURIE 1998). Hip malpositions like hip dysplasia (HD) can lead to subluxation of the hip. Sooner or later, this constant incongruent workload causes osteoarthritic processes in the hip. Both HD and ED

2.1. Osteoarthritis

(14)

are believed to have genetic triggers, but also predispositions like weight, growth, hormones and exercise take part in their genesis (CLEMENTS et al. 2006).

Another joint which is often affected by OA is the knee (or stifle). It is either affected in young dogs due to congenital abnormity or trauma or, in older patients, due to chronic degenerative processes. The luxation of the patella can occur exclusively or in combination with rupture of the cranial cruciate ligament (RCCL). In younger dogs the luxation mostly appears due to hereditary anatomic anomalies like hip malformation and displacement of the quadriceps mechanism. The hind limb develops a malposition and the patellar tendons become lax so that the patella can luxate. This often happens to be at the medial side of the stifle and in rather small dogs. Without the required pressure of the patella during growth the trochlear groove develops a shallow or even plain shape; this causes a lifelong problem.

As an example for acquired prepositions, trauma due to a sudden misload can cause patellar dislocation as well, which is independent of age (MCLAUGHLIN 2001). If tendons become loose and the instability grows, the patella luxates more often. This stimulus can lead to osteoarthritic processes.

RCCL itself is also a common traumatic reason for OA development. In young dogs it mostly occurs as injury because of a sudden hyperextension, whereas in older dogs it is often triggered by degenerative processes and finalized by trauma (MCLAUGHLIN 2001; ROUSH 2001). When the cranial cruciate ligament ruptures partly or completely, the position of the tibial plateau is no longer fixated to the femur in cranial and medial direction. During movement this instability causes cartilage lesions which promote OA, but also the damaged tissue like the ruptured ligament are believed to be initiators of OA as they maintain the inflammatory reactions (RYCHEL 2010; MCLAUGHLIN 2001). The rupture or deliberate transection of the cranial cruciate ligament (ACLT) in dogs and other animals is a frequent model for osteoarthritis research with a benefit for humans as well (LITTLE and SMITH 2008;

GHARBI et al. 2013; BRONNER and FARACH-CARSON 2007).

A frequent acquired risk factor for OA in humans and dogs is overweight. In dogs high weight might occur because puppies are fed excessively and grow too fast, but it is also the breed which predisposes for a certain weight and physique

(15)

(MELE 2007). Therefore, some breeds are affected more frequently and in younger ages. For example, the average age of Rottweilers is 3,5 years when OA is diagnosed, whereas it is 9,5 years for Poodles. German Shepherd Dogs, Golden Retrievers, Rottweilers and Labradors have a higher predisposition for OA than other breeds. It is estimated that 45 % of dogs with OA are categorized as “big breeds” and about 50 % among these are even “giant-sized dogs”, but only 27 % of all OA cases count as small breeds (MELE 2007).

The most important risk factor for OA in humans is the age (GLYN-JONES et al. 2015), and as it is believed that degenerative processes are the reason, it seems likely that it has high importance for dogs as well because aging of dogs is similar to aging of humans.

Lastly, fractures and other instabilities that lead to imbalanced workload are OA-facilitating factors as they cause inadequate movement of the joint (RYCHEL 2010; BRONNER et al. 2007).

As previously mentioned, OA is a disease that not only affects cartilage, but also the entire joint, meaning bone, synovial membrane, muscles, tendons and ligaments (EGLOFF et al. 2012). Whichever activator is responsible for the onset of OA, the disease always develops by synovial inflammation, cartilage degradation, subchondral bone remodeling and osteophyte formation in areas with the greatest movement. This process continues until the joint stiffens progressively and ankylosis begins (HUNTER 2011; NEUMANN 2015). Today mostly the tissues cartilage, synovial membrane and subchondral bone are considered to play key roles in the pathogenesis of OA. There are a lot of cytokines and growth factors which originate from various tissues and are involved in disease progression:

In cartilage dysregulation of repair is thought to be the initiating process of OA.

Presumably activated by fissures which occur during aging and mechanical stress, chondrocytes secrete cytokines like interleukin-1β (IL-1β), IL-6, tumor-necrosis-factor α (TNF-α) and proteinases (mainly matrix metalloproteinases (MMPs) like MMP-13 and aggrecanases) which are leading to matrix degradation.

Instead of collagen II and aggrecan, more type X collagen is produced. Additionally,

(16)

pathologic calcification takes place. These factors are responsible for degradation of cartilage resistance and buffering ability. Augmented toll-like receptors are expressed, and the complement system is activated by damage-associated triggers like extracellular matrix molecules, so the innate immune system partakes as well.

Eicosanoids (like prostaglandins) and reactive oxygen species are also thought to contribute to cartilage degradation. Consequently, chondrocytes become hypertrophic and decline. Cartilage volume initially increases while its surface becomes erosive. Beginning with microcracks, lesions are growing towards subchondral bone until vertical clefts have destroyed the protective function of the subchondral bone cartilage. Fragments of cartilage can even detach and move into articular cavity (KLIPPEL et al. 2008, GLYN-JONES et al. 2015, HAQ et al. 2003).

In subchondral bone mechanisms of endochondral ossification take place.

Osteoblasts and osteoclasts are activated and start enhanced bone turnover. This bone turnover leads to thickened subchondral bone mass of inferior quality, as the new bone is less mineralized (BRONNER and FARACH-CARSON 2007). Growth factors like vascular endothelial growth factor (VEGF) initiate vascular infiltration and neo-angiogenesis towards cartilage. Additionally, osteophytes are built at predisposed positions and subchondral cysts develop. Histologically, microfractures can be seen in areas with the greatest cartilage damage. Because subchondral bone is highly innervated an important generation of pain in OA might be localized here (GLYN-JONES et al. 2015, HAQ et al. 2003).

Synovial inflammation plays a pivotal role in OA. Synovial lining cells activate macrophages that secrete proinflammatory cytokines like IL-1β, IL-6, IL-17, TNF-α, TGF-β (transforming growth factor-β), IGF-1 (insulin-like growth factor-1) and LIF (leukemia inhibitory factor). Subsequently, acute-phase proteins like the C-reactive protein are produced (HASSANALI and OYOO 2011; HAQ et al. 2003).

Synovitis supports joint degeneration in a positive feedback circle and makes its contribution to OA symptoms. The synovial membrane thickens and the produced lubricants for synovial fluid display functional loss so that deficiency of lubrication takes its part in joint space narrowing and cartilage friction (BRONNER et al. 2007).

(17)

In final stage OA the most common sign is cartilage degeneration. If the innervated bones get into contact, movement becomes painful. Damaged structures can cause crepitus. Osteophytes grow and play their part in joint immobilization, supported by pain and loss of gliding function. Other tissues like the cranial cruciate ligament can be affected as well, although a rupture mostly precedes OA (ROUSH 2001).

If the inflammation, caused by any reason, activates the chondrocytes to start OA or if degradation of cartilage is the beginning of OA procedure remains unproven. Both pathways are plausible. But in recent studies the idea emerged that subchondral bone metabolism might start and/or promote OA (LAJEUNESSE and REBOUL 2003).

Fracture healing or bone healing describes all pathophysiological and biochemical processes which occur after a disconnection of a bone. A fracture is always accompanied by damage of the surrounding soft tissue. The aim of the healing is to heal the soft tissues and to connect the fracture ends in order to obtain the former physiologic stability (THOMAS and ADLER 1996; HIRNER and WEISE 2008).

After a bone has broken, the tissues surrounding it immediately start inflammation and repairing processes with the aim of reconnecting severed fragments with solid bone. Fracture healing can occur in two forms: primary bone healing is rare and in need of perfect conditions like absolute immobility of the concerned limb and direct contact of fracture ends. In this case bone tissue can be built directly without inflammation and cartilage formation. If these perfect conditions are not given, secondary bone healing takes place. It can be described in a 4-stage model in which inflammation, proliferation, remodeling and occurring biochemical factors are considered (LIEBERMAN and FRIEDLAENDER 2005; SCHINDELER et al. 2008).

Naturally, stages are partly overlapping:

The first stage is called inflammation. It starts immediately after injury and lasts about 1 week. Initiated by a trauma that has torn a bone into two parts, haematoma fills the damaged tissue. Platelets secrete chemoattractants and induce chemotaxis

2.2. Bone healing

(18)

18 2.2. Bone healing

of macrophages, granulocytes and lymphocytes. These release different cytokines (IL-1, IL-6 and TNF-α), which start the inflammatory process. Inflammation is believed to be important to keep the healing cascade going by activating other proteins and organizing the tissues. New vessels grow into the clot that in turn is transformed into granulation tissue. This process is supported by an increase of VEGFs and fibroblast growth factor-2 (FGF-2). Macrophages remove debris, whereas transforming growth-factor-β (TGF-β), platelet-derived growth factor (PDGF) and bone morphogenetic protein-2 (BMP-2) increase, initiating callus formation.

Mesenchymal stem cells are recruited as well (LIEBERMAN and FRIEDLAENDER 2005, AI-AQL et al. 2008).

Next step in bone healing is soft callus formation or cartilage formation, which starts a few days after injury and lasts up to 2 weeks. It starts with chondrogenesis:

chondrocytes and fibroblasts start to produce cartilaginous matrix and fibrous tissue to replace the granulation tissue with it. The resulting soft callus connects bone fragments and generates basal stability in the fracture line. In the end, the cartilage is mineralized and chondrocytes undergo apoptosis. Soft callus formation is accompanied by angiogenesis and further vascular invasion, although cartilaginous callus itself is avascular. These processes are caused by other cytokines and growth factors, for example insulin-like growth factor-1 (IGF-1), BMP-5 and -6, angiopoietins and VEGFs (SCHINDELER et al. 2008).

Cartilage formation is followed by primary bone formation or hard callus formation, which lasts from 2 weeks until months after injury. Now osteogenesis is started and woven bone is built by osteoblasts while mineralized cartilage is resorbed by chondroclasts (endochondral ossification). Cartilage resorption is associated with an increase of TNF-α, which also promotes recruitment of mesenchymal stem cells.

Other important factors for endochondral ossification are BMP-3, -4, -7 and -8, RANKL and MCSF. As BMPs increase, antagonists like chordin, DAN (differential screening-selected gene aberrant in neuroblastoma) and BAMBI (BMP and activin membrane bound inhibitor) rise, especially when formation of new bone material is completed and only remodeling processes remain to be done. In some regions close to the fracture site, even direct intramembranous ossification by differentiated

(19)

osteoblasts takes place. During intramembranous and endochondral ossification an increase of BMP-5 and -6 is seen and neo-angiogenesis still is upregulated by VEGFs and other angiogenic factors (SCHINDELER et al. 2008; AI-AQL et al. 2008;

GHIASI et al. 2017; DEAN et al. 2010).

The fourth and last phase of fracture healing is secondary bone formation or remodeling. It starts months after injury and lasts until years after it. At this point the woven bone is being replaced by cortical and trabecular bone, meaning the irregular structure of woven bone is reorganized to original lamellar bone. Osteoclasts play an important role in demineralizing the matrix and resorbing tissue while osteoblasts can produce new bone material. Important factors for osteoclastogenesis and their activation are RANKL, osteoprotegerin, ILs, TNF-α, BMPs and TGF-β (LIEBERMAN and FRIEDLAENDER 2005, SCHINDELER et al. 2008, GHIASI et al. 2017).

Some publications indicate that OA might resemble bone healing, but no author has vocalized this idea until 2015 when NEUMANN published the thesis that OA might be bone healing in a misguided context.

For example, a review which pointed out the idea that subchondral bone and articular cartilage are a functional unit and that they might be causing the initiation and progression of OA was written by LAJEUNESSE and REBOUL in 2003. It highlights different studies that proposed the idea of subchondral bone and cartilage as a unit and, more importantly, of subchondral bone as the initiator of OA. The authors state that changes in bone density and osteoid volume are often more severe than cartilage degeneration in spontaneous OA, whereas the severity of cartilage loss only exceeds bone changes in advanced OA (e.g. CARLSON et al. 1994). Thus, OA might begin in subchondral bone. Other sources mentioned in this review have shown that various cytokines and other proteins that are specific for bone metabolism increase in OA (e.g. SEIBEL et al. 1989, SHARIF et al. 1995, DEQUEKER et al.

1993, TARDIF et al. 1999). In addition, channels between subchondral bone and 2.3. Comparison of fracture healing and osteoarthritis

(20)

20 2.4. Biochemical markers

cartilage have been discovered which could explain the initiation of cartilage degradation (SOKOLOFF 1993, MITAL and MILLINGTON 1970).

MURATOVIC et al. (2018) investigated bone marrow lesions in the tibial plateau of osteoarthritic knees. They found a higher microcrack burden, more bone resorption, less cartilage volume and a higher numerical density of osteocytes compared to normal subchondral bone tissue. Furthermore, they found an increased arteriolar density and increased wall thickness of blood vessels. All these findings are reminiscent of the processes taking place during fracture healing.

In 2006 HUNTER et al. wrote about bone marrow lesions facilitated by mechanical stress. These lesions could be strongly associated with subsequent cartilage degradation, which is a part of OA. Another study connected OA with subchondral bone cysts that led to increased bone turnover (CHEN et al. 2015). In addition, the study of GEVERS and DEQUEKER (1987) resulted in the idea that osteoarthritis must be a more generalized bone disease instead of a localized cartilage disease (cf. 2.4.3.3). In summary, these studies support the idea of transformed subchondral bone as the initiator of OA.

There are several markers that are promising candidates for a comparison of biochemical reactions in fracture repair and OA. The proteins investigated in this study should ideally play varying roles in the process of bone healing so that a reliable comparison of these mechanisms to osteoarthritis can be made.

Furthermore, markers with available canine-specific immunoassays were chosen.

Thus, sRANKL, chordin and osteocalcin were analyzed in the synovial fluid and serum of dogs.

2.4. Biochemical markers

(21)

2.4.1. RANKL / sRANKL 2.4.1.1. Biology of RANKL

Receptor activator of nuclear factor-kappaB ligand (RANKL), also known as tumor necrosis factor ligand superfamily member 11 (TNFSF11), Osteoprotegerin ligand (OPGL) and TNF-related activation-induced cytokine (TRANCE) (FULLER et al. 1998; MUELLER and HESS 2012), is a cytokine that belongs to the tumor necrosis factor family. It exists either as a transmembrane or a soluble form (LACEY et al. 1998) called soluble RANKL (sRANKL).

The canine RANKL is composed of 315 amino acids (predicted on www.ensembl.org, Ensembl-ID: ENSCAFT00000007513.4). Interaction of RANKL with RANK, its receptor, is crucial for osteoclastogenesis, activation and differentiation of osteoclasts and for their delayed apoptosis (LACEY et al. 1998; KECK and PECHERSTORFER 2003; BOYCE and XING 2008). Another receptor for RANKL is Osteoprotegerin (OPG), which plays an antagonizing role to RANK and therefore inhibits osteoclast differentiation and activation after binding to RANKL on the osteoclasts (LACEY et al. 1998). These facts point out that the role of RANKL not only lies in bone resorption but also in bone turnover. Thus, it could play a role in fracture healing as well as in inflammatory bone and joint diseases like rheumatoid arthritis and OA.

RANKL was first found in lymph nodes and thymus, mostly expressed in T-cells (WONG et al. 1997). In bone and cartilage, RANKL is mostly secreted by osteoblasts, osteoclasts, osteocytes and hypertrophic chondrocytes (NAKASHIMA et al. 2011; XIONG et al. 2011; BOYCE and XING 2007; WANG et al. 2005). In joints, it is also expressed by synovial fibroblasts (TAKAYANAGI et al. 2000). RANKL induces osteoclast formation in these tissues and leads to resorption of calcified cartilage during endochondral bone formation (BOYCE and XING 2007; XIONG and O’BRIEN 2012). Hence, LACEY et al. could demonstrate hypercalcemia and reduced bone volume in mice after injection of recombinant RANKL in 1998. Mice lacking RANKL in various cells have developed increased bone mass due to lack of osteoclast function (XIONG et al. 2011). Interestingly, RANKL was also determined

(22)

as a part of the structural bone changes of osteoporosis in women: Denosumab, an anti-RANKL antibody which blocks the binding of RANKL to RANK, was injected into women with osteoporosis and these women had a significant lower risk of fractures than women who had received a placebo. This led to the conclusion that RANKL is involved in bone resorption and loss of bone density in osteoporosis (CUMMINGS et al. 2009).

However, osteoclast function is related to bone resorption, which in turn is part of bone remodeling. Therefore, RANKL concentration cannot only be correlated negatively to bone volume.

2.4.1.2. RANKL in fracture healing

In mice, KON et al. found an immediate increase of RANKL mRNA expression after tibia fracture, and the level remained higher than normal for the whole healing period except for a little drop until day 7 when OPG expression was elevated (2001). This supported the idea that RANKL is not only important for bone remodeling but also responsible for early mechanisms in fracture healing like endochondral resorption.

A little differing, KIDD et al. (2010) found elevated RANKL mRNA expression in rats at 4 days after stress fracture, its peak at 7 days and persistent higher levels at 14 days after injury which were exactly the moments when remodeling in the fracture line started and woven bone already had been produced (proven here by collateral histology).

In diabetic patients whose fracture healing is delayed and bone formation is diminished, significantly higher mRNA expression levels of RANKL were found compared to the normal group. Additionally, on day 16 after fracture the fracture callus and cartilage tissue were twofold smaller in the diabetic group (KAYAL et al.

2007). This example underlines the role of RANKL during the cartilage resorption of fracture healing.

(23)

2.4.1.3. RANKL in osteoarthritis

In 2008, PILICHOU et al. found elevated sRANKL levels and sRANKL/OPG-ratios in serum of human patients with primary knee OA. Interestingly, they also observed a negative correlation between severity of the disease and serum sRANKL levels. No difference was found between sRANKL levels in serum and synovial fluid.

In rheumatoid arthritis, RANKL has been the object of investigation in joint material for several times in the last few years. In a study of DANKS et al. from 2016 for example, RANKL gene was deleted in synovial fibroblasts of mice, which led to a significant decrease of osteoclast formation and bone erosions. Deletion in T-cells had different outcomes. Articular cartilage loss, however, has not been prevented.

As far as it is known, there has been no investigation of RANKL or sRANKL in osteoarthritic material of dogs so far, so it was a point of interest if and how RANKL concentrations vary amongst healthy and osteoarthritic joints of canine patients.

2.4.2. Chordin

2.4.2.1. Biology of chordin

Chordin is a large polypeptide protein consisting of 964 amino acids (in dogs, predicted on www.ensembl.org, Ensembl-ID: ENSCAFT00000020435.4) and has four cysteine-rich domains. In embryonic development, it is responsible for dorsalization (SASAI et al. 1994) and an important regulator of chondrocyte maturation and limb growth (ZHANG et al. 2002). In adults, it was initially found in liver, cerebellum, female genital tract, heart, skeletal muscle, medulla and spinal cord (MILLET et al. 2001), later also in cartilage and synovial membrane (TARDIF et al.

2004 and 2006). Its function is to antagonize bone morphogenetic proteins that are growth factors belonging to the transforming growth factor beta (TGF-β) superfamily (LAVERY et al. 2008; BRAGDON et al. 2011). More precisely, chordin binds to BMP-2, BMP-4 and BMP-7 (ZHANG et al. 2007; PICCOLO et al. 1996). Besides fetal

(24)

development of neural tissues, cartilage and heart, BMPs are important for formation of postnatal bone (CHEN et al. 2004).

BMP-2, BMP-4 and BMP-7, the ligands of chordin, are able to induce osteoblastic differentiation in primary human mesenchymal stem cells (LAVERY et al. 2008).

Large segmental bone defects in rabbit humeri could be repaired by the presence of BMP-2 in titanium implants within 6 weeks (MURAKAMI et al. 2002). There also have been several clinical trials on the influence of BMP-2 and BMP-7 treatment on open fractures and non-unions, but most of them could not find any significant differences when compared to patients not treated with BMPs, except that BMP-2 was found to accelerate fracture healing and wound healing and to decrease interventions and infections in open fractures (GAUTSCHI et al. 2007). YIN et al. (2015) could also show a positive correlation of BMP-2 levels in synovial fluid and serum with the severity of knee osteoarthritis in human patients.

Chordin is therefore believed to play an influencing role in bone formation and OA by antagonizing these BMPs.

2.4.2.2. Chordin in fracture healing

In 2012, KLOEN et al. could show chordin expression in fracture callus of different healing stages by immunohistochemistry. They found chordin in chondrocytes, osteoblasts, osteoclasts and fibroblasts. In non-unions, they also detected chordin in osteoblasts and osteoclasts. Unfortunately, they did not describe the individual status of the healing cascade. DEAN et al. (2010) analyzed BMPs and BMP-antagonists (chordin among others) in fracture tissues of mice. They found downregulated BMP-antagonists during the whole time of fracture repair as well as an increase of other BMP-antagonists. The latter included chordin, which increased from day 7 to 3 weeks after injury with its peak after 2 weeks. Therefore, they concluded that there are two groups of BMP-inhibitors: those whose suppression may be crucial for new bone formation and those who are important for bone remodeling after formation of new bone has taken place.

(25)

KWONG et al. (2009b) found chordin expression mostly in areas of cartilage formation and barely in areas of bone formation during human fracture healing. As they did not take differing stages of bone repair into account, the detected chordin concentrations could be normal as expression of chordin automatically occurs with rise of BMPs.

Some other studies investigated the ratio of BMP-concentrations and BMP-antagonist concentrations (among them chordin) in fracture healing and non-unions. Like they had hypothesized, they found an elevation of antagonists with their respective opponent BMPs in healing bone as much as in non-unions, but BMP-concentrations were decreased in non-unions compared to concentrations in healing fractures (KWONG et al. 2009 b; KLOEN et al. 2012).

Altogether, chordin is important in fracture healing as an antagonist of BMPs and it seems to increase especially during bone remodeling and hard callus formation.

2.4.2.3. Chordin in osteoarthritis

A study from 2006, utilizing real-time PCR and immunohistochemistry, reports increased chordin levels in osteoarthritic cartilage and synovial fibroblasts in comparison to healthy joints (TARDIF et al. 2006). Another finding was the downregulation of chordin by various growth factors (e.g. BMP-2) in osteoarthritic cartilage, but not in healthy cartilage. Interferon-γ caused an increase of chordin in healthy but not in OA cartilage. No influence on chordin expression was found in synovial fibroblasts. Unfortunately, the authors did not describe the OA-stages of the patients. Interestingly, in their former study from 2004, they had found no difference in gene expression of chordin between normal and osteoarthritic chondrocytes and synovial fibroblasts by use of real-time PCR-quantitation.

These results illustrate that osteoarthritis has a very complex pathogenesis with a plethora of regulating proteins being involved. Logically, joint tissues and body fluids like serum and synovial fluid have to be analyzed separately.

As far as it is known, chordin has not been detected in synovial fluid so far and there have not been any investigations in canine material at all, so that the level of chordin

(26)

in synovial fluid and serum of healthy dogs and dogs with OA poses an interesting area of research.

2.4.3. Osteocalcin

2.4.3.1. Biology of osteocalcin

Osteocalcin (OC), also known as bone gla protein (BGP), is a non-collagenous, vitamin K- and vitamin D3-dependent, calcium-binding protein consisting of 49 amino acids (in dogs) and is part of the extracellular bone matrix (COLOMBO et al. 1993;

BOIVIN et al. 1990, FRASER et al. 1988). The protein contains three gla (γ-carboxyglutamic acid) residues, which are able to bind calcium ions. This cation binding leads to an alpha-helical conformation and consequently the affinity of osteocalcin to hydroxyapatite surfaces increases (HAUSCHKA and CARR 1982).

Therefore, it is connected to bone calcification. Recent studies pointed out systemic roles beyond bone mineralization for osteocalcin: it is also believed to work hormonally and seems to be involved in glucose metabolism, fertility and cognition (ZOCH et al. 2016).

Osteocalcin is mostly secreted by osteoblasts. Nevertheless, low amounts were observed in osteocytes and it could be found in progressively increasing concentrations along the mineralization front in bone matrix (BOIVIN et al. 1990).

This implies a connection between osteocalcin presence and the calcification process. Indeed, the expression level of osteocalcin was found to affect maturation of mineral species during osteogenesis of mesenchymal stromal cells and to influence hydroxyapatite concentrations in matrix (TSAO et al. 2017). It has also been found in connection with fracture healing in numerous studies (cf. 2.4.3.2). But, as osteocalcin also takes part in bone resorption by recruiting osteoclasts and osteoclast precursors (GLOWACKI and LIAN 1987), a rather suppressive function of osteocalcin in bone formation could be detected as well. For example, in 1996 DUCY et al. observed an increase of bone formation with elevated mineralized bone mass in genetic modified

(27)

mice lacking osteocalcin, compared to wildtype mice. PRICE et al. (1980) found elevated osteocalcin plasma levels in several bone diseases, including diseases with increased bone resorption like secondary hyperparathyroidism but also with increased bone formation like Paget’s disease. As Paget’s disease is not just characterized by bone formation but rather by foregone bone resorption and remodeling (ALBAGHA and RALSTON 2016) and as secondary hyperparathyroidism is also characterized by bone remodeling, it seems likely that osteocalcin is a marker for bone remodeling rather than a marker for bone mineralization in bone formation.

This is also supported by a study of LUMACHI et al. (2009). They analyzed osteocalcin and bone-specific alkaline phosphatase concentrations in serum of menopausal women with osteoporosis. In women older than 59 years, they saw a significant increase of osteocalcin levels in relation to the age, but there was no connection to a loss of bone mineral density. This shows that osteoporosis is not necessarily related to the loss of bone mineral density but that it is rather associated with bone remodeling. It also shows that osteocalcin is associated with these processes.

Indeed, osteocalcin is used in medical laboratories as a blood marker for bone remodeling with associated elevated osteoblast activity like fracture healing, primary hyperparathyroidism, bone metastases and osteoporosis or osteomalacia (e.g.

BIOSCIENTIA).

2.4.3.2. Osteocalcin in fracture healing

In patients with bone fractures, osteocalcin increases within weeks after injury in serum and stays higher than normal during the whole time of fracture repair (TANIGUCHI et al. 2003; ÅKESSON et al. 1995). Comparing osteotomy healing to distraction bone repair, which has to build up a lot more callus, LAMMENS et al.

(1998) found progressively increasing osteocalcin concentrations in the serum of the distraction model group compared to the osteotomy group. In the latter, osteocalcin was initially increased but decreased later on.

(28)

In another study, fractures treated with PTH (parathyroid hormone) injections showed, despite elevated osteocalcin concentrations, a higher bone mineral content, higher bone mineral density and a higher ultimate load to failure of calluses compared to the control group (NAKAJIMA et al. 2002). There are more studies that investigated increased osteocalcin concentrations in combination with accelerated fracture repair (e.g. ROZEN et al. 2007).

PRICE et al. (1981) studied osteocalcin concentrations in serum and bone of rats treated with warfarin and vitamin K and discovered that the origin of serum osteocalcin is actual cellular synthesis alone and that no osteocalcin is transferred from bone resorption. This led to the conclusion that serum osteocalcin directly reflects osteoblast synthetic activity. Therefore, it should be a useful biochemical marker for the comparison of bone healing and osteoarthritis.

2.4.3.3. Osteocalcin in osteoarthritis

Already in 1987 GEVERS and DEQUEKER measured elevated osteocalcin levels in iliac crest bone and serum of women with generalized and local OA. They categorized OA as local if only the hip was affected and as generalized if the hand joints were affected. In these cases, the iliac crest bone, where samples came from, did not have to be strongly affected by OA. What they found were increased osteocalcin levels in cortical bone, cancellous bone and in the serum of OA patients, compared to healthy patients. Because of their findings in bone matrix and serum they presumed that osteoarthritis must be part of a more generalized bone disease instead of a purely cartilage disease.

In 1995, SHARIF et al. analyzed osteocalcin in synovial fluid of human osteoarthritic patients and saw a correlation between osteocalcin values and joint abnormalities in scintigraphic scans. In contrast, CSIFO et al. (2014) measured osteocalcin levels in serum and synovial fluid of variously staged osteoarthritic joints in humans and detected lower values in the advanced osteoarthritic group than in the early osteoarthritic group. They based their findings on the decreased bone metabolism in advanced osteoarthritis.

(29)

ARICAN et al. (1996) measured osteocalcin concentrations in serum and synovial fluid of dogs, comparing healthy joints to different joint diseases including osteoarthritis. They discovered increased osteocalcin concentrations in serum of osteoarthritic patients but no difference between healthy and osteoarthritic synovial fluid by radioimmunoassay. As far as it is known, no other investigations of osteocalcin in canine synovial fluid have been published.

(30)

30 3.1. Study design

3. Material and methods

For this study, dogs were examined that either had been euthanized or died in the Small Animal Clinic of the University of Göttingen or dogs that had surgery of the stifle joint at that clinic. The latter patients mostly had stifle surgery because of RCCL, some had a dislocation of the patella. Surgery patients usually presented with different stages of OA, depending on how much time had passed since the trauma or if even OA had contributed to the trauma by joint degeneration. Additionally, age differed among the presenting patients. Radiographs of the affected joints of all dogs were obtained in mediolateral and craniocaudal point of view to evaluate the stage of osteoarthritis. Clinical data like age, weight, breed, sex, date of trauma and degree of lameness were noted. Serum (from surgery patients only) was taken as well as synovial fluid from the concerned joints to analyze levels of sRANKL, chordin and osteocalcin. Concentrations and appearance of proteins were compared to clinical data of the patients and radiological findings of the joints.

The aim of this study was to gain knowledge about the occurrence of bone healing markers in an OA joint combined with the progress of their rise and fall during osteoarthritis development. The local concentrations of these proteins in synovial fluid were correlated to the systemic concentrations in serum. Besides, the progression of OA in dogs was analyzed and compared to clinical findings and the marker concentrations.

The results were expected to reveal how osteoarthritis is initiated and if there is a measurable marker that could be helping in early osteoarthritis screening and therapy. Secondly, they were hoped to show how clinical findings can partake in the diagnosis of OA.

3.1. Study design

(31)

In this study, 87 participants were involved: 12 of them were deceased dogs and 75 were surgery patients. 10 of them were counted twofold as samples were taken from both stifles. Because the patients’ clinical parameters like age, weight, degree of lameness, duration of lameness (surgery patients) and grade of OA (deceased dogs) distinguished, both examinations were included independently into data analysis.

According to the Veterinary Institute of LAVES, this study involved no animal experiments (confirmation from April 20st, 2016).

3.2.1. Deceased dogs

Dogs which had been euthanized or had died in the Small Animal Clinic of the University of Göttingen were to be used as control group, but some of them had osteoarthritic changes in their stifles. Furthermore, no blood could be taken from the deceased dogs as they mostly had been euthanized with Pentobarbital and Embutramid, so the serum already was hemolytic and cell destruction had taken place. Often, the coagulation process had already started. Therefore, deceased dogs and surgery patients both were considered in either the healthy or the osteoarthritis group. Tests were performed to see if there was any significant difference between marker concentrations in synovial fluid of deceased dogs and patients, so that these samples could be evaluated equally (cf. 4.1.4).

Members of this group did not die because of joint problems, nevertheless some of them showed osteoarthritic changes in their stifles, maybe due to the higher average age. The range of age in this group was 9–14 years, average age was 11,25 years.

Breeds occurring in this group were as follows: Golden Retriever (n = 1), Small Munsterlander (n = 1), German Shepherd (n = 1; 2 samples), Boxer (n = 1;

2 samples), Border Collie (n = 1; 2 samples), Rhodesian Ridgeback (n = 1;

2 samples) and Old German Shepherd Dog (n = 1; 2 samples). 4 of the dogs were male, 1 of them neutered, 3 were female, 1 of them neutered. The dogs weighed from 19–40 kg, the average weight was 28,7 kg. Most prevalent reasons for

3.2. Study population

(32)

32 3.2. Study population

euthanasia or death were neoplasia (n = 2, 4 samples), involving spleen and mammary gland tumor, and discopathy or other algetic problems in the spinal column (n = 2, 4 samples). One dog was renal insufficient, one suffered from epileptic seizures and one from chronic enteritis.

3.2.2. Patients in surgery

All dogs of which samples were taken had surgery of one stifle joint at the Small Animal Clinic of the University of Göttingen. Mostly (n = 71), these were patients with rupture of the cranial cruciate ligament (RCCL); 1 patient even had the cranial and the caudal cruciate ligament ruptured. 3 patients suffered from patellar dislocation.

Like for the deceased dogs, every patient’s breed, age, weight and sex was noted.

The patients were mostly mixed breed (n = 24), other breeds were Labrador Retriever (n = 7), Old English Bulldog (n = 4), Bichon à poil frisé (n = 4), Doberman (n = 2), Dogue de Bordeaux (n = 2), Biewer Yorkshire Terrier (n = 2) and other breeds that occurred just once (Boxer, Golden Retriever, German Shepherd Dog, Dalmatian, Renaissance Bulldog, Korthals Griffon, Australian Shepherd, West Highland White Terrier, Cairn Terrier, Beagle, Bernese Cattle Dog, Maltipoo, Appenzell Mountain Dog, English Bulldog, Rottweiler, other Terrier, French Bulldog, American Staffordshire Terrier, Poodle, Entlebucher Cattle Dog, American Pit Bull Terrier, Jack Russel Terrier, Greek Harehound, German Long Hair, Tibet Terrier, Staffordshire Bullterrier, Malinois, Tibet Spaniel, Cane Corso and Hovawart.). 30 of the dogs were male, 11 of them neutered, 45 were female, 16 of them neutered. The dogs weighed from 3–50 kg, the average weight was 23,9 kg. In this group, the range of age was 0,8–13 years and average age was 6,4 years.

(33)

No clinical examination could be performed anymore, so a preliminary report was noted including all characteristics mentioned in 3.2.1. Radiographs of the stifles were obtained from craniocaudal and mediolateral point of view to classify them as healthy or to determine the stage of osteoarthritis (cf. 3.5).

3.3.2. Joint surgery patients

Before surgery, a clinical examination was performed as usual to check suitability for anesthesia. This mainly included assessment of mucous membranes, auscultation of heart and lungs and palpation of the abdomen and lymph nodes. Furthermore, a specific orthopedic examination was done, depending on the kind of surgery. The degree of lameness of each patient was determined (degree 0–4 according to ARNOCZY and TARVIN (1981), Table 1). Swelling of the stifle was noted, too. If not already existing from earlier investigations, radiographs of the diseased joint were obtained from both mediolateral and craniocaudal point of view. The degree of osteoarthritis was determined as described below (3.4, Table 3). From patients’

preliminary reports, information was noted to classify subgroups. For this purpose, the duration of lameness was defined as the time span from the suspected trauma (mostly RCCL) until the day of surgery. The duration was divided into five subgroups (from days up to more than a year, Table 2).

3.3. Clinical examination and further diagnostics

(34)

34 3.4. Staging of osteoarthritis severity by radiographs

Table 1: Degree of lameness (ARNOCZY and TARVIN 1981).

Classification Manifestation Degree 0 No lameness

Degree 1 Hardly visible lameness (vaguely low grade)

Degree 2 Visible lameness, but loading of limb (clearly low grade) Degree 3 No full loading, just touching of the limb (moderate) Degree 4 No touching of the ground (high grade)

Table 2: Duration of lameness.

Classification Time

Duration 0 Days: suspected trauma a few days ago, less than a week

Duration 1 Week(s): Suspected trauma 1 to 2 weeks ago Duration 2 Weeks: Suspected trauma 2 to 8 weeks ago Duration 3 Months: Suspected trauma 2 to 12 months ago Duration 4 Year(s): Suspected trauma more than a year ago

At the beginning of this study, evaluation of stifle radiographs was to be categorized by means of the KELLGREN-LAWRENCE score for humans from 1957 (Table 3) as no established system for dogs was available. There has been an attempt to generate a radiographic scoring system for canine stifle joints by VASSEUR and BERRY (1992), but it was a 96-point scoring system with only 3 grades of OA and it could not prevail. KELLGREN and LAWRENCE performed several studies to elaborate an evaluating system for human joint radiographs that should be as

3.4. Staging of osteoarthritis severity by radiographs

(35)

independent as possible of the observer. Their grading system for knee radiographs considered osteophyte building, joint space narrowing, sclerosis and bony deformity.

During the preliminary works and sample collection of this study these published examples of different OA grades in knees were compared to various radiographs of healthy and arthritic canine stifles that were made at the small animal clinic. The outcome was that a complete transferability of this system to the dog is not possible.

This was confirmed by the intra-articular view during surgery: it was observed that osteophytes mostly occur between the femoral condyles and the medial and lateral ends of femur and tibia in dogs but rarely in the joint space like it is described in the KELLGREN-LAWRENCE system. The radiographic evaluation of beginning osteophyte growth between femoral condyles is not trivial because these are points that are overlain by bone from the lateral view. Another differing factor is that hardly any joint space narrowing could be seen in dog radiographs. This might be due to divergent statics in humans and dogs as humans carry much more load on the knees compared to a dog standing on four limbs. Furthermore, no stress images could be made from dogs as radiographs were performed in lying and stretched position (HD- position) whereas in humans it is useful to evaluate the joint with its natural load (standing). Lastly, big anomalies in bone deformation like bone cysts could hardly be found in dog radiographs although grade 3 and 4 (according to osteophytic building) were detected. These were the reasons why a modified scoring system for the evaluation of canine stifle joints on radiographs was elaborated. During this elaboration a lot of radiographs were assessed in which small beginning osteophytes could be detected (which was proven by the inspection during surgery). Roughness on predisposed places like condyles, patella and the facies poplitea had often been visible as well. These joints were just at the ultimate beginning of OA and were categorized as grade 1. Grade 2 was defined for definite but still small osteophytes.

Here, sometimes even sclerosis could be seen. For grade 3, multiple and moderate osteophytes had to be visible in combination with sclerosis and possible initiating bony deformity like indentations in femur condyles. If there were large osteophytes, definite sclerosis and even bony deformity OA was categorized as grade 4. As stated

(36)

before, high grade bony deformity could rarely be detected and grade 4 was only seen once.

In Table 3, the original KELLGREN-LAWRENCE system is opposed to the modified scoring system for dogs. In Figure 1 and Figure 2, exemplary pictures for each grade of arthritis are shown like they have been detected during this study.

(37)

Table 3: Radiographic osteoarthritis scores: KELLGREN-LAWRENCE system (1957) and a new modified system for dogs.

KELLGREN-LAWRENCE system Modified system for dogs Grade 0 Absence of radiographic

changes

Grade 0 Absence of radiographic changes

Grade 1 Doubtful joint space narrowing and possible osteophytic lipping

Grade 1 Slightly detectable, small osteophytes and/or

roughness, no sclerosis, no bony deformity

Grade 2 Definite osteophytes and possible joint space narrowing

Grade 2 Definite, small to moderate osteophytes, possible sclerosis, no bony deformity Grade 3 Multiple osteophytes,

definite joint space narrowing, sclerosis, possible bony deformity

Grade 3 Multiple and moderate to big osteophytes, initiating

sclerosis, possible initiating bony deformity

Grade 4 Large osteophytes, marked joint space narrowing, severe

sclerosis and definite bony deformity

Grade 4 Multiple and large osteophytes, sclerosis, advanced bony deformity

(38)

Grade 1 Grade 2

Grade 3 Grade 4

Grade 0

Figure 1 : Stages of osteoarthritis in radiographs of dog stifles, frontal view Grade 0: healthy stifle. No osteophytes can be seen, smooth bone ends.

Grade 1: beginning osteophytes and roughness can be seen on femoral condyles and tibia (arrows).

Grade 2: multiple, small to moderate osteophytes can be detected (arrows).

Grade 3: multiple moderate osteophytes and roughness visible on bone ends; sclerosis begins.

Grade 4: multiple and large osteophytes have been formed, bony deformity has started.

(39)

The evaluation of stifle osteoarthritis by means of the mediolateral view was limited as most arthritic activities are overlapping with normal bone material from this point of view. However, lateral views were consulted to complete arthritis evaluation and have been useful if the front view alone was not sufficient.

Grade 0 Grade 1

Grade 2 Grade 3

Figure 2: Stages of osteoarthritis in radiographs of dog stifles, lateral view Grade 0: smooth bone ends, no roughness or osteophytes can be seen.

Grade 1: beginning roughness, mostly detectable on the patella and the facies poplitea (arrows).

Grade 2: beginning osteophytes on femoral trochlea, condyles and on sesamoid bones (arrows).

Grade 3: roughness, osteophytes and sclerosis (stars) can be seen.

No lateral radiograph of a grade 4 stifle could be obtained.

(40)

40 3.5. Evaluation of osteoarthritis severity by intraarticular view during surgery

During stifle surgery, irrespective of whether it was RCCL or patellar dislocation, the joint is usually assessed by the surgeon, based on the following observations:

osteophyte growth and size, color of synovial membrane (redness yes/no), color and viscosity of synovial fluid (bright/sanguineous, viscous/liquid), thickness of synovial membrane (normal/slightly thickened/strongly thickened), shape of articular cartilage (erosions yes/no) and shape of the menisci (normal/deformed/torn). A liquefied and sanguineous synovial fluid and a thickened synovial membrane are signs of acute inflammation. Changes of cartilage and osteophyte building are more typical for chronic osteoarthritic processes.

In order to compare radiographic OA scores with the surgical assessment, an OA scoring system for the findings in surgery was elaborated. Naturally, during these surgeries only a limited view of the joint is given as it is opened at the dorsolateral side. Therefore, the patella, the femur condyles and the tibial plateau can be examined from the dorsal side and there is also a limited view of the medial and lateral side and the inner surfaces. No caudal view is given.

As synovial inflammation is an important part of osteoarthritis but does not give much information about the grade of OA, it was only adducted to the scoring system if the outcome ranged between two grades. Hence, for osteoarthritis evaluation during surgery the osteophyte growth was most import with cartilage and menisci degradation being of subordinated importance. An overview of the evaluation system is shown in Table 4.

3.5. Evaluation of osteoarthritis severity by intraarticular view during surgery

(41)

Table 4: Osteoarthritis scoring via intraarticular view during surgery.

Classification Manifestation

Grade 0 No osteophytes, no cartilage degradation

Grade 1 Beginning, small osteophytes, possible cartilage degradation and meniscus impair (low grade)

Grade 2 Small to moderate osteophytes, possible cartilage degradation and meniscus impair (moderate).

Grade 3 Moderate to big Osteophytes, cartilage degradation and/or meniscus impair (high grade)

Grade 4 Large Osteophytes, cartilage degradation and/or meniscus impair (severe)

Samples were taken from 75 dogs during surgery and from 12 deceased dogs. From patients and deceased synovial fluid was taken whereas blood could only be sampled from patients.

From dogs which were euthanized or died in the clinic, no blood was taken because either they had been dead for some time or because they had been euthanized with Pentobarbital and Embutramid so that the serum already was hemolytic and cell destruction had taken place.

Radiographs of both stifles were made to evaluate the grade of OA. Subsequently, synovial fluid was taken from the evaluated joints and following instructions of ELISA manufacturer BlueGene it was centrifuged at 1000 x g for 15 minutes to remove debris and blood. If the blood had not been slumped entirely the centrifugation was

3.6. Sample collection and preparation

(42)

42 3.7. Quantification of sRANKL, osteocalcin and chordin by enzyme linked immunosorbent assay (ELISA)

repeated at 1500 x g for 6–10 minutes. The supernatant synovial fluid was aspirated, aliquoted as 300 µl and frozen at -80 °C.

3.6.2. Patients in surgery

All dogs of which samples were taken presented for stifle surgery. As usual blood was taken for a preanesthetic blood tests so that serum could be obtained for these investigations, too. Serum was prepared for the ELISA following instructions of BlueGene: the blood was kept at room temperature for 2 hours to clot. Then it was centrifuged at 1000 x g for 15 minutes. Serum was aspirated, aliquoted and frozen at -80 °C. Radiographs of the injured stifle were made to evaluate the grade of OA.

In surgery, the synovial fluid was aspirated by opening the joint. The bones, cartilage and synovium were examined to evaluate the degree of osteoarthritis. The synovial fluid was centrifuged at 1000 x g for 15 minutes to remove debris and blood. If the blood had not been slumped entirely the centrifugation was repeated at 1500 x g for 6–10 minutes. The clarified synovial fluid was aspirated, aliquoted and frozen at -80 °C.

3.7.1. Assay procedure

Commercially available quantitative competitive ELISA kits were used. All three kinds of ELISA to determine canine osteocalcin, canine sRANKL and canine chordin had been manufactured by BlueGene Biotech (Shanghai, China) and were done in the same procedure: after bringing all reagents to room temperature, 100 µl of standard or sample were added to each well (coated with polyclonal anti-sRANKL/

-osteocalcin/-chordin antibody), for blank control 100 µl PBS were used. Then 10 µl 3.7. Quantification of sRANKL, osteocalcin and chordin

by enzyme linked immunosorbent assay (ELISA)

(43)

of balance solution were dispensed into synovial fluid (not into serum) to buffer the reagents. Into samples and standards 50 µl of enzyme conjugate were added and they were mixed well. Enzyme conjugate solution contains a specific antigen-horseradish peroxidase (HRP) conjugate (meaning osteocalcin-, chordin- or sRANKL-specific) which can bind to the antibody binding site. This way it competes with sample antigen. The plate was then covered, incubated for 1 hour at 37 °C and was washed 5 times manually with a provided wash solution after incubation. 50 µl of substrate A and substrate B were added to each well and the plate was incubated for another 15 minutes. The substrates contain 3,3’,5,5’-tetramethylbenzidine (TMB) and citric acid. The bound HRP catalyzes the oxidation of TMB by hydrogen peroxide into a blue colored product. The more antigen samples contain, the less enzyme conjugate can bind to the antibody binding site and therefore less chromogenic turnover of the substrate can take place. Finally, 50 µl of stop solution containing sulfuric acid were added to each well, affectively stopping the enzymatic reaction and changing the color of TMB to yellow. The optical absorption of the product at 450 nm (with a wavelength correction of 570 nm) was then measured spectrophotometrically in a microplate reader (TECAN Group Ltd., Männedorf, Switzerland).

3.7.2. Calculation of sample concentrations

The standard series solutions contained predetermined concentrations of canine osteocalcin (0–25 ng/ml), canine chordin (0–10 ng/ml) and canine sRANKL (0–1000 pg/ml). Standards and samples were measured in duplicate (except when sample volume did not suffice for more than one well). Mean values of standards and samples were calculated and the average blank control value was subtracted from all values. Standard curves were generated by plotting the given concentrations of each standard against the measured absorption. For a possibly fitting curve a four-parameter logistic regression curve was generated using a commercial software (Curve Expert Professional, version 2.6 by Daniel G. Hyams). Because competitive ELISAs have been used, absorption and concentration presented inversely