Dynamic Stiffness of Articular Cartilage and Potential Repair Materials

Home , Sesamoid bone

Dynamic stiffness of articular cartilage and potential repair materials

Inauguraldissertation

zur Erlangung der Würde eines Doktors der Philosophie vorgelegt der Medizinischen Fakultät der Universität Basel

von Sarah Ronken aus Nederweert, Niederlande

Basel, 2012

Genehmigt von der Medizinischen Fakultät auf Antrag von

PD Dr.med. Markus P. Arnold, PhD, Bruderholz Dissertationsleiter

Prof. Dr.med. Magdalena Müller-Gerbl, Basel Fakultätsverantwortliche

Prof. Dr. Ivan Martin, Basel Koreferent

Prof. Dr. Urs Staufer, Delft, Niederlande Externer Gutachter

Prof. Dr.med. Dr.med.dent. Dr.h.c. Hans-Florian Zeilhofer, Basel Prüfungsvorsizender

Basel, den 25. Mai 2012

Prof. Dr.med. Christoph Beglinger Dekan

Contents

Abstract 9

Publications arising from this thesis 13

1 Introduction 17

2 Experimental verification of a non-linear model for computing cartilage modulus from micro-indentation data 43 Journal of Computational and Mathematical Methods in Medicine

3 A comparison of healthy human and swine articular cartilage dynamic indentation mechanics 51 Journal of Biomechanics and Modeling in Mechanobiology

4 Acrylamide polymer double-network hydrogels: Candidate cartilage repair materials with cartilage-like dynamic stiffness and attractive surgery-related attachment mechanics 65 Cartilage

5 Double network acrylamide hydrogel compositions adapted to achieve cartilage-like dynamic stiffness 81 Journal of Biomechanics and Modeling of Mechanobiology

6 Discussion and outlook 93

Acknowledgements 103

List of publications and curriculum vitae 105

Abstract

Cartilage has limited potential for self repair. were performed because indentation minimizes Therefore, articular cartilage lesions often lead specimen preparation and has been shown by to early osteoarthritis. Early clinical results of others to produce meaningful cartilage stiffness cartilage replacement procedures such as autolo- data. However, a mathematical model is needed gous matrix-induced chondrogenesis (AMIC) to calculate stiffness data out of those experi- seem promising, but long term results are not ments. These models are always a simplification available to date. Only longitudinal studies of of the real situation, since cartilage is a complex 20 years and more will show whether the car- structure with complex mechanical properties. tilage repair procedures currently in evaluation To determine the influence of the mathemati- will prevent the treated patients from develop- cal model used on the results the conventional ing early osteoarthritis and if the progression of model (Hayes) is compared with a novel method osteoarthritis will be halted. In order to be able (Kren) in chapter 2, which has as a main ad- to evaluate and compare different methods of vantage that it does not assume linear elasticity. cartilage treatment, a thorough understanding Although a difference was found in absolute val- of the mechanical properties of intact cartilage ues calculated with these models, the trends they and cartilage with early degenerations is needed. show were similar when used to evaluate the The prime function of cartilage is load same set of data. Thus experimental data can- bearing. Cartilage absorbs and spreads the ap- not be compared between these different mod- plied energy thereby protecting the underlying els, but for comparisons within one model, both bone. It also provides diarthrodial joints with an models give similar results. almost frictionless gliding surface. It has been shown in clinic that there is a correlation be- In order to determine cartilage behaviour, pref- tween the histological quality, the load bearing erably healthy human specimens are tested. capacity and the durability of the repair. Thus it Unfortunately, these specimens were extremely seems logical to search for a repair with proper- difficult to obtain. Therefore chapterin 3, we in- ties close to that of normal cartilage. vestigated whether swine cartilage could serve as a model for human cartilage for mechanical test- The mechanical behaviour of cartilage is coming. At equivalent anatomic locations, dynamic plex, since the tissue structure is a combination modulus was similar for human and swine speci- of partly porous, viscous and elastic components. mens, but a small difference was found in the This results in deformation rate-dependent stiff- loss angle. Keeping these differences in mind, ness, i.e. how much energy is needed to deform swine specimens can be used for ex-vivo testing. the cartilage (dynamic modulus) and energy dis- Since mechanical behaviour of cartilage sipation (loss angle) properties. The water move- depends on the applied deformation rate and in- ment through or out of the cartilage under a ter- and intra-individual heterogeneity, in chap- given loading condition makes its response to ter 3 the behaviour of cartilage in swine knee loading more complex compared to an ordinary joints was determined as a function of loading viscoelastic solid. To determine these proper- mode and anatomic location. We observed a ties, several tests can be performed, i.e. uncon- larger heterogenity at fast compared to slow de- fined or confined compression, or indentation formation rates. Moreover, no differences were tests. In this thesis, dynamic indentation tests found in the loss angle at slow deformation rate

9 between locations. These differences highlight ter 4 by altering the water content in chapter 5. the need for using multiple test modes, i.e. load- The dynamic modulus increased with decreasing ing cartilage at different strain rates. water content. No difference in the loss angle was found in slow deformation whereas in fast After expanding the knowledge of dynamic stiff- deformation loss angle was higher in DN-gels ness properties of cartilage, in chapter 4 and 5 with lower water content. The DN-gel with low- we explored whether double network hydrogels est water content had higher stiffness in slow de- (DN-gels) are suitable as a cartilage repair ma- formation and lower stiffness in fast deformation terial. It already has been shown by others that compared to native cartilage. This difference is these DN-gels look promising to serve as a car- caused by the lower loss angle of this DN-gel. tilage repair material because of its low sliding Overall it looks promising that DN-gel stiffness friction, high wear resistancy, high thoughness can come close to that of native cartilage. How- and biocompatibility. Current focal repairs have ever, loss angle differences should be further in- a much lower initial stiffness and strength than vestigated. the surrounding tissue, which increases early failure potential. In chapter 4, we tested the me- In summary, cartilage is a complex structure and chanical properties related to surgical use of two it was shown that not only stiffness, but also en- kinds of DN-gels. Both DN-gels showed good ergy dissipation is an important mechanical pa- suture tear-out strength and also pull-off tests rameter. Both parameters should be investigated with tissue adhesive showed promising results. at multiple deformation rates to get a complete However, dynamic stiffness of both DN-gels was picture of cartilage mechanics. Also, healthy only about 10% of cartilage stiffness and also its swine cartilage was shown to be a reasonable loss angle was much lower. substitute for human cartilage in dynamic stiffness evaluations. Finally, DN-gels look promis- To increase the potential of these DN-gels as ing to serve as a cartilage repair material, since cartilage repair material, its stiffness has to be they have good surgical handling properties and increased. To achieve this, we adapted the stiff- their stiffness is close to that of native cartilage. ness of one of the two DN-gels tested in chap-

Publications Arising From This Thesis

Journal Papers Conference Abstracts

D. Wirz, S. Ronken, A.P. Kren, A.U. Daniels, Oral Presentations: Experimental verification of a non-linear model S. Ronken, D. Wirz, M. Stolz, A.P. Kren, A.U. for computing cartilage modulus from micro- Daniels, Experimental verification of a viscoelas- indentation data, submitted to Computational tic model for computing cartilage modulus from and Mathematical Methods in Medicine microindentation data, 17th Congress of the European Society of Biomechanics, Edinburgh, S. Ronken, M.P. Arnold, H. Ardura García, A. United Kingdom, July 5 - 8 2010 Jeger, A.U. Daniels, D. Wirz, A comparison of healthy human and swine joint cartilage dy- S. Ronken, M.P. Arnold, A.U. Daniels, D. Wirz, namic compression behaviour, Biomechanics Swine joint cartilage as an ex-vivo standard of and Modeling in Mechanobiology, 2011, DOI comparison for human articular cartilage, 7. 10.1007/s10237-011-0338-7 Jahrestagung der Deutschen Gesellschaft für Biomechanik, Murnau, Deutschland, May 19 M.P. Arnold, A.U. Daniels, S. Ronken, H. Ar- - 21, 2011, nomination for the young investigator dura García, N.F. Friederich, T. Kurokawa, J.P. award Gong, D. Wirz, Acrylamide polymer double- network hydrogels: candidate cartilage repair S. Ronken, M.P. Arnold, A.U. Daniels, D. Wirz, materials with cartilage-like dynamic stiffness A comparison of healthy human and swine joint and attractive surgery-related attachment me- cartilage dynamic compression behaviour in chanics, accepted for publication in Cartilage modes emulating joint function, International 2011 Society of Biomechanics, Brussels, Belgium, July 3 - 7, 2011 S. Ronken, D. Wirz, A.U. Daniels, T. Kuro- kawa, J.P. Gong, M.P. Arnold, Double network Poster presentations: acrylamide hydrogel compositions adapted to S. Ronken, M.P. Arnold, A.U. Daniels, D. Wirz, achieve cartilage-like dynamic stiffness, accept- Can swine joint cartilage serve as an ex-vivo sub- ed for publication in Biomechanics and Model- stitute for human articular cartilage in mechani- ing in Mechanobiology, 2012 cal tests?, 5th Basel International Knee Congress and Instructional Course, Basel, Switzerland, May 9 - 10, 2011, best poster award

1 Introduction

Young patients with “old knees” are one of constructs and other possible repair materials the most demanding patient groups in the with native cartilage one has to determine its outpatient clinic of an orthopeadic surgeon. behaviour. Preferably, these constructs and Cartilage lesions due to traumatic injuries are other repair materials are additionally com- common, especially in those practicing sports. pared with conventional methods for carti- Total or partial knee replacement results in lage repair. To be able to measure mechanical satisfactory clinical results for most older pa- properties of those clinically applied meth- tients. However, younger patients have higher ods, a biopsy is taken from the defect site. expectations and an artificial joint cannot get The disadvantage of this method is that the to their standards [99]. Besides artificial joint newly formed cartilage is disrupted and can surgery, other treatment protocols are ap- only be measured once. An arthroscopic de- plied in clinic [69, 109, 113, 141]. The results vice which is able to measure the mechanical of these treatments are very variable and the properties non-destructively in vivo would long-term outcome is unsatisfactory. The dif- be able to overcome this. Such a device might ferent types of cartilage lesions and patient’s also be used to detect osteoarthritis. However, potential to heal are important factors for the before developing such a device it needs to be success of cartilage repair [7]. However, sev- known what parameters are crucial and have eral lesions lead to osteoarthritis in a later to be measured. stadium [38]. This heavily increases social and economic burden on the health care sys- Quantifying mechanical properties of com- tems around the world. Therefore, prevention, plex structures like cartilage is not straight treatment and long-term cartilage repairs are forward. Mechanical properties of linear elas- widely investigated by researchers and clini- tic materials can be easily determined, but for cians [111]. linear viscoelastic materials it is already more challenging due to their time dependent and In the past decades a tremendous effort has rate dependent behaviour. In this thesis we been made in optimizing tissue engineered tried to come a step closer to treatment or di- articular cartilage constructs. However, in agnosis of osteoarthritis by investigating the order to compare those tissue-engineered mechanical behaviour of articular cartilage.

CARTILAGE Chondrocytes The ECM is built and maintained by the chon- Articular cartilage is a connective tissue at the drocytes. Chondrocytes are metabolically active ends of the subchondral bone in diarthrodial and respond to various environmental stimuli. joints. It does not have a blood supply, neither a They generally maintain a stable matrix; howev- lymphatic drainage, nor a connection to nerves. er, some stimuli may lead to degradation of the The specific microstructure and composition ECM. of cartilage is thought to give the tissue its remarkable mechanical properties and durability: Water it provides an almost frictionless joint motion Healthy cartilage has water contents ranging and it absorbs and distributes the applied load from 65% to 80% of its total wet weight. About to reduce localized stress concentrations in the 30% of the water is found in the intrafibrillar underlying bone. In most individuals, cartilage space within the collagen, but the majority is is able to do this for 8 decades or even longer [34, found in the molecular pore space of the ECM. 82, 110]. By applying a pressure gradient across the tissue, water may move through the ECM. Due to the Composition and structure high frictional resistance against this flow, the permeability of cartilage is very low. This resis- Articular cartilage is primarily a tissue of extra- tance together with the pressurization of the wa- cellular matrix (ECM) with a small number of ter within the ECM ensures its ability to support chondrocytes - specialized cells which are only very high joint loads. Nutrients are transported found in cartilage. The ECM mainly consists of within the tissue due to the water flow. water (65-80%) and the remaining wet weight of the tissue is accounted for principally by two Collagen macromolecular materials: collagen and proteo- Tissue’s tensile and shear properties as well as the glycans (Table 1.1). Beside these main compo- immobilization of the proteoglycans within the nents, cartilage also consists of lipids, phospho- ECM is determined by collagen. Collagen has lipids, proteins and glycoproteins. a triple-helical structure and the fibres vary in width from 10 to 100 nm, although it may in- Table 1.1: Composition of articular cartilage. Reprinted crease with age and disease. The collagen in the from [82] with author’s permission. cartilage tissue is cross-linked, which is thought Quantitatively Major Quantitatively Minor to add stability to the fibril network. The colla- Components Components gen fibre network does not offer significant resistance to compression, but it is stiff and strong in % wet (less than 5%) * tension and provides resistance to swelling and weight tensile strains [144].

Water 65-80% Proteoglycans Proteoglycans Collagen 10-20% Biglycan The size, structural rigidity and molecular confir- (type II) Decorin mation of the proteoglycans affect the mechani- Aggrecan 4-7% Fibromodulin cal behaviour of articular cartilage. Proteoglycans Collagens consist of a protein core with covalently bound Type V polysaccharide (glycosaminoglycan) chains (Fig- Type VI ure 1.1a). Aggrecan is the most common proteo- Type IX glycan in cartilage (80-90%). Aggrecan consist Type X of up to 100 chondroitin sulphate and 50 kera- Type XI tan sulphate glycosaminoglycan chains covalently bound to a long protein core. The N-terminal * Although these components are present in lower overall of this protein core is able to bind to hyaluronate amounts, they may be present in similar molar amounts compared to type II collagen and aggrecan (for example, and a macromolecular complex is formed when link protein), and may have major roles to play in the many aggrecan molecules are bound to a chain functionality of the matrix of hyaluronate (Figure 1.1b). This macromolecu-

19 Figure 1.1: (a) Schematic diagram of the aggrecan molecule and its binding to hyaluronate. This binding is stabilized by a link protein. Keratan sulphate and chondroitin sulphate glycosaminoglycan chains are bound to the protein core. (b) Diagram of a proteoglycan aggregate; aggrecan molecules bound to a chain of hyaluronate. Reprinted from [82] with author’s permission.

lar complex is effectively immobilized within the The superficial zone is the (gliding) sur- collagen network. face of the cartilage. It has the highest water con- All the glycosaminoglycan chains are tent (~80%) and the lowest proteoglycan content. ionized in solution. Positive counter ions are re- The collagen fibrils are relatively thin and are quired in the physiologic environment to achieve aligned parallel to the surface. The interconnec- overall electro neutrality. These free-floating tion between the proteoglycans and the collagen ions within the interstitial water cause osmotic fibrils is very strong. The chondrocytes are - ex pressure. Because the proteoglycans are packed tended with their long axis parallel to the sur- within one fifth of their free-solution volume in face. In the middle zone the collagen fibres have cartilage, fixed-charge groups are only 10 to 15 Å a larger diameter and are randomly distributed. apart. This results in a strong repulsive force. Further, the chondrocytes have a more rounded shape. In the deep zone the collagen fibres have a Location dependency large diameter and are organized perpendicular to the surface. The chondrocytes are arranged in Depth dependency columns and have a spherical shape. In this zone Cartilage is not a uniform tissue: its structure the lowest water content (~65%) and the high- and composition vary throughout its depth (Fig- est concentration of proteoglycans is found. The ure 1.2). It can be divided into four zones: the calcified cartilage separates the hyaline cartilage superficial zone, the middle zone, the deep zone from the subchondral bone and consists of small and the calcified zone. cells distributed in a cartilaginous matrix.

Figure 1.2: Chondrocyte (left) and collagen fiber (right) organization in the different zones: STZ = superficial tangential zone, the middle zone, the deep zone and the calcified zone. Reprinted from [25] with author’s permission.

20 Chondrocyte proximation Traumatic injuries The ECM structure is not only depth-dependent, it also changes depending on the proximity to The amount of stress transmitted to a joint by the chondrocytes. Collagen fibres get thinner indirect impact or torsional loading depends closer to the chondrocytes and in close vicinity on whether the load is expected or unexpected. hardly any collagen fibres are present. The ECM If the stress is expected, especially during slow proteoglycan concentration is increased by a fac- movements and impacts, the muscles absorb a lot tor of two adjacent to a chondrocyte. Since the of energy through contraction while simultane- majority of the ECM is not in close proximity of ously stabilizing the joint. If unexpected move- the chondrocytes, the material properties of the ments and sudden impacts occur, the muscles articular cartilage are mainly determined by that are not prepared to stabilize the joint and cannot part [82, 144]. absorb the energy. Consequently, sudden and unexpected movement or impacts transmit more For cartilage to function normally and provide stress to joint surfaces and are thus more likely protection to the bone and joint, each of the to cause articular surface injuries [27]. However, components described above must be present in cartilage can be damaged without disrupting the the proper amounts and in the right structure. articular surface or surrounding soft tissue [26]. The chondrocytes must be present for supervis- Alterations of the cartilage matrix can occur due ing the concentration and condition of the ECM to impact loading which is higher than the level and maintaining the equilibrium between syn- during normal activities, but lower than the level thesis and degradation. The collagens provide a necessary to produce cartilage disruption [31]. framework to resist tensile forces and the pro- The risk of an articular cartilage lesion teoglycans must sustain the hydration and re- in the knee joint after an anterior cruciate liga- sist compressible forces. The unique mechanical ment injury is 43% [124]; and in 60% of pa- properties of articular cartilage depend on these tients undergoing an arthroscopy, a chondral components and are sensitive to disruption [21]. lesion is found [141]. The most affected areas are the patella and the (medial) femoral condyle. In the majority of the cases the onset of the le- CARTILAGE INJURIES sion was traumatic and often occurring during sports participation [8, 53, 124, 141]. However, Loading and movement of joints are important damage may remain unnoticed, especially in to maintain the composition, structure and me- a younger population as long as the individual chanical properties of human articular cartilage. does not experience sequelae [38]. The intensity and frequency of this loading vary over a broad range. The balance between syn- Degenerative osteoarthritis thesis and degradation will be disturbed if the intensity and frequency is above or below certain Healthy articular cartilage can self-repair and tresholds. This causes changes in composition maintain the ECM, but with age this capacity and structure of articular cartilage. Reduced declines. In osteoarthritic (OA) cartilage the joint loading, e.g. due to immobilization, re- equilibrium of degradation and synthesis of the sults in atrophy or degeneration of the cartilage. ECM is disturbed. Part of the ECM starts to Whereas increased joint loading increases the degrade, which causes an increased synthesis of magnitude of the loading or impacts and may other matrix components. Many of the mecha- damage the cartilage [82]. Thus basically there nisms responsible for the origin and progression are two types of articular cartilage damage: of OA are still unknown. In early OA the water 1) Traumatic injuries which are the result of ex- content of the superficial zone increases. How- cessive loading, e.g due to a sport-accident or a ever, whether this water increase is due to a de- bad fall. 2) Biological disorders, which causes crease of proteoglycan content or due to damage deterioration of the articular cartilage. This can in the collagen network is still subject of debate. be initiated by e.g. avascular necrosis and osteo- Due to this water content increase, the cartilage chondritis dissecans. Besides this, cartilage dam- is less able to withstand compressional loading. age occurs without knowing the cause. When OA progresses, disruption and fibrillation

21 of the cartilage surface commences. As collagen Repair techniques: clinical applications fibers degrade, the proteoglycan molecules are less trapped in the structure and a decrease in Chondroplasty cartilage tissue is observed. Eventually, all carti- The cartilage in and around a chondral lesion is lage will have disappeared and the subchondral abnormal. Chondroplasty, also called debride- bone is exposed. As a side effect, inflammation ment, is a procedure where all unstable cartilage can occur due to breakdown products of the and the calcified layer are removed from the le- ECM, which are released in the synovium [24, sion. 68, 82, 137]. This procedure improves symptoms for five years or more, however, results gradually deteriorate over the five-year period. Whether CARTILAGE LESION REPAIR chondroplasty also improves symptoms in OA is still subject of debate [43, 57, 61, 89, 122, 125]. Due to the limited healing capacity of cartilage [58, 81], several techniques were invented to treat Microfracture articular cartilage lesions - e.g. microfracture, During an arthroscopy the microfracture tech- autologous chondrocyte transplantation, mo- nique is performed. This technique has gained saicplasty and recently, tissue engineered con- popularity because of its low costs, its limited structs. Depending on patient specific variables surgical morbidity and the technical simplicity - such as age, demand and other injuries - and [116, 142]. The lesion is prepared for microfrac- defect specific variables - such as size, depth and ture by removing the loosely attached cartilage location of the defect - the most suitable tech- from the surrounding rim and debriding the ex- nique is chosen [17, 130]. In general, treatment posed bone of all remaining cartilage tags. Then of fresh defects has a higher chance of healing multiple holes are made with an arthroscopic awl. compared to old defects. The goal of any inter- Those holes are approximately 3 to 4 mm apart vention would be the formation of a durable re- and have a depth of about 4 mm so fat droplets pair tissue providing symptomatic relief, allow- and blood is released from the subchondral bone. ing high physical activity and delaying partial A blood clot is formed to provide the optimal en- or total joint replacement surgery. To achieve vironment for a viable population of pluripotent this, the joint needs a stable equilibrium (joint marrow cells to differentiate into a stable tissue homeostasis), not only of the articular carti- within the lesion. The rehabilitation protocol is lage, but also of the synovium. A cartilage le- an important part of the microfracture proce- sion might only be a consequence of a disturbed dure. Early mobility of the joint with continuous equilibrium. In this case treating only the carti- passive motion is advocated in conjunction with lage lesion will not recover the equilibrium and reduced weight-bearing for 6-8 weeks. the treatment will be less effective [7, 119]. Until Most patients suffer less pain and show now it is not completely understood how to re- an increased capacity for activities of daily living store joint homeostasis. It is not known whether after microfracture therapy. Clinical results are the cartilage structure needs to be completely satisfactory in lesions up to 2cm2. However, mi- normal. It might be that an 80% normal struc- crofracture technique does not induce growth of ture is already good enough to restore joint ho- hyaline-like cartilage, but results in fibrous car- meostasis. Saris et al. [120, 136] treated patients tilage. Whether this defect filling will become with symptomatic cartilage defects with chon- stable over time and support weight bearing is drocyte implantation or microfracture. Clinical still unclear. Clinical data shows that the results results are comparable for both treatments after begin to decline 3-5 years after surgery, showing 5 years. However, chondrocyte implantation led a limited longevity of the repair tissue correlat- to better results when the lesion was treated in ing with the clinical results [120]. The outcome an early stage. Next, the most commonly used has a higher chance of success in young patients techniques are described, as well as the most with small cartilage lesions due to trauma who recent developments and research areas to treat had a high activity level before surgery [18, 59, cartilage defects. 69, 88, 96, 127-130].

22 Autologous chondrocyte transplantation (ACT) and long term follow-up is not yet available [2, To improve joint function a two-step surgical 15, 32, 41, 62, 101]. procedure called autologous cultured chondrocyte transplantation can be performed. In a first Mosaicplasty arthroscopic operation 300 to 500 mg of carti- Osteochondral autografting, or mosaicplasty, lage is obtained from the injured knee. Cells are is usually performed as an open procedure. Al- isolated within 6 hours after the operation in a though, depending on the location of the defect, cell-culture laboratory. The isolated cells are cul- it is possible to perform it arthroscopically. Os- tured for 14 to 21 days in patient’s own serum. teochondral plugs are taken with a cylindrical In a second surgery the lesion is debrided back cutting device and used to fill an articular carti- to the best cartilage available without penetrat- lage defect. The defect is first debrided and then ing the subchondral bone plate. In the classical measured to determine the number and size of technique a periosteal flap is harvested, fitted the grafts. Cylindrical osteochondral plugs of and sutured to the surrounding rim of the le- about 6 to 11 mm in diameter and 15 to 20 mm sion, after which the cultured chondrocytes are long are harvested from non-weight-bearing ar- injected under this periosteal flap [108]. eas with similar curvature as the defect site. In In most cases the integration into the the defect 1 mm smaller sockets are drilled, in surrounding cartilage is good and the failure order to press-fit the previously harvested grafts rate is only about 16%. In 67% of the patients in the right location. At least 80% of the defect the defect is filled with hyaline-like cartilage, needs to be covered by this procedure. Depend- which is twice as stiff as the fibrous tissue in the ing on the defect, weight should not be fully ap- other patients. Weight-bearing seems to promote plied for 1-8 weeks. the formation of hyaline repair tissue. The long- Hyaline cartilage is formed with nice in- term results show that treatment with autologous tegration into the surrounding cartilage in about chondrocyte transplantation results in a durable 80% of the cases. The advantage of mosaicplasty repair for the majority of patients. However, the is that it is a one-stage procedure, it has low costs time required for the tissue to form and the long and morbidity and already living cartilage is im- rehabilitation period until pain-free weight-bear- planted. However, it should not be used when ing is a main disadvantage of this technique [22, (pre)osteoarthritis, inflammatory arthropathies 33, 42, 86, 87, 107, 108, 114]. or tumours are present, and in patients over 50. Other concerns of this method are donor site Matrix-Induced Autologous Chondrocyte Implan- morbidity and the difficulty to produce a smooth, tation (MACI) perfectly congruent joint surface [19, 43, 49, 50, A similar procedure to ACT as described above is 84, 113, 125]. matrix-induced autologous chondrocyte implantation. Cells are harvested in a first arthroscopic Osteochondral allografts operation and cultured in a laboratory as with Besides osteochondral grafts from the patient’s ACT. However, the cultured cells are then seed- own joint, grafts from donors can be used. ed onto a collagen I/III matrix membrane before These osteochondral allografts are preferably implantation. After preparing the lesion in a sec- fresh; otherwise they need to be fresh stored, to ond surgery, the graft is cut in the correct size preserve metabolically active chondrocytes [16, and secured in place with fibrin sealant [41]. 143]. Whole joint specimens are stored in nutri- Short term follow-up shows reliable re- tive medium prior to transplantation. Allograft sults for treated cartilage defects. Treated lesions tissue is size-matched to the host with the use have shown formation of hyaline-like or mixed of radiographs or magnetic resonance imaging hyaline and fibrocartilage repair tissue. The -ad studies. Cylindrical osteochondral allografts of vantages of MACI are that it allows a more mini- 8-15 mm height are harvested from the donor mal access approach surgery, less operating time and press-fitted in the host defect. Weight-bear- and cell distribution can be ensured. Besides ing is restricted after surgery and a rehabilitation this, the number of revision surgeries is reduced program is needed to restore normal gait. due to less hypertrophy and donor-site morbidity. The functionality of the joint increased However, the costs of this procedure is higher after surgery and the cartilage thickness of the

23 allograft is maintained. Even five and fourteen influence cell migration, differentiation, tissue years after surgery, 75% and 63%, respectively, growth and diffusion of oxygen, nutrients, waste of the grafts show a good result. However, re- products and signalling molecules [54, 101, 135]. sults are poor for patients age 60 and older. Graft Big efforts are made to have ECM com- failure can occur due to necrotic bone and ar- ponents in the tissue engineered construct close ticular cartilage fragmentation. The main disad- to native cartilage. It has been shown that pro- vantages of this technique is the availability of teoglycan content can come close to native carti- the donor grafts, the possibility of disease trans- lage [76]. Unfortunately it is not yet possible to mission between donor and host and the limited engineer a construct with close to native amount cartilage viability over time [16, 40, 45–48, 78, and orientation of collagen [54]. Therefore nec- 80, 85, 105, 138, 143]. essary load bearing mechanical properties of the constructs are still much lower than native car- Repair techniques: research areas tilage [118]. Also depth depending matrix content, orientation and stiffness still needs to be Tissue engineering improved [70]. The general drawback of e.g. microfracture and autologous cultured chondrocyte injection is Double network hydrogels that the newly formed tissue lacks the structural Double network hydrogels (DN-gels) are devel- organisation of cartilage. This tissue has inferior oped as possible implant materials for the repair mechanical properties compared to native tissue of soft tissues by Gong and her colleagues [11, and is therefore prone to failure [60]. One ul- 44, 56, 94, 95, 133, 145, 146]. A whole family of timate goal of cartilage tissue engineering is to DN-gels can be made, composed of two kinds develop a replacement that has a structure and of independently interpenetrated polymers (Fig- composition resembling native cartilage, yield- ure 1.3). Of which the first network is stiff and ing similar mechanical behaviour and which brittle and the second network is soft and duc- fully restores joint functionality [70]. tile [95]. In this respect they resemble cartilage Chondrocytes are the most used cell and other skeletal system soft tissues, which are source, but cells harvested from diseased joints also high water-content materials or structures are relatively inactive and less good in forming with a double-network strategy. Cartilage con- cartilage [14, 28]. Chondrocytes from older (os- sists of highly crosslinked collagen-fibres with teoarthritic) patients are metabolically less active proteoglycan gel. DN-gels can be created using compared to young (animal) chondrocytes [30, various synthetic and biological materials, such 52, 106, 140]. To overcome the limited supply as acrylamides, collagen and bacterial cellulose of chondrocytes, multipotent stem cells are also [44, 95]. The double network structure results used [20, 126]. in a high water content material with a much There are several factors which influence higher stiffness than one of the two components the growth and tissue formation during cell cul- separately. For example the DN-gel consist- ture. First, to promote chondrogenic phenotype and to stimulate ECM production a number of growth factors, including transforming growth factor (TGF-b), insulin-like growth factor (IGF- 1) and others can be used [1]. Second, to improve the mechanical properties of the tissue-engineered cartilage, the cells are mechanical stimu- lated during culturing. This is mostly done under direct confined or unconfined compression or hydrostatic pressure [118, 123]. Third, cells can be seeded onto a scaffold, which replaces the function of the native matrix. Scaffold can be synthetic or made from natural materials and Figure 1.3: Structure of double network hydrogels with a can have various architecture, porosity and stiff- stiff and brittle first network and a second soft and ductile ness. These properties are important, since they network entangled within the first network.

24 ing of poly(2-acrylamido-2-methylpropansul- MECHANICAL PROPERTIES OF SOFT fonic acid) (PAMPS) as the first network and MATERIALS poly(acrylamid) (PAAm) as the second network is 43 times stronger than the PAMPS single net- Elastic behaviour work gel [44]. To be suitable for clinical use, the im- When a force is applied to a material, it will plant material has to be biocompatible. Tanabe deform the material. This deformation can oc- et al. [133] implanted four different DN-gels: cur in tension, bending, compression, torsion or PAMPS/PAAm, PAMPS/poly(N,N-Dimetyl shear. The resistance to this elastic deformation acrylamide) (PDMAAm), Cellulose/PDMAAm is a measure of the stiffness of a material. A mea- and Cellulose/Gelatine, in the muscle and the sure of this stiffness in compression and tension subcutaneous tissue of rabbits. The Cellulose/ is the Young’s modulus (E; Equation 1.1). The Gelatin did not show an inflammation reaction, Young’s modulus is the stress (s) divided by the but was gradually absorbed after 4 and 6 weeks strain (e). Where stress is the force (F) normal- of implantation. Thus the Cellulose/Gelatin gel ized to the area (A) over which it is applied and has the potential to be used as an absorbable the strain is the change in height or length (ΔL)

implant. The PAMPS/PAAm and Cellulose/PD- normalized to the original length (L0). MAAm gels showed significant inflammation at s FA0 both time points and are therefore not suitable E = = (1.1) e ∆LL as an implant material. The PAMPS/PDMAAm 0 gel induced only a mild inflammation after 1 E is the Young’s modulus (modulus of week, and decreased at the same degree as the elasticity) negative control at 4 and 6 weeks. The stiffness, F is the force applied on an object

strength and strain to failure did not change af- A0 is the original cross-sectional area ter implantation. This short experiment shows through which the force is applied promising results for the PAMPS/PDMAAm ΔL is the amount by which the length of gel, however, whether it is suitable for implanta- the object changes

tion needs to be further investigated [133]. L0 is the original length of the object To further investigate the biological response on the PAMPS/PDMAAm gel, Yasuda The relationship between stress and et al. [145] created osteochondral defects in strain for linear elastic materials is in general re- rabbits and inserted the DN-gel, poly(vinyl al- ferred to as Hooke‘s Law (Equation 1.2). It also cohol) (PVA) gel or ultrahigh molecular weight applies in small, elastic deformations of other polyethylene (UHMWPE) plugs at the bottom materials [75]. of the defect. Hyaline cartilage was regenerated F= − k ⋅ x (1.2) in the defects with the implanted DN-gel, but F is the force applied rarely by the PVA gel or the UHMWPE plugs. k is a constant called the spring constant However, the depth of the implanted DN-gel x is the displacement of the spring plug affected the regeneration effect, which -im plies that the physical environment may affect When the stress of an elastic material is plotted hyaline-cartilage regeneration. The cells in the versus the strain, this results in a straight line. defect with the DN-gel plug highly expressed The Young’s modulus is the resulting slope of type-2 collagen, aggrecan and the regenerated that line. A material with a higher Young’s mod- matrix was rich in proteoglycan and type-2 col- ulus has a steeper slope and is referred to as stiffer lagen at 4 weeks [145]. This is a promising re- material and vice versa (Figure 1.4a). sult for cartilage regeneration. Nevertheless, the When a material is compressed in one mechanical properties of the regenerated tissue direction, it usually expands in the other two were not determined and it is unclear whether directions perpendicular to the direction of this mechanism also takes place in older human compression. To what extent this takes place, is patients. dependent on the material. For an incompressible material, the expansion in the perpendicular

25 Figure 1.4: Stress-strain diagram of a linearly elastic (a) and linearly viscoelastic (b) material. For an elastic material the slope of the stress-strain curve corresponds to the Young’s modulus. For a viscoelastic material the hysteresis-loop shows the energy loss in a loading and unloading cycle.

directions will be larger than in a compressible als. And the loss angle (d) is a measure of which material. A measure of this effect is the Poisson’s part of the energy is dissipated. The correlation ratio (u). The Poisson’s ratio is the ratio of the between these properties is shown in Figure 1.5. fraction of expansion divided by the fraction of This shows that at least two of those parameters compression, for small values of these changes are needed in order to describe a viscoelastic ma- (Equation 1.3). A perfectly incompressible mate- terial properly. rial has a Poisson’s ratio of exactly 0.5 when it is Since viscoelastic materials have ele- deformed elastically at small strains. ments of both viscous and elastic properties, they exhibit time dependent strain. Three main detrans u = − (1.3) characteristics of viscoelastic materials are: creep, de axial stress relaxation and strain-rate dependent proper- u is the Poisson’s ratio ties. e trans is transverse strain, perpendicular to When a constant stress is applied on a vis- the applied stress coelastic material, the deformation increases in e axial is axial strain, in parallel with the beginning, but slows down until it becomes applied stress nearly constant (Figure 1.6). This phenomenon is called creep. Stress relaxation is observed when Viscoelastic behaviour a viscoelastic material is deformed and held under a constant strain, the stress will rise to a peak Viscoelastic materials exhibit both viscous and and decreases continuously with time (Figure elastic characteristics when undergoing defor- 1.7). Viscoelastic materials behave strain-rate de- mation. Elastic materials deform instantaneously when a force is applied and return to their original state once the force is removed; all the energy is stored (Figure 1.4a). Whereas when a force is applied to a viscous material, it does not deform, it flows like a liquid. When the force is removed it does not return to its original shape, because the force (energy) is dissipated. Viscoelastic materials dissipate part of the energy when a load is applied and then removed. In a stress-strain curve this is observed as hysteresis (Figure 1.4b). There are several ways to describe these properties. The storage (E’) and the loss modulus Figure 1.5: Schematic representation of the correlation (E’’) represent the energy which is stored and dis- between the dynamic modulus E*, the storage modulus E’ sipated respectively. The dynamic modulus (E*) (elastic part), the loss modulus E’’ (viscous part) and the is a measure of stiffness for viscoelastic materi- loss angle d in a viscoelastic material.

26 Figure 1.6: Creep: A viscoelastic material where a constant stress is applied (a) shows a continuous deformation (b) until equilibrium is reached (dotted line).

Figure 1.7: Stress relaxation: A viscoelastic material under constant strain (a) shows a decrease in stress with time (b) until equilibrium is reached (dotted line). pendent. The stiffness increases with increasing als. The loss angle d( ) represents the phase shift deformation rate whereas the part of the energy between stress and strain (Figure 1.9). A purely which is dissipated decreases with increasing de- elastic material has a loss angle of 0°. The more formation rate (Figure 1.8). Viscoelastic materi- energy is dissipated, the higher the loss angle als behave more elastic at high deformation rates will be. and more viscous at low deformation rates [104]. With dynamic mechanical analysis vis- Mechanical properties of articular cartilage coelastic behaviour can be measured. Stress and strain are in phase in elastic materials. The stress- The response of cartilage on mechanical loading strain response is instantaneous and all energy is determines how load is absorbed and distrib- stored. Strain lags stress in viscoelastic materi- uted to the underlying bone. If cartilage stiff-

Figure 1.8: Stress versus strain of viscoelastic materials Figure 1.9: Stress (solid line) and strain (dashed line) ver- is strain-rate dependent. The stiffness increases and the sus time. The loss angle d ( ) is the phase shift between energy dissipation decreases with increasing strain-rate. stress and strain. Solid line: high strain-rate; dashed line: intermediate strain rate; dotted line: low strain-rate.

27 Figure 1.10: Schematic view of measurements performed in (a) confined compression, (b) unconfined compression and (c) indentation to define mechanical properties of cartilage. Within this figure, dark grey indicates impermeable platens or confining chamber, light grey indicates permeable platens and white indicates a cartilage sample. The cartilage sample in indentation (right) is still attached to the subchondral bone (light grey).

ness is too low, the load is transmitted directly flows through the cartilage and across the articu- to the underlying bone. If cartilage stiffness is lar surface [77, 83]. So, besides being viscoelastic, too high, a load is likely to become focused in a cartilage also behaves as a sponge. small region, which can cause tissue damage or pain. Also the energy dissipation of cartilage is Test methods for determining cartilage stiffness important, since it reduces the peak loads in the Determination of the mechanical properties of cartilage itself and the underlying bone. cartilage and other (bio)materials is typically When cartilage is compressed, the nega- done in confined compression [64, 71, 121], un- tively charged aggrecan molecules are pushed confined compression [35, 65, 67, 71, 104] or in- closer together, which increases the repulsive dentation [3, 65, 71, 79, 90, 102, 103] (Figure force. Non-aggregated proteoglycans would not 1.10). be as effective in resisting compressive loads, In confined compressionthe cartilage layer since they are not as easily trapped in the col- is placed in a confining chamber and compressed lagen matrix. When the collagen matrix is dam- with a permeable piston (Figure 1.10a). Expan- aged, the compressive stiffness is also reduced, sion perpendicular to the direction of compres- due to less efficiently trapped proteoglycans. sion is restrained. It is only possible for fluid to Where the compressive stiffness of cartilage is move through the permeable piston, which cre- mainly coming from the proteoglycans, the ates an artificial porous environment. The mod- tensile stiffness of articular cartilage reflects the ulus is determined from the slope of a linear fit stiffness of the collagen network in tension [82]. of the equilibrium stress versus the strain. Another response of cartilage on loading is fluid In unconfined compression the cartilage flow through the tissue. When cartilage is de- layer is placed between two smooth frictionless formed or a pressure difference is applied, fluid impermeable plates (Figure 1.10b). The cartilage

Table 1.2: The main advantages and disadvantages of articular cartilage measurements performed in confined compression, unconfined compression or indentation. Confined Unconfined Indentation compression compression Data processing/modeling + + -- Sample preparation -- - ++ Mapping - - ++ Original geometry -- -- ++ Original (osteochondral) environment -- -- +(+) Measurement flexibility - - ++

28 layer can expand perpendicular to the direction described advantages and disadvantages of all of compression. Here, the modulus can easily be three test methods are summarized in Table 1.2. calculated from the stress strain ratio. In both confined and unconfined compression the cartilage Cartilage response on loading specimens need to be prepared for the measure- Two mechanisms are responsible for the me- ments. The subchondral bone has to be removed chanical properties of articular cartilage; a from the cartilage, which destroys the original flow-dependent and a flow-independent mecha- osteochondral environment. Further the carti- nism. The flow-independent viscoelastic behav- lage needs to fit perfectly in the measurement iour comes from the intermolecular friction in set-up to get proper results. The contact surfaces the collagen-proteoglycan matrix. Whereas the have to be completely flat and parallel to each flow-dependent behaviour originates from the other. Due to the sample preparation the initial interstitial fluid flow, which can be seen in creep geometry is interrupted and likely induces struc- and stress relaxation experiments [9, 35, 64, 65, tural changes at the edges of the sample. This 67, 71, 72, 82, 90, 97, 121]. It is shown that car- interruption of the initial geometry is even worse tilage has a long equilibration time and therefore in unconfined compression due to expansion no equilibrium state occurs in daily living be- perpendicular to the direction of compression of cause the joints are always moving. This implies the specimens. that, in normal cartilage, fluid pressurization is During indentation measurements a part an important load-support mechanism. of the cartilage is compressed with an indenter (Figure 1.10c). This indenter can be plane-ended The mechanical properties of articular carti- or spherical and permeable or impermeable. In- lage are deformation rate-dependent. Cartilage dentation measurements do not need thorough stiffness increases and its energy dissipation specimen preparation and can even be per- decreases with increasing deformation rate [82, formed in vivo [79]. The theoretical analysis of 104]. Park et al. [104] showed that the dynamic the data is complex and requires a mathemati- modulus of cartilage increases by a factor 2 due cal model. Hayes’ method is frequently used to to its viscoelasticity when the deformation rate determine cartilage stiffness [51]. However, it is is increased from 0.1 to 40 Hz. Their theoretical based on the assumption that cartilage is a ho- studies suggest that flow-dependent viscoelastic- mogeneous, linear elastic material rather than a ity is less significant than flow-independent vis- complex structure. But even viscoelastic models coelasticity at higher frequencies [55]. At 40Hz do not describe cartilage properly, since besides the loss angle reduced to zero, which implies being viscoelastic, cartilage is also porous and that above this frequency cartilage behaves prin- water can move within the structure. Indenta- cipally as an elastic solid. No further increase in tion behaviour is affected by the inhomogeneity dynamic modulus is to be expected in this case and anisotropy of the cartilage and preferably [104]. Fulcher et al. [37] showed that the stor- the model takes all of this into account. But even age modulus increases with increasing frequency. though these models are a simplification, when However, this increase levels out into a plateau the same mathematical model and measurement before a frequency of 92 Hz is reached. How- method is used, data can be used for compari- ever, the loss modulus stayed constant over the sons. Besides this, it is also useful in increasing whole frequency range tested (1 to 92 Hz). Thus, the understanding of cartilage behaviour. the dynamic modulus increased with increasing frequency, but levelled out into a plateau. The Differences found in the mechanical proper- loss angle decreased with increasing frequency ties determined by the different measurement until the plateau is reached, but did not de- methods can partly be explained by the differ- crease to zero. At all frequencies a much higher ent specimen preparation (intact versus not in- storage modulus than viscous modulus was ob- tact cartilage) and mathematical models used. served, which means that more energy is stored In addition to that, in confined and unconfined by the tissue than dissipated and that this effect compression the whole cartilage tissue is mea- is greater at higher frequencies [37]. A possible sured, whereas in indentation mainly the super- reason for increased stiffness, with increasing de- ficial zone is measured [63, 68, 71]. The above formation rate, is that the time for fluid flow in

29 the proteoglycan gel is short compared with the important, since they determine the functional periodic time associated with high-frequency si- behaviour of cartilage. Magnetic Resonance Im- nusoidal loading. However, these viscous effects aging (MRI) is shown to be useful in obtaining cannot only be explained by fluid flow. morphology data of healthy and progressed OA Not only are the mechanical properties cartilage. Unfortunately, early OA does not lead of cartilage deformation rate dependent, they are to detectable morphological changes. also depth dependent. The equilibrium confined It is widely accepted that the mechanical compression modulus is the lowest at the surface properties of cartilage depend on its composition and increases with depth. Thus the actual me- and structural characteristics. Thus a lot of effort chanical behaviour of cartilage is different from is made to determine whether changes in these that predicted under the assumption of tissue characteristics due to OA could be detected in homogeneity [121]. However, the tensile modu- mechanical tests. Several studies were performed lus is the highest at the surface and decreases to investigate the changes in mechanical proper- with depth, because of the high concentration ties due to cartilage degradation. There are three and high degree of orientation of collagen fibrils types of studies performed; 1) OA-like changes in the superficial zone [82]. were investigated, where some components of the cartilage was modified by using a degradation Cartilage is an adaptive tissue, which can be medium; 2) OA-like changes in animal models, seen when looking at the mechanical proper- with induced or spontaneous OA; 3) spontane- ties in different joints, and on different loca- ous occurring OA in vivo in humans [68]. tions within one joint. Several studies found lo- In the first group the proteoglycan con- cation dependent properties [3, 10, 39, 79, 117, tent decreased between 60% and 90%, mainly 132]. In several studies the highest stiffness was in the superficial zone. A correlation was found found in the load bearing areas of the condyles, between cartilage stiffness and the proteoglycan while the tibial and patellar joint surfaces had content. In some cases an increase in collagen softest cartilage [10, 79, 132]. Cartilage of the type II was found. And structural changes were tibial plateau was thinner and the instantaneous observed when a collagen degenerating medium stiffness was higher on locations covered by the was used [74, 97, 98, 112, 139]. In the second meniscus [134]. Appleyard et al. found the high- group an increase in water content was observed est dynamic shear modulus in the lateral outer and a proteoglycan content decrease, whereas no region and the lowest in the medial inner region collagen changes were found. In most cases a de- of ovine tibial plateaus. A high variation in carti- crease in cartilage stiffness was found, although lage stiffness was found between individuals and in some cases this decrease was only temporarily between locations, whereas the energy dissipa- and stiffness increased to near normal again [3, tion was found to be relatively constant on ovine 103]. In the third group the results were com- tibial plateaus [3]. parable with those found in the animal models A possible explanation for these differ- concerning the decrease in mechanical proper- ences is that there are variations in the cartilage ties and proteoglycan content and the increase composition. It is shown that the compressive of water content. More structural changes were stiffness of cartilage in creep experiments in- observed as well as degraded collagen [13, 100, creases as a function of the glycosaminoglycan 115]. content. However, no correlation was found be- These studies showed that a decrease in tween compressive stiffness and collagen content stiffness was found in static and dynamic mea- [66, 117]. Besides the ECM also the water con- surement methods. However, a mechanical test- tent influences the mechanical properties of car- ing device should have a high accuracy and re- tilage. As the water content increases, cartilage producibility to detect small changes in stiffness becomes less stiff and more permeable [6]. in early OA. Besides this, the high inter-subject and location variation in stiffness of articular Changes in mechanical properties in osteoarthritis cartilage will complicate the detection of early Early detection of osteoarthritis (OA) is necessary OA as is shown by Brown et al. [23] The stiffness to prevent or reduce long-term disability. Both of visually normal, artificially degraded and nat- morphological and mechanical properties are urally osteoarthritic articular cartilage of bovine

30 patellae using a micro-indentation device was tissue strain, in order to predict damage which investigated. They found a 25% decrease in stiff- can occur performing these indentation tests. As ness after proteoglycan depletion, however, when expected, indenting deeper into the cartilage in- compared to the stiffness of visually normal car- creases the strain magnitude, whereas indenter tilage, only 17% of the samples lie outside the geometry only slightly influences the peak strain. normal range. Because of the high variability in Above described models give more insight in the stiffness of normal samples, indentation data articular cartilage behaviour. However, a major cannot accurately distinguish between normal drawback is that a model always will be a sim- and abnormal articular cartilage samples [23]. plification of the real situation. Therefore the re- Stolz et al. [131] examined cartilage bi- sults will be just an approximation of what really opsies from seven patients undergoing total hip happens. or knee replacement using atomic force microscopy. It was not possible to distinguish between healthy and osteoarthritic cartilage by determin- AIM OF THIS THESIS ing the micro-stiffness, whereas the nano-stiffness decreased from 83 kPa (healthy) to 5.6 kPa Because of the particular micro-architecture of (osteoarthritic). These changes were clearly de- biologic materials, techniques to measure their picted before any morphological changes could mechanical properties are complex. Indentation be observed using current diagnostic methods. measurements need a mathematical model to Although this is a clear difference, it might be calculate the stiffness out of the force and dis- challenging to be able to detect early changes placement data. The most commonly used mod- of osteoarthritis due to an increase in cartilage el (Hayes) assumes that cartilage is a linear elas- nano-stiffness with age. It might be difficult to tic material and it requires calculation of a factor distinguish healthy cartilage from early osteo- k, a complex function depending on indentation arthritic cartilage from older patients, since age depth, cartilage thickness and Poisson’s ratio [51]. increases the stiffness, whereas osteoarthritis de- Since cartilage is not linear elastic, the results creases the stiffness [131]. are compared with another mathematical model (Kren) in which linear elasticity is not assumed. Modelling cartilage behaviour However, this other model does not take carti- In order to improve understanding of cartilage lage thickness into account and treats the car- behaviour several groups made an effort in mod- tilage as a material rather than a structure [73]. elling articular cartilage [4, 5, 12, 29, 36, 90- Many studies have been performed to 93]. Mow et al. [90-93] modelled cartilage as a study the mechanical properties of healthy and mixture of fluid and solid components. In this osteoarthritic cartilage. Most of these studies fo- modelling, all of the solid-like components, e.g. cussed on how cartilage stiffness changes due to proteoglycans, collagen and cells, are taken to- variation in ECM components, structure, grade gether to constitute the solid phase. The fluid of degradation or location within a joint. But phase, consisting of the interstitial fluid, is free since cartilage is a viscoelastic material, not only to move through the matrix. Typically, the solid stiffness is an important material property, also phase is modelled as an incompressible elastic energy handling is a key property (Figure 1.5). material, and the fluid phase as incompressible Detecting early osteoarthritis might be improved and without viscosity [92]. The biggest drawback by not only looking at the stiffness, but also at of this and other biphasic/poroelastic models [5, the energy dissipation. The variation in energy 36, 92] is that cartilage is seen as a homogeneous dissipation is smaller [3] and early changes in material. Therefore efforts were made to incor- human osteoarthritic cartilage are increased wa- porate the inhomogeneous nature of cartilage ter content and decreased proteoglycan content into these models [29]. Here it was clearly seen [82]. Besides this, cartilage has strain rate depen- that the stiffness increased from the superficial dent behaviour (as other viscoelastic materials) to the deep zone and that the value of the homo- and it undergoes different loading patterns in geneous model lies in between these values. daily living. Thus, both cartilage stiffness and Bae et al. [12] modelled the effects of in- energy dissipation at different deformation rates denter geometry and indentation depth on intra- were investigated.

31 However, this does not solve the problem conditions related to function: fast impact and of the lack of sufficient treatment possibilities of slow sinusoidal mode. For equivalent anatomic cartilage defects. Therefore, other possible carti- locations, there was no difference in dynamic lage repair materials need to be further investi- modulus. However, the loss angle of the human gated. Since DN-gels showed promising results cartilage was ~35% lower in fast impact and to function as a cartilage repair material [11, 133, ~12% higher in slow sinusoidal mode. These dif- 145], dynamic stiffness and surgery-related - at ferences seem attributable to age (young swine tachment mechanics were determined. Further, cartilage and older human cartilage) but also to it would be advantageous when the stiffness pa- species anatomy and biology. Test mode-related rameters of those DN-gels would be tuneable to differences in human-swine loss angle support mimick the mechanical properties of cartilage. use of multiple function-related test modes. And keeping loss angle differences in mind, swine The aim of this thesis is to increase the under- specimens could serve as a standard of compari- standing of the mechanical behaviour in articu- son for mechanical evaluation of e.g. engineered lar cartilage specimens, including energy dissi- cartilage or synthetic repair materials. pation and to further investigate the feasibility of DN-gels to become a cartilage replacement Nowadays in focal repair of joint cartilage and material. meniscus, initial stiffness and strength of repairs are generally much less than the surrounding tissue. This increases early failure potential. Secure OUTLINE OF THIS THESIS primary fixation of the repair material is also a problem. In chapter 4 it was evaluated whether Hayes’ method [51] is widely used to determine acrylamide polymer double-network hydrogels articular cartilage mechanical properties. The (DN-gels) could serve as a repair material. Me- main disadvantages are that it assumes linear chanical properties related to surgical use were elastic behaviour and requires calculation of a tested in two types of DN-gels and the results factor k, which is a complex function of indenta- were compared to that of swine meniscus and car- tion depth, cartilage thickness and Poisson’s ratio. tilage. Remarkably, these >90%-water DN-gels In chapter 2 Hayes’ method is compared with exhibited dynamic modulus values approaching a new, simplified mathematical model to deter- swine meniscus (up to 50%). However, the en- mine stiffness by Kren [73]. This model does nei- ergy-absorbing capability of these DN-gels was ther assume linear elasticity nor determination much lower than that of swine meniscus. Also, of k. Both models were applied on indentation fine 4/0 suture tear-out strength approached car- data on swine cartilage specimens. Differences tilage. Initial strength of attachment of DN-gels were found between the determined dynamic to cartilage with acrylic tissue adhesive was also modulus. The modulus determined by the mod- high. DN-gel strength and toughness properties els correlated well together, which confirms that stem from optimized entanglement of the two one model should be used in order to be able network components. DN-gels thus have obvi- to compare different specimens. To reduce the ous structural parallels with cartilaginous tissues, number of assumptions and to be able to use the and their surgical handling properties make same model for both articular cartilage and non- them ideal candidates for clinical use. However, linear elastic materials, Kren’s model is used in the initial stiffness of these DN-gels is still lower the rest of this thesis to calculate the mechanical than cartilage stiffness. properties. DN-gels have shown to be an attractive repair In chapter 3 the mechanical properties of swine material for skeletal system soft tissues. They ex- cartilage were compared with those of human ist in a very wide range, with different compo- cartilage. Since fresh healthy human articu- sitions, with corresponding differences in stiff- lar cartilage is not readily available, we tested ness, biocompatibility, etc. In chapter 4 it has whether swine cartilage could serve as a suitable been shown that their surgical handling proper- substitute for mechanical comparisons. Carti- ties as well as the ability to attach them to the lage stiffness was tested under different loading surrounding tissue make them very good can-

32 didates, but in the stiffness and energy handling increased with decreasing water content in fast- properties there was still room for improvement. impact, but not in slow-sinusoidal deformation, In chapter 5 it was investigated whether it was and is still much lower compared to cartilage. possible to create a DN-gel, which is as stiff as This results in a different rate dependency. It is cartilage. Stiffness properties of three different possible to adapt the chemical composition of water content DN-gels were determined and DN-gels in such a way that most of their biome- compared. The dynamic modulus increased with chanical properties are close to those of healthy decreasing water content in both testing modes cartilage. and resembles that of cartilage. The loss angle

33 REFERENCES Indentation testing of human articular cartilage: effects of probe tip geometry and inden- 1. Ahmed TAE, Hincke MT. Strategies for ar- tation depth on intra-tissue strain. Journal of ticular cartilage lesion repair and functional Biomechanics 39: 1039-1047, 2006. restoration. Tissue Engineering. Part B 16: 13. Bank RA, Soudry M, Maroudas A, Mizrahi 305-329, 2010. J, TeKoppele JM. The increased swelling and 2. Albrecht C, Tichy B, Nürnberger S, Hosiner instantaneous deformation of osteoarthritic S, Zak L, Aldrian S, Marlovits S. Gene ex- cartilage is highly correlated with collagen pression and cell differentiation in matrix-as- degradation. Arthritis and Rheumatism 43: sociated chondrocyte transplantation grafts: 2202-2210, 2000. a comparative study. Osteoarthritis and Carti- 14. Barbero A, Grogan S, Schäfer D, Heberer M, lage 19: 1219-1227, 2011. Mainil-Varlet P, Martin I. Age related chang- 3. Appleyard R. Topographical analysis of the es in human articular chondrocyte yield, pro- structural, biochemical and dynamic biome- liferation and post-expansion chondrogenic chanical properties of cartilage in an ovine capacity. Osteoarthritis and Cartilage 12: 476- model of osteoarthritis. Osteoarthritis and 484, 2004. Cartilage 11: 65-77, 2003. 15. Bauer S, Khan RJK, Ebert JR, Robertson 4. Argatov I, Mishuris G. Elliptical contact of WB, Breidahl W, Ackland TR, Wood DJ. thin biphasic cartilage layers: exact solution Knee joint preservation with combined neu- for monotonic loading. Journal of Biomechan- tralising High Tibial Osteotomy (HTO) and ics 44: 759-761, 2011. Matrix-induced Autologous Chondrocyte 5. Armstrong CG, Lai WM, Mow VC. An Implantation (MACI) in younger patients analysis of the unconfined compression of with medial knee osteoarthritis: A case series articular cartilage. Journal of Biomechanical with prospective clinical and MRI follow-up Engineering 106: 165-173, 1984. over 5years. The Knee (July 20, 2011). 6. Armstrong CG, Mow VC. Variations in the 16. Beaver R, Mahomed M, Backstein D, Davis intrinsic mechanical properties of human ar- A, Zukor D, Gross A. Fresh osteochondral ticular cartilage with age, degeneration, and allografts for post-traumatic defects in the water content. The Journal of Bone and Joint knee. A survivorship analysis. Journal of Bone Surgery 64: 88-94, 1982. and Joint Surgery 74: 105-110, 1992. 7. Arnold MP, Hirschmann MT, Verdonk 17. Bedi A, Haidukewych GJ. Management of PCM. See the whole picture: knee preserving the Posttraumatic Arthritic Knee. Journal therapy needs more than surface repair. Knee of the American Academy of Orthopaedic Sur- Surgery, Sports Traumatology, Arthroscopy geons 17: 88-101, 2009. 20(2): 195-197, 2012. 18. Blevins FT, Steadman JR, Rodrigo JJ, Silli- 8. Aroen A. Articular Cartilage Lesions in 993 man J. Treatment of articular cartilage de- Consecutive Knee Arthroscopies. American fects in athletes: an analysis of functional Journal of Sports Medicine 32: 211-215, 2004. outcome and lesion appearance. Orthopedics 9. Athanasiou KA, Agarwal A, Dzida FJ. Com- 21: 761-768, 1998. parative study of the intrinsic mechanical 19. Bobić V. Arthroscopic osteochondral au- properties of the human acetabular and tograft transplantation in anterior cruciate femoral head cartilage. Journal of orthopaedic ligament reconstruction: a preliminary clini- research 12: 340-349, 1994. cal study. Knee Surgery, Sports Traumatology, 10. Athanasiou KA, Rosenwasser MP, Buckwal- Arthroscopy 3: 262-264, 1996. ter JA, Malinin TI, Mow VC. Interspecies 20. Boeuf S, Richter W. Chondrogenesis of mes- comparisons of in situ intrinsic mechanical enchymal stem cells: role of tissue source properties of distal femoral cartilage. Journal and inducing factors. Stem Cell Research and of Orthopaedic Research 9: 330-340, 1991. Therapy 1: 1–9, 2010. 11. Azuma C, Yasuda K, Tanabe Y, Taniguro H, 21. Borrelli J, Ricci WM. Acute effects of car- Kanaya F, Nakayama A, Chen YM, Gong JP, tilage impact. Clinical Orthopaedics and Re- Osada Y. Biodegradation of high-toughness lated Research : 33-39, 2004. double network hydrogels as potential ma- 22. Brittberg M, Lindahl A, Nilsson A, Ohlsson terials for artificial cartilage. Journal of Bio- C, Isaksson O, Peterson L. Treatment of deep medical Materials Research. Part A 81: 373- cartilage defects in the knee with autologous 380, 2007. chondrocyte transplantation. New England 12. Bae WC, Lewis CW, Levenston ME, Sah RL. Journal of Medicine 331: 889–895, 1994.

34 23. Brown CP, Crawford RW, Oloyede A. In- A, Thonar EJ, Masuda K, Chen AC, Sah RL, dentation stiffness does not discriminate Davol A, Klisch SM. Articular cartilage me- between normal and degraded articular car- chanical and biochemical property relations tilage. Clinical Biomechanics 22: 843-848, before and after in vitro growth. Journal of 2007. Biomechanics 40: 3607-3614, 2007. 24. Buckwalter J, Mankin H, Grodzinsky A. Ar- 36. Frank EH, Grodzinsky AJ. Cartilage elec- ticular cartilage and osteoarthritis. Instruc- tromechanics--I. Electrokinetic transduction tional Course Lectures 54: 465-480, 2005. and the effects of electrolyte pH and ionic 25. Buckwalter J, Mow V, Ratcliffe A. Restora- strength. Journal of Biomechanics 20: 615- tion of Injured or Degenerated Articular 627, 1987. Cartilage. The Journal of the American Acad- 37. Fulcher GR, Hukins DWL, Shepherd DET. emy of Orthopaedic Surgeons 2: 192-201, 1994. Viscoelastic properties of bovine articular 26. Buckwalter JA. Mechanical injuries of artic- cartilage attached to subchondral bone at ular cartilage. The Iowa Orthopaedic Journal high frequencies. BMC Musculoskeletal Dis- 12: 50-57, 1992. orders 10: 61, 2009. 27. Buckwalter JA. Articular cartilage: injuries 38. Gaissmaier C, Fritz J, Schewe B, Weise K, and potential for healing. The Journal of Or- Mollenhauer J, Aicher W. Cartilage Defects: thopaedic and Sports Physical Therapy 28: 192- Epidemiology and Natural History. Osteosyn- 202, 1998. thesis and Trauma Care 14: 188-194, 2006. 28. Candrian C, Miot S, Wolf F, Bonacina E, 39. Garcia-Seco E, Wilson DA, Cook JL, Kuroki Valderrabano V, Dickinson S, Wirz D, Jakob K, Kreeger JM, Keegan KG. Measurement of M, Daniels AU, Martin I, Barbero A. Are an- articular cartilage stiffness of the femoropa- kle chondrocytes from damaged fragments a tellar, tarsocrural, and metatarsophalangeal suitable cell source for cartilage repair ? In: joints in horses and comparison with bio- EORS. 2008, p. 110054-110054. chemical data. Veterinary Surgery 34: 571- 29. Chen AC, Bae WC, Schinagl RM, Sah RL. 578, 2005. Depth- and strain-dependent mechanical 40. Ghazavi M, Pritzker K, Davis A, Gross and electromechanical properties of full- A. Fresh osteochondral allografts for post- thickness bovine articular cartilage in con- traumatic osteochondral defects of the knee. fined compression. Journal of Biomechanics Journal of bone and joint surgery 79: 1008– 34: 1-12, 2001. 1013, 1997. 30. Dehne T, Karlsson C, Ringe J, Sittinger M, 41. Gibson A, Mcdonnell S, Price A. Matrix- Lindahl A. Chondrogenic differentiation Induced Autologous Chondrocyte Implanta- potential of osteoarthritic chondrocytes and tion. Operative Techniques in Orthopaedics 16: their possible use in matrix-associated autol- 262-265, 2006. ogous chondrocyte transplantation. Arthritis 42. Gillogly SD, Voight M, Blackburn T. Treat- Research and Therapy 11, 2009. ment of articular cartilage defects of the knee 31. Donohue JM, Buss D, Oegema T, Thomp- with autologous chondrocyte implantation. son R. The effects of indirect blunt trauma The Journal of Orthopaedic and Sports Physical on adult canine articular cartilage. Journal of Therapy 28: 241-251, 1998. Bone and Joint Surgery 65: 948–957, 1983. 43. Gobbi A, Francisco RA, Lubowitz JH, Al- 32. Enea D, Cecconi S, Busilacchi A, Manzotti legra F, Canata G. Osteochondral lesions of S, Gesuita R, Gigante A. Matrix-induced au- the talus: randomized controlled trial com- tologous chondrocyte implantation (MACI) paring chondroplasty, microfracture, and in the knee. Knee Surgery, Sports Traumatol- osteochondral autograft transplantation. Ar- ogy, Arthroscopy (August 12, 2011). throscopy 22: 1085-1092, 2006. 33. Erggelet C, Steinwachs MR, Reichelt A. The 44. Gong JP, Katsuyama Y, Kurokawa T, Osada operative treatment of full thickness carti- Y. Double-Network Hydrogels with Extreme- lage defects in the knee joint with autologous ly High Mechanical Strength. Advanced Ma- chondrocyte transplantation. Saudi Medical terials 15: 1155-1158, 2003. Journal 21: 715-721, 2000. 45. Gross AE, McKee NH, Pritzker KP, Langer 34. Fetter NL, Leddy HA, Guilak F, Nunley JA. F. Reconstruction of skeletal deficits at the Composition and transport properties of hu- knee. A comprehensive osteochondral trans- man ankle and knee cartilage. Journal of Or- plant program. Clinical Orthopaedics and Re- thopaedic Research 24: 211-219, 2006. lated Research : 96-106, 1983. 35. Ficklin T, Thomas G, Barthel JC, Asanbaeva 46. Gross AE, Shasha N, Aubin P. Long-term

35 followup of the use of fresh osteochondral articulating cartilages. Philosophical Transac- allografts for posttraumatic knee defects. tions of The Royal Society 42: 514-521, 1743. Clinical Orthopaedics and Related Research : 59. Hunziker EB. Articular cartilage repair: ba- 79-87, 2005. sic science and clinical progress. A review of 47. Gross AE, Silverstein EA, Falk J, Falk R, the current status and prospects. Osteoarthri- Langer F. The allotransplantation of partial tis and Cartilage 10: 432-463, 2001. joints in the treatment of osteoarthritis of the 60. Hunziker EB. The elusive path to cartilage knee. Clinical Orthopaedics and Related Re- regeneration. Advanced Materials 21: 3419- search : 7-14, 1975. 3424, 2009. 48. Gross AE. Fresh osteochondral allografts for 61. Jackson RW, Dieterichs C. The results of post-traumatic knee defects: Surgical tech- arthroscopic lavage and debridement of os- nique. Operative Techniques in Orthopaedics teoarthritic knees based on the severity of 7: 334-339, 1997. degeneration: a 4- to 6-year symptomatic 49. Hangody L, Füles P. Autologous osteochon- follow-up. Arthroscopy 19: 13-20, 2003. dral mosaicplasty for the treatment of full- 62. Jacobi M, Villa V, Magnussen RA, Neyret P. thickness defects of weight-bearing joints: MACI - a new era? Sports Medicine, Arthros- ten years of experimental and clinical experi- copy, Rehabilitation, Therapy and Technol- ence. The Journal of Bone and Joint Surgery ogy 3: 10, 2011. 85-A Suppl: 25-32, 2003. 63. Julkunen P, Korhonen RK, Herzog W, Jurv- 50. Hangody L, Kish G, Kárpáti Z, Szerb I, Ud- elin JS. Uncertainties in indentation testing varhelyi I. Arthroscopic autogenous osteo- of articular cartilage: a fibril-reinforced po- chondral mosaicplasty for the treatment of roviscoelastic study. Medical Engineering and femoral condylar articular defects. A prelimi- Physics 30: 506-515, 2008. nary report. Knee Surgery, Sports Traumatol- 64. Jurvelin J, Buschmann M, Hunziker E. Op- ogy, Arthroscopy 5: 262-267, 1997. tical and mechanical determination of pois- 51. Hayes W, Keer L, Herrmann G, Mockros L. son’s ratio of adult bovine humeral articular A mathematical analysis for indentation tests cartilage. Journal of Biomechanics 30: 235- of articular cartilage. Journal of Biomechanics 241, 1997. 5: 541-551, 1972. 65. Jurvelin J, Kiviranta I, Säämänen A-M, Tam- 52. Hidaka C, Cheng C, Alexandre D, Bhargava mi M, Helminen HJ. Indentation stiffness of M, Torzilli PA. Maturational differences in young canine knee articular cartilage—In- superficial and deep zone articular chondro- fluence of strenuous joint loading. Journal of cytes. Cell and Tissue Research 323: 127-135, Biomechanics 23: 1239-1246, 1990. 2006. 66. Kempson GE, Muir H, Swanson SA, Free- 53. Hjelle K, Solheim E, Strand T, Muri R, Britt- man MA. Correlations between stiffness and berg M. Articular cartilage defects in 1000 the chemical constituents of cartilage on the knee arthroscopies. Arthroscopy 18: 730-734, human femoral head. Biochimica et Biophysi- 2002. ca Acta 215: 70-77, 1970. 54. Hu JC, Athanasiou K a. A self-assembling 67. Kleemann RU, Krocker D, Cedraro A, process in articular cartilage tissue engineer- Tuischer J, Duda GN. Altered cartilage me- ing. Tissue Engineering 12: 969-979, 2006. chanics and histology in knee osteoarthritis: 55. Huang C-Y, Mow VC, Ateshian GA. The relation to clinical assessment (ICRS Grade). Role of Flow-Independent Viscoelasticity in Osteoarthritis and Cartilage 13: 958-963, the Biphasic Tensile and Compressive Re- 2005. sponses of Articular Cartilage. Journal of Bio- 68. Knecht S, Vanwanseele B, Stüssi E. A review mechanical Engineering 123: 410, 2001. on the mechanical quality of articular carti- 56. Huang M, Furukawa H, Tanaka Y, Nakaji- lage - implications for the diagnosis of osteo- ma T, Osada Y, Gong JP. Importance of En- arthritis. Clinical Biomechanics 21: 999-1012, tanglement between First and Second Com- 2006. ponents in High-Strength Double Network 69. Knutsen G, Engebretsen L, Ludvigsen TC, Gels. Macromolecules 40: 6658-6664, 2007. Drogset JO, Grøntvedt T, Solheim E, Strand 57. Hubbard MJ. Articular debridement versus T, Roberts S, Isaksen V, Johansen O. Autolo- washout for degeneration of the medial femo- gous chondrocyte implantation compared ral condyle. A five-year study. The Journal of with microfracture in the knee. A random- Bone and Joint Surgery 78: 217-219, 1996. ized trial. The Journal of Bone and Joint Sur- 58. Hunter W. Of the structure and disease of gery 86-A: 455-464, 2004.

36 70. Kock L, van Donkelaar CC, Ito K. Tissue 82. Mankin H, Mow V, Buckwalter J, Iannotti engineering of functional articular cartilage: J, Ratcliffe A. Articular cartilage structure, the current status. Cell and Tissue Research composition, and function. In: Orthopaedic (October 27, 2011). basic science: biology and biomechanics of 71. Korhonen RK, Laasanen MS, Töyräs J, Ri- the musculoskeletal system. American Acad- eppo J, Hirvonen J, Helminen HJ, Jurvelin emy of Orthopaedic Surgeons Publishers J. Comparison of the equilibrium response Rosemont, IL, 443–470, 2000 of articular cartilage in unconfined compres- 83. Maroudas A, Bullough P. Permeability of sion, confined compression and indentation. Articular Cartilage. Nature 219: 1260-1261, Journal of Biomechanics 35: 903-909, 2002. 1968. 72. Korhonen RK, Wong M, Arokoski J, Lind- 84. Matsusue Y, Yamamuro T, Hama H. Ar- gren R, Helminen HJ, Hunziker EB, Jurv- throscopic multiple osteochondral trans- elin J. Importance of the superficial tissue plantation to the chondral defect in the knee layer for the indentation stiffness of articular associated with anterior cruciate ligament cartilage. Medical Engineering and Physics 24: disruption. Arthroscopy 9: 318-321, 1993. 99-108, 2002. 85. McDermott AG, Langer F, Pritzker KP, 73. Kren AP, Rudnitskii VA, Deikun IG. Deter- Gross AE. Fresh small-fragment osteochon- mining the viscoelastic parameters of vulca- dral allografts. Long-term follow-up study nisates by the dynamic indentation method on first 100 cases. Clinical Orthopaedics and using a non-linear deformation model. Inter- Related Research : 96-102, 1985. national Polymer Science and Technology 32: 86. Micheli LJ, Browne JE, Erggelet C, Fu F, 19-23, 2005. Mandelbaum B, Moseley JB, Zurakowski D. 74. Laasanen MS, Töyräs J, Hirvonen J, Saarak- Autologous chondrocyte implantation of the kala S, Korhonen RK, Nieminen MT, Kivi- knee: multicenter experience and minimum ranta I, Jurvelin JS. Novel mechano-acoustic 3-year follow-up. Clinical Journal of Sport technique and instrument for diagnosis of Medicine 11: 223-228, 2001. cartilage degeneration. Physiological Measure- 87. Minas T. Autologous chondrocyte implanta- ment 23: 491-503, 2002. tion for focal chondral defects of the knee. 75. Lakes RS. Viscoelastic solids. 1999. Clinical Orthopaedics and Related Research: 76. Lima EG, Bian L, Mauck RL, Byers BA, S349-361, 2001. Tuan RS, Ateshian GA, Hung CT. The ef- 88. Mithoefer K, Williams RJ, Warren RF, fect of applied compressive loading on tissue- Wickiewicz TL, Marx RG. High-impact ath- engineered cartilage constructs cultured with letics after knee articular cartilage repair: a TGF-beta3. Conference proceedings: Annu- prospective evaluation of the microfracture al International Conference of the IEEE En- technique. The American Journal of Sports gineering in Medicine and Biology Society. 1: Medicine 34: 1413-1418, 2006. 779-782, 2006. 89. Moseley JB, O’Malley K, Petersen NJ, Men- 77. Linn FC, Sokoloff L. Movement and compo- ke TJ, Brody BA, Kuykendall DH, Hol- sition of interstitial fluid of cartilage. Arthri- lingsworth JC, Ashton CM, Wray NP. A tis and Rheumatism 8: 481-494, 1965. controlled trial of arthroscopic surgery for 78. Locht R, Gross A, Langer F. Late osteochon- osteoarthritis of the knee. New England Jour- dral allograft resurfacing for tibial plateau nal of Medicine 347: 81–88, 2002. fractures. The Journal of Bone and Joint Sur- 90. Mow V, Gibbs M, Lai W, Zhu W, Athana- gery 66: 328-335, 1984. siou KA. Biphasic indentation of articular 79. Lyyra T, Kiviranta I, Väätäinen U, Helmin- cartilage--II. A numerical algorithm and an en HJ, Jurvelin J. In vivo characterization of experimental study. Journal of Biomechanics indentation stiffness of articular cartilage in 22: 853-861, 1989. the normal human knee. Journal of Biomedi- 91. Mow V, Holmes M, Lai W. Fluid transport cal Materials Research 48: 482-487, 1999. and mechanical properties of articular car- 80. Malinin T, Temple HT, Buck BE. Transplan- tilage: a review. Journal of Biomechanics 17: tation of osteochondral allografts after cold 377-394, 1984. storage. The Journal of Bone and Joint Surgery 92. Mow VC, Kuei SC, Lai WM, Armstrong 88: 762-770, 2006. CG. Biphasic creep and stress relaxation of 81. Mandelbaum B, Browne J, Fu F. Articular articular cartilage in compression? Theory cartilage lesions of the knee. American Jour- and experiments. Journal of Biomechanical nal of Sports Medicine 26: 853-861, 1998. Engineering 102: 73-84, 1980.

37 93. Mow VC, Lai WM. Recent Developments in tilage in a sheep model of osteoarthritis. Os- Synovial Joint Biomechanics. SIAM Review teoarthritis and Cartilage 12: 667-679, 2004. 22: 275-317, 1980. 104. Park S, Hung C, Ateshian G. Mechanical 94. Nakajima T, Furukawa H, Tanaka Y, Ku- response of bovine articular cartilage under rokawa T, Osada Y, Gong JP. True Chemi- dynamic unconfined compression loading at cal Structure of Double Network Hydrogels. physiological stress levels. Osteoarthritis and Macromolecules 42: 2184-2189, 2009. Cartilage 12: 65-73, 2004. 95. Nakayama A, Kakugo A, Gong JP, Osada Y, 105. Parrish F. Allograft Replacement of All or Takai M, Erata T, Kawano S. High Mechani- Part of the End of a Long Bone Following cal Strength Double-Network Hydrogel with Excision of a Tumor. The Journal of Bone and Bacterial Cellulose. Advanced Functional Ma- Joint Surgery 55: 1-22, 1973. terials 14: 1124-1128, 2004. 106. Pestka JM, Schmal H, Salzmann G, Hecky 96. Negrin L, Kutscha-Lissberg F, Gartlehner G, J, Südkamp NP, Niemeyer P. In vitro cell Vecsei V. Clinical outcome after microfrac- quality of articular chondrocytes assigned ture of the knee: a meta-analysis of before/ for autologous implantation in dependence after-data of controlled studies. International of specific patient characteristics. Archives of Orthopaedics (October 4, 2011). Orthopaedic and Trauma Surgery 131: 779- 97. Nieminen MT, Töyräs J, Laasanen MS, Sil- 789, 2011. vennoinen J, Helminen HJ, Jurvelin J. Pre- 107. Peterson L, Brittberg M, Kiviranta I, Åker- diction of biomechanical properties of ar- lund EL, Lindahl A. Autologous chondro- ticular cartilage with quantitative magnetic cyte transplantation: Biomechanics and resonance imaging. Journal of Biomechanics long-term durability. The American Journal of 37: 321-328, 2004. Sports Medicine 30: 2-12, 2002. 98. Nieminen MT, Töyräs J, Rieppo J, Ha- 108. Peterson L, Minas T, Brittberg M, Lindahl A. kumäki JM, Silvennoinen J, Helminen HJ, Treatment of osteochondritis dissecans of the Jurvelin JS. Quantitative MR microscopy of knee with autologous chondrocyte transplan- enzymatically degraded articular cartilage. tation: results at two to ten years. The Journal Magnetic Resonance in Medicine 43: 676-681, of Bone and Joint Surgery 85-A Suppl: 17-24, 2000. 2003. 99. Nilsdotter AK, Toksvig-Larsen S, Roos EM. 109. Peterson L, Minas T, Brittberg M, Nilsson Knee arthroplasty: are patients’ expectations A, Sjögren-Jansson E, Lindahl A. Two- to fulfilled? A prospective study of pain and 9-year outcome after autologous chondrocyte function in 102 patients with 5-year follow- transplantation of the knee. Clinical Ortho- up. Acta Orthopaedica 80: 55-61, 2009. paedics and Related Research : 212-234, 2000. 100. Nissi MJ, Töyräs J, Laasanen MS, Rieppo 110. Poole AR, Kojima T, Yasuda T, Mwale F, J, Saarakkala S, Lappalainen R, Jurvelin JS, Kobayashi M, Laverty S. Composition and Nieminen MT. Proteoglycan and collagen structure of articular cartilage: a template for sensitive MRI evaluation of normal and de- tissue repair. Clinical Orthopaedics and Re- generated articular cartilage. Journal of Or- lated Research : S26-33, 2001. thopaedic Research 22: 557-564, 2004. 111. Richardson SM, Hoyland JA, Mobasheri R, 101. Nuernberger S, Cyran N, Albrecht C, Redl Csaki C, Shakibaei M, Mobasheri A. Mesen- H, Vécsei V, Marlovits S. The influence of chymal stem cells in regenerative medicine: scaffold architecture on chondrocyte dis- opportunities and challenges for articular tribution and behavior in matrix-associated cartilage and intervertebral disc tissue engi- chondrocyte transplantation grafts. Biomate- neering. Journal of Cellular Physiology 222: rials 32: 1032-1040, 2011. 23-32, 2010. 102. Nugent GE, Law AW, Wong EG, Temple 112. Rieppo J, Töyräs J, Nieminen MT, Kovanen MM, Bae WC, Chen AC, Kawcak CE, Sah V, Hyttinen MM, Korhonen RK, Jurvelin RL. Site- and exercise-related variation in JS, Helminen HJ. Structure-function rela- structure and function of cartilage from tionships in enzymatically modified articular equine distal metacarpal condyle. Osteoar- cartilage. Cells, Tissues, Organs 175: 121-132, thritis and Cartilage 12: 826-833, 2004. 2003. 103. Oakley SP, Lassere MN, Portek I, Szomor Z, 113. Robert H. Chondral repair of the knee joint Ghosh P, Kirkham BW, Murrell GAC, Wulf using mosaicplasty. Orthopaedics and Trau- S, Appleyard RC. Biomechanical, histologic matology, Surgery and Research 97: 418-429, and macroscopic assessment of articular car- 2011.

38 114. Roberts S, McCall IW, Darby AJ, Menage J, S, Tourville TW, Johnson RJ, Beynnon BD. Evans H, Harrison PE, Richardson JB. Au- Geographic mapping of meniscus and carti- tologous chondrocyte implantation for cartilage lesions associated with anterior cruciate lage repair: monitoring its success by magnet- ligament injuries. The Journal of Bone and ic resonance imaging and histology. Arthritis Joint Surgery 91: 2094-2103, 2009. Research amd Therapy 5: R60-73, 2003. 125. Smith GD, Knutsen G, Richardson JB. A 115. Saarakkala S, Laasanen M., Jurvelin J., Tör- clinical review of cartilage repair techniques. rönen K, Lammi M., Lappalainen R, Töyräs The Journal of Bone and Joint Surgery 87: 445- J. Ultrasound indentation of normal and 449, 2005. spontaneously degenerated bovine articu- 126. Song L, Baksh D, Tuan RS. Mesenchymal lar cartilage. Osteoarthritis and Cartilage 11: stem cell-based cartilage tissue engineering: 697-705, 2003. cells, scaffold and biology. Cytotherapy 6: 116. Safran MR, Seiber K. The Evidence for Sur- 596-601, 2004. gical Repair of Articular Cartilage in the 127. Steadman JR, Briggs KK, Rodrigo JJ, Kocher Knee. Journal of the American Academy of Or- MS, Gill TJ, Rodkey WG. Outcomes of mi- thopaedic Surgeons 18: 259-266, 2010. crofracture for traumatic chondral defects of 117. Samosky JT, Burstein D, Eric Grimson W, the knee: average 11-year follow-up. Arthros- Howe R, Martin S, Gray ML. Spatially-lo- copy 19: 477-484, 2003. calized correlation of dGEMRIC-measured 128. Steadman JR, Miller BS, Karas SG, Schlegel GAG distribution and mechanical stiffness TF, Briggs KK, Hawkins RJ. The microfrac- in the human tibial plateau. Journal of Ortho- ture technique in the treatment of full-thick- paedic Research 23: 93-101, 2005. ness chondral lesions of the knee in National 118. Santoro R, Olivares AL, Brans G, Wirz D, Football League players. The Journal of Knee Longinotti C, Lacroix D, Martin I, Wendt D. Surgery 16: 83-86, 2003. Bioreactor based engineering of large-scale 129. Steadman JR, Rodkey WG, Rodrigo JJ. Mi- human cartilage grafts for joint resurfacing. crofracture: surgical technique and reha- Biomaterials 31: 8946-8952, 2010. bilitation to treat chondral defects. Clinical 119. Saris DBF, Dhert WJA, Verbout AJ. Joint ho- Orthopaedics and Related Research : S362-369, meostasis: the discrepancy between old and 2001. fresh defects in cartilage repair. The Journal of 130. Steadman JR, Rodkey WG, Singleton SB, Bone and Joint Surgery 85: 1067-1076, 2003. Briggs KK. Microfracture technique forfull- 120. Saris DBF, Vanlauwe J, Victor J, Haspl M, thickness chondral defects: Technique and Bohnsack M, Fortems Y, Vandekerckhove B, clinical results. Operative Techniques in Or- Almqvist KF, Claes T, Handelberg F, Lagae thopaedics 7: 300-304, 1997. K, van der Bauwhede J, Vandenneucker H, 131. Stolz M, Gottardi R, Raiteri R, Miot S, Mar- Yang KGA, Jelic M, Verdonk R, Veulemans tin I, Imer R, Staufer U, Raducanu A, Düg- N, Bellemans J, Luyten FP. Characterized gelin M, Baschong W, Daniels AU, Fried- chondrocyte implantation results in better erich NF, Aszodi A, Aebi U. Early detection structural repair when treating symptomatic of aging cartilage and osteoarthritis in mice cartilage defects of the knee in a randomized and patient samples using atomic force mi- controlled trial versus microfracture. The croscopy. Nature Nanotechnology 4: 186–192, American Journal of Sports Medicine 36: 235- 2009. 246, 2008. 132. Swann AC, Seedhom BB. The stiffness of 121. Schinagl RM, Gurskis D, Chen AC, Sah normal articular cartilage and the predomi- RL. Depth-dependent confined compression nant acting stress levels: implications for the modulus of full-thickness bovine articular aetiology of osteoarthrosis. British Journal of cartilage. Journal of Orthopaedic Research 15: Rheumatology 32: 16-25, 1993. 499-506, 1997. 133. Tanabe Y, Yasuda K, Azuma C, Taniguro 122. Schonholtz GJ, Ling B. Arthroscopic chon- H, Onodera S, Suzuki A, Chen YM, Gong droplasty of the patella. Arthroscopy 1: 92-96, JP, Osada Y. Biological responses of novel 1985. high-toughness double network hydrogels in 123. Schulz RM, Bader A. Cartilage tissue engi- muscle and the subcutaneous tissues. Journal neering and bioreactor systems for the culti- of Materials Science. Materials in Medicine 19: vation and stimulation of chondrocytes. Eu- 1379-1387, 2008. ropean Biophysics Journal 36: 539-568, 2007. 134. Thambyah A, Nather A, Goh J. Mechanical 124. Slauterbeck JR, Kousa P, Clifton BC, Naud properties of articular cartilage covered by

39 the meniscus. Osteoarthritis and Cartilage 14: pressure increases apoptosis in cartilage- 580-588, 2006. constructs produced from human osteoar- 135. Tognana E, Chen F, Padera RF, Leddy H a, thritic chondrocytes. Frontiers in Bioscience Christensen SE, Guilak F, Vunjak-Novakov- 11: 1690-1695, 2006. ic G, Freed LE. Adjacent tissues (cartilage, 141. Widuchowski W, Widuchowski J, Trzaska T. bone) affect the functional integration of en- Articular cartilage defects: study of 25,124 gineered calf cartilage in vitro. Osteoarthritis knee arthroscopies. The Knee 14: 177-182, and Cartilage 13: 129-138, 2005. 2007. 136. Vanlauwe J, Saris DB, Victor J, Almqvist KF, 142. Williams RJ, Harnly HW. Microfracture: Bellemans J, Luyten FP. Five-year outcome indications, technique, and results. Instruc- of characterized chondrocyte implantation tional Course Lectures 56: 419-428, 2007. versus microfracture for symptomatic carti- 143. Williams RJ, Ranawat AS, Potter HG, Carter lage defects of the knee: early treatment mat- T, Warren RF. Fresh stored allografts for the ters. The American Journal of Sports Medicine treatment of osteochondral defects of the 39(12): 2566-2574, 2011. knee. The Journal of Bone and Joint Surgery 137. Verzijl N, Degroot J, Zaken CB, Braun- 89: 718-726, 2007. benjamin O, Maroudas A, Bank RA, Miz- 144. Wilson W, van Donkelaar CC, van Rietber- rahi J, Schalkwijk CG, Thorpe SR, Baynes gen R, Huiskes R. The role of computational JW, Bijlsma JWJ, Lafeber FPJG, Tekoppele models in the search for the mechanical be- JM. Crosslinking by advanced glycation end havior and damage mechanisms of articular products increases the stiffness of the col- cartilage. Medical Engineering and Physics 27: lagen network in human articular cartilage: 810-826, 2005. a possible mechanism through which age is 145. Yasuda K, Kitamura N, Gong JP, Arakaki K, a risk factor for osteoarthritis. Arthritis and Kwon HJ, Onodera S, Chen YM, Kurokawa Rheumatism 46: 114–123, 2002. T, Kanaya F, Ohmiya Y, Osada Y. A novel 138. Volkov M. Allotransplantation of joints. double-network hydrogel induces spontane- Journal of Bone and Joint Surgery 52: 49, 1970. ous articular cartilage regeneration in vivo in 139. Wayne JS, Kraft KA, Shields KJ, Yin C, Owen a large osteochondral defect. Macromolecular JR, Disler DG. MR imaging of normal and Bioscience 9: 307-316, 2009. matrix-depleted cartilage: correlation with 146. Yasuda K, Ping Gong J, Katsuyama Y, Na- biomechanical function and biochemical kayama A, Tanabe Y, Kondo E, Ueno M, composition. Radiology 228: 493-499, 2003. Osada Y. Biomechanical properties of high- 140. Wenger R, Hans MG, Welter JF, Solchaga toughness double network hydrogels. Bioma- LA, Sheu YR, Malemud CJ. Hydrostatic terials 26: 4468-4475, 2005.

2 Experimental verification of a non-linear model for computing cartilage modulus from microindentation data

Neither linear-, visco- nor poro-elasticity is a correct model for cartilage stiffness behaviour. Therefore, a new, simplified mathematical model is proposed for determining from indentation data the stiffness of hyaline cartilage and related materials with non-linear elastic behaviour. In the literature a traditional mathematical model proposed by Hayes et al. [4]--which assumes linear elasticity--is in general use for calculating cartilage stiffness, although cartilage is not linearly elastic. The Hayes approach also requires calculation of a factor k which is a complex function of indentation depth, cartilage thickness and Poisson’s ratio. The newer Kren [7] model does not require either the linear elasticity assumption or determination of k. Sinusoi- dal indentation data were obtained at 0.1 Hz from osteochondral plugs of fresh swine cartilage from 8 knees. Using the same assumed Poisson’s ratio (0.45) in both cases, the calculated moduli at indentation depths of 0.05 and 0.1 mm were respectively 4.2 ± 1.1 MPa and 4.9 ± 1.3 MPa (Hayes) and 7.0 ± 2.1 and 8.7 ± 3.0 MPa (Kren). Linear regression analysis between the two models showed a correlation of 0.97 (99% confidence interval). For the Kren model in contrast to other models, it is not necessary to assume linear-, visco- or porovisco-elasticity to calculate a modulus from indentation data. This is a strength An adapted version of this chapter is submit- since cartilage is not a material but instead an ted as: D. Wirz, S. Ronken, A.P. Kren and A.U. inhomogeneous, non-isotropic structure. The Daniels. Experimental verification of a non- Kren model is perhaps best used for inden- linear model for computing cartilage modulus tations <10% of cartilage thickness to avoid from microindentation data. Computational and effects on modulus of the stiffness of underly- Mathematical Methods in Medicine ing bone.

INTRODUCTION coordinate. One of the major drawbacks of the Hayes In recent years, a lot of effort is put in finding model is the fact that linear elasticity is assumed. an articular cartilage repair material, e.g. tissue At the beginning of the discussion, Hayes warns engineered constructs. To be successful, such that: “Articular cartilage is a viscoelastic mate- engineered tissues must approach not only the rial and any dynamic analysis must treat it as morphology but also the mechanical properties such.”[4]. of healthy natural cartilage, in order to bear the heavy loads occurring in the human body. The primary mechanical measurements have been THEORETICAL ANALYSIS AND MODEL- non-destructive assessments of stiffness since this ING determines how cartilage bears and distributes loads which are functional (i.e. non-traumatic) We therefore propose a simplified, alternative when applied to healthy tissue. Methods for approach for calculating stiffness parameters of measuring the stiffness of cartilage in confined linear elastic and viscoelastic materials from in- or unconfined compression have been widely dentation data gathered with a spherical indenter. used but are disadvantageous for several reasons One first assumes that the force acting against (i) it is very demanding to prepare geometrically the (spherical) indenter may be separated into an

correct specimens (e.g. cylinders), a condition elastic component Pe and a viscous component

that has to be met in order to get proper mea- Pv (Equation 2.2) [7]: surement results, (ii) the cartilage layer is usually PPP= + (2.2) only 1-3 mm thick which makes it difficult to e v measure the stiffness for small strains, (iii) initial It is then necessary to assume that the elastic specimen geometry is lost during unconfined component of contact force obeys the Hertz law compression, due to lateral bulging of the speci- [5] (Equation 2.3): mens, (iv) confined compression requires use of 4E* = a 3 artificial porous confinements and (v) prepara- Pe 2 r (2.3) 3(1−n ) tion of geometric samples destroys the specimen’s original osteochondral environment. In addition, the viscous component (Pv) h Another approach to determine the stiff- depends directly on viscosity , and the velocity ness of cartilage is to measure with indenters. No of intrusion V (Equation 2.4):

preparation of the cartilage is needed and stiff- Pv = g()a V (2.4) ness mapping of a cartilage surface is possible on where g(a) is an arbitrary positive function and a smaller scale than with compression specimens. V the current velocity of the indenter. Also, a comparison by Korhonen et al. [6] sug- Obviously P is equal to zero when P is gests that indentation provides more function- v equal to zero and/or V is equal to zero. For in- ally meaningful results than either confined or dentation measurements P is equal to zero at the unconfined compression. Further, to allow in moment of indentation--i.e. where a = 0 and V vivo measurements e.g. during a knee arthros- is equal to zero at the end of the active stage of copy, indentation is the only possibility. indentation when a = a . Consequently the to- Hayes et al. [4] proposed a mathemati- max tal force at a is as follows (Equation 2.5). cal model allowing stiffness calculations (e.g. E- max * modulus) from indentation measurements using 4E 3 P = ra (2.5) flat or spherical indenters. Equation 2.1 shows amax 3(1−n 2 ) max the Hayes’ solution for spherical indenters: For a given spherical indenter of radius r, if Pamax 1 and a are measured and n is known or as- 2 P max E =(1 −n ) a() td t (2.1) sumed, then by rearrangement of (Equation 2.5) 2ka a (2r − a ) ∫ 0 the dynamic modulus can be determined: With E = E-Modulus, n = Poisson’s ratio, r = 3P indenter radius, a = indentation depth. k is a E* =(1 −n 2 ) amax 3 (2.6) nondimensional parameter depending on carti- 4 ramax lage thickness, a and r. t is a normalized radial

45 MATERIALS AND METHODS Table 2.1: Moduli determined using Hayes’ and Kren’s model Eight pig knees were obtained at a local butcher Indentation E Hayes E* Kren shop. All pigs were of similar size and age and depth [mm] [MPa] [MPa] were freshly slaughtered on the day of testing. Af- ter preparation of the knee joints, cylindrical os- 0.05 4.2 ± 1.1 7.0 ± 2.1 teochondral plugs from the lateral condyles were obtained with a diamond core drill with a diam- 0.1 4.9 ± 1.3 8.7 ± 3.0 eter of 7.45 mm (Synthes, Oberdorf, Switzerland). Cartilage thickness was measured to the nearest 0.1 mm using a calliper. The indentation tests DISCUSSION were performed with a mechanical test machine (Synergie 100, MTS Systems, Eden Prairie, MN) In 1972 Hayes et al. [4] proposed a mathemati- equipped with a 2.5 N loadcell (Burster 8432, cal model to measure stiffness of cartilage in Burster Praezisionsmesstechnik GmbH & Co, indentation mode. In this model cartilage is Gernsbach, Germany) and a spherical indenter assumed to be linearly elastic, although Hayes with a radius of 1.585 mm. Force and displace- acknowledges that this is not actually the case. ment data were measured under sinusoidal dis- The Hayes approach also requires calculation of placement control at 0.1 Hz until a maximum a factor k which is a complex function of inden- indentation depth of either 0.05 or 0.1 mm. In- tation depth, cartilage thickness and poisson’s dentation depth was kept below 10% of cartilage ratio. The model has been widely and is still thickness. Cartilage modulus was calculated us- used to calculate stiffness from indentation data ing the method described above as well as the at various strain rates and for stress relaxation method of Hayes et al. [4]. A Poisson’s ratio of measurements [3]. This was done even though 0.45 was assumed for both methods. It needs to Hayes et al. warned that this model is only valid be pointed out that values different Poisson’s ra- for predicting the instantaneous response to step tio would not change the relative values of the loads and predicting asymptotic deformation. moduli obtained by the two methods. Statistical Much later, different models where proposed in analysis (Wilcoxon rank sum test and linear re- order to elaborate a constitutive model of car- gression analysis) was made with R (R Founda- tilage, including a poroelastic model reinforced tion for Statistical Computing, Austria). with a network of elastic fibres [2]. But it should be noted that the actual fibres in cartilage, i.e. the collagen fibres, are also not linearly elastic. RESULTS Because neither linear-, visco- nor poro-elasticity is a correct model for cartilage we The moduli from swine hyaline cartilage at two propose a new method taking non-linear elastic different indentation depths (0.05 and 0.1 mm) behaviour of cartilage into account, based on the calculated using Hayes’ and Kren’s models are work of Kren [7]. With this approach only Pois- shown in Table 2.1. The paired Wilcoxon rank son’s ratio has to be assumed. When moduli are sum test showed that there is a significant differ- calculated from the same data using the Kren ence (p<0.01) between the values of the modulus approach and the traditional Hayes approach, measured at the two different indentation depths there is a significant difference between the using either the Hayes or Kren model. However, modulus values. However, both models showed the moduli at a given depth determined by the similar behaviour. two models differ significantly. The modulus First, both models showed a higher determined by the Hayes model is ~40% smaller modulus at an indentation depth of 0.1 mm at both depths, but it is a proportional differ- compared to 0.05 mm. The deformation rate is ence. That is, linear regression analysis shows also higher at an indentation depth of 0.1 mm, a correlation coefficient of 0.97 between Hayes’ since the frequency used is the same. This higher and Kren’s modulus values that is statistically modulus is expected since viscoelastic materials significant (99% confidence interval). and cartilage both become stiffer with increasing deformation rate [8]. In addition, it has been

46 shown elswhere that articular cartilage stiffness ensures that the indentation force/displacement increases with depth. Cartilage closer to the sur- response is not influenced by the presence of a face is less stiff compared to that close to the sub- much stiffer underlying support (e.g. subchon- chondral bone [9]. dral bone). Second, the results of the two models correlate perfectly. When a high modulus is calculated using Hayes’ model, Kren’s model also CONCLUSION gives a high modulus and vice versa. This suggests that changes in moduli due to differences The results suggest that a much simplified ap- in e.g. specimens or deformation rate calculated proach based on the work of Kren can be used for with one model will have similar effects on mod- calculating dynamic stiffness of cartilage from uli calculated using the other model. Clearly indentation data--providing that indentation however, absolute values of moduli determined depth does not exceed 10% of specimen thick- using one model should not directly compared ness. Another strength of the new model is that with values obtained using the other model to in contrast to other models, it is not necessary to determine effects of variables. assume linear-, visco- or porovisco-elasticity to There is a possible explanation for why calculate a modulus from indentation data. the Kren model yields higher modulus values Finally, as show through the results re- at both depths than the Hayes model. As noted ported here, older data determined using the above, cartilage stiffness increases with depth; Hayes model can still be considered highly ac- cartilage closer to the surface is less stiff com- curate and useful in evaluating the response to pared to that close to the subchondral bone [9]. variables such as indentation depth. However, Kren’s model only uses the force and deforma- when evaluating the effects of controlled vari- tion values obtained at maximum indentation ables the absolute values of moduli based on dif- depth. ferent models should not be compared. A limitation of Kren’s model is the fact that the thickness of the cartilage is not taken into account. This is in contrast to Hayes’ model, ACKNOWLEDGEMENTS which includes a factore k which depends on cartilage thickness, indentation depth and ra- The study was partially supported by the dius of the indenter. As a result, when using the Deutsche Arthrose-Hilfe e.V., Saarlouis, Ger- new model, it seems appropriate that the inden- many, the Hardy and Otto Frey-Zünd Stiftung, tation depth should not exceed 10% of the car- Basel, Switzerland and the funds donated to the tilage thickness. According to Bueckle [1], this University of Basel by H.J. Wyss.

47 REFERENCES 6. Korhonen RK, Laasanen MS, Töyräs J, Ri- eppo J, Hirvonen J, Helminen HJ, Jurvelin 1. Bückle H. The science of hardness testing J. Comparison of the equilibrium response and its research applications. American Soci- of articular cartilage in unconfined compres- ety for Metals, Metals Park, OH, 1973. sion, confined compression and indentation. 2. Carter DR, Wong M. Modelling cartilage Journal of Biomechanics 35: 903-909, 2002. mechanobiology. Philosophical Transactions 7. Kren AP, Rudnitskii VA, Deikun IG. Deter- of the Royal Society of London. Series B 358: mining the viscoelastic parameters of vulca- 1461-1471, 2003. nisates by the dynamic indentation method 3. Hall ML, Krawczak DA, Simha NK, Lewis using a non-linear deformation model. Inter- JL. Effect of dermatan sulfate on the indenta- national Polymer Science and Technology 32: tion and tensile properties of articular carti- 19-23, 2005. lage. Osteoarthritis and Cartilage 17: 655-661, 8. Park S, Hung C, Ateshian G. Mechanical 2009. response of bovine articular cartilage under 4. Hayes W, Keer L, Herrmann G, Mockros L. dynamic unconfined compression loading at A mathematical analysis for indentation tests physiological stress levels. Osteoarthritis and of articular cartilage. Journal of Biomechanics Cartilage 12: 65-73, 2004. 5: 541-551, 1972. 9. Schinagl RM, Gurskis D, Chen AC, Sah 5. Hertz H. Ueber die Berührung fester elast- RL. Depth-dependent confined compression ischer Körper. Journal fur die reine und ange- modulus of full-thickness bovine articular wandte Mathematik 92, 1882. cartilage. Journal of Orthopaedic Research 15: 499-506, 1997.

3 A comparison of healthy human and swine articular cartilage dynamic indentation mechanics

Articular cartilage is a multicomponent, poroviscoelastic tissue with nonlinear mechanical properties vital to its function. A consequent goal of repair or replacement of injured cartilage is to achieve mechanical properties in the repair tissue similar to healthy native cartilage. Since fresh healthy human articular cartilage (HC) is not readily available, we tested whether swine cartilage (SC) could serve as a suitable substitute for mechanical comparisons. To a first approximation, cartilage tissue and surgical substitutes can be evaluated mechanically as viscoelastic materials. Stiffness measurements (dynamic modulus, loss angle) are vital to function and are also a non-destructive means of evaluation. Since viscoelastic material stiffness is strongly strain rate dependent, stiffness was tested under different loading conditions related to function. Stiffness of healthy HC and SC specimens was determined and compared using two nondestructive, mm-scale indentation test modes: fast impact and slow sinusoidal deformation. Deformation resistance (dynamic modulus) and energy handling (loss angle) were determined. For equivalent anatomic locations, there was no difference in dynamic modulus. However, the HC loss angle was ~35% lower in fast impact and ~12% higher in slow sinusoidal mode. Dif- ferences seem attributable to age (young SC, older HC) but also to species anatomy and An adapted version of this chapter has been pub- biology. Test mode-related differences in hu- lished as: S. Ronken, M.P. Arnold, H. Ardura man-swine loss angle support use of multiple García, A. Jeger, A.U. Daniels and D. Wirz. A function-related test modes. Keeping loss comparison of healthy human and swine artic- angle differences in mind, swine specimens ular cartilage dynamic indentation mechanics. could serve as a standard of comparison for Biomechanics and Modeling in Mechanobiology, mechanical evaluation of e.g. engineered car- 2011, DOI 10.1007/s10237-011-0338-7 tilage or synthetic repair materials.

INTRODUCTION likely to become focused in a small region and can cause pain or tissue damage. The second re- Joint surface hyaline cartilage is comprised sponse is energy handling, or the extent to which mostly of extracellular material that is produced energy imparted by deformation is either stored by a small number of chondrocytes. These cells or dissipated. If cartilage stores energy it rapidly have no direct blood supply and receive nutrition springs back to shape when a load is removed. from synovial fluid and the subchondral bone If it instead dissipates some of the deformation plate [28]. Cartilage must withstand millions energy (as heat) it returns to shape more slowly, of dynamic loads each year that are often mul- and the peak loads in the cartilage itself and in tiples of body weight, and it protects the sensi- underlying bone are reduced. This is a means of tive, nerve-filled underlying bone from receiving damage protection. loads which are concentrated enough to result Cartilage is a complex “material” because in pain and/or damage. In the majority of indi- it behaves in some ways like a sponge (porous), viduals, joint cartilage is able to do this for de- like a spring (elastic) and like a liquid (viscous). cades after skeletal maturity without undergoing Thus cartilage can be described as a “porovisco- appreciable damage or wear itself. elastic solid”. The stiffness and energy handling One consequence of the above is that no of such complex materials depends on the rate synthetic implantable materials or structures are at which they are deformed. As a result there are yet available which come close matching the me- special definitions for stiffness--E* = dynamic chanical properties and durability of joint carti- modulus, and energy handling--d = loss angle, lage. As a result, surgical repair of the joint sur- methods for measuring them, and reasons for faces is reliant on either (a) joint replacement with doing so under different conditions which mim- metal, ceramic and polymer components, (b) ic cartilage dynamic mechanical function. The autograft, allograft or xenograft transplantation, correlation between the storage modulus (E’), (c) medical and surgical treatments designed to the loss modulus (E”), the dynamic modulus rejuvenate cartilage or (d) surgical resurfacing of (E*) and the loss angle (d) for a viscoelastic ma- portions of the joint surfaces with engineered tis- terial is shown in Figure 1.6. This shows that at sue [6, 9, 12, 17, 30]. Assuming that repair prod- least two of those parameters are needed in order ucts for damaged cartilage should mimic the to describe a viscoelastic material properly. In biomechanical properties of healthy human car- previous studies, the focus was mainly on deter- tilage, the best relative measure of their success is mining the stiffness [10, 15, 22, 24, 32, 33], and to determine to what extent the resultant struc- only a few studies also determined which part of tures have mechanical properties which resemble the energy was dissipated [1, 27, 29]. Determin- those of healthy human joint cartilage. But this ing the mechanical properties of cartilage is typi- poses another problem--such human tissue is not cally done in confined compression [13, 16, 32], readily available for ex-vivo use as a standard of unconfined compression [8, 13, 15, 16, 29] or comparison--i.e. to be subjected to the same me- indentation [1, 14, 16, 22, 23, 26, 27]. Most of chanical tests as candidate synthetic structures, these studies determine the stiffness of the carti- grafts, rejuvenated cartilage or engineered cartilage in slow compression, e.g. stress relaxation [8, lage. In contrast, healthy animal joint cartilage 13, 16, 25, 32] or creep [4, 14, 15, 23]. However, is readily available for such purposes. In order to besides undergoing slow quasi-cyclic deforma- draw conclusions from animal models, one then tions (e.g. while someone stands in place), carti- has to ascertain the extent to which such animal lage also functions in sudden transient deforma- cartilage has the same mechanical properties as tions which occur during gait. This is important, human cartilage. because as mentioned above dynamic stiffness The response to non-damaging dynamic parameters of poroviscoelastic materials like car- compressive loads has two aspects. The first is tilage are extremely dependent on deformation stiffness, or the amount of deformation in re- rate. sponse to load. If cartilage stiffness is too low it The aims of this study were to (a) fur- can become so thin under load that the load is ther develop the authors’ methods of measuring effectively transmitted directly to the underlying E* and d of cartilage in both slow quasi-cyclic bone. If cartilage stiffness is too high, a load is deformation and fast impact indentation, and

53 (b) to determine and compare results for healthy swine specimens and healthy human cartilage. The overall goal was to establish whether swine specimens can serve as a standard of comparison for evaluating the extent to which joint cartilage replacement and repair strategies achieve cartilage-like mechanical behaviour.

Figure 3.1: MATERIALS AND METHODS Example of a swine (left) and a human (right) piece of cartilage. White bar is 1 cm. Materials cartilage, are extremely strain rate dependent. Swine cartilage Besides that, cartilage functions in two different Swine cartilage (SC) was obtained from the knee loading regimes -- the sudden transient defor- of 10 nine months old swine. Cylindrical osteo- mations which occur during gait, and the slow chondral plugs of 7.6 mm in diameter (Figure quasi-cyclic deformations which cause fluid to 3.1; left) were harvested using a standard dia- move in and out of cartilage and thus provide mond core-drill designed for mosaicplasty (Syn- a means for nutrition. Thus both a Fast Impact thes, Oberdorf, Switzerland). Plugs were harvest- Mode and a Slow Sinusoidal Mode test method ed from the lateral patella (LP), the medial and were developed. The dynamic modulus was cal- lateral patellar groove (MPG & LPG) and from culated as described by Wirz [35] and Kren [18]. the medial and lateral condyles (MC & LC). The The loss angle was calculated directly from the samples were stored in phosphate buffered saline phase shift between the displacement and load (PBS) and frozen at -24°C. Prior to testing, the curves. samples were thawed until room temperature and kept wet with PBS during testing. Fast Impact (FI) mode To simulate the impact velocity in gait a fast Human cartilage impact micro-indentation instrument was used. Human articular knee cartilage (HC) tissues This is a modified version of an instrument de- were collected from full-thickness biopsies of the veloped at the Minsk Institute of Physics [18]. A lateral femoral condyle of 8 fresh human cadav- pendulum-mounted spherical indenter (diam- ers (2 male and 6 female, median age: 57 years, eter: 1.0 mm; 1.7 g) falls down on the specimen range: 42-69) at the Department of pathology of under gravitational force. The motion of the in- the local University Hospital following informed denter is captured electromagnetically during consent by relatives and in accordance with the indentation and rebound. On each specimen 10 requirements of the Local Ethical Committee. replicate measurements were performed on the Tissue was only harvested from knees without same spot at ~20 second time intervals. Resul- macroscopic signs of degenerative arthritis (Fig- tant E* and d were calculated for each impact ure 3.1; right). The samples were stored in PBS and then each set was averaged to get one set of and frozen at -24°C for later use. Prior to testing, specimen values. the samples were thawed in PBS at room temperature and kept wet during testing. Slow Sinusoidal (SS) mode To simulate nutrition in cartilage specimens a Methods Synergie 100 MTS® mechanical testing instrument was used to perform slow sinusoidal micro- Two micro-indentation methods were used as indentations. A spherical indenter (diameter ~3.2 previously described [2, 3, 35] to determine the mm) was moved sinusoidally under computer dynamic stiffness parameters (dynamic modulus software control. The frequency was 0.1 Hz and E* and loss angle d) of cartilage, meniscus and the indentation depth was ~0.05 mm. The maxi- possible implant materials. The dynamic stiff- mum speed was ~0.015 m/s. The same specimens ness parameters of poroviscoelastic materials, i.e. were measured as in FI-mode. On each specimen

54 5 replicate measurements were performed at in- coelastic behaviour of cartilage at high strain tervals of ~2 minutes on the same spot. Resultant rates. E* and d were averaged to get one set of speci- In SS-mode the applied displacement men values. was sinusoidal as specified by the software driv- ing displacement. There was again a difference Statistical analysis in the time at which force and displacement A one-sided Wilcoxon Rank Sum test was per- reached a maximum, again attributable to visco- formed on both the E* and loss angle d data sets elasticity. In addition, however, for both HC and (p < 0.05). To quantify the spread in the data, the SC specimens the resultant force-time curves median absolute deviation was calculated and were not sinusoidal. divided by the median to normalize and thus al- The E* andd calculated from the raw SC low comparisons of the spreads within the data. data are shown in Figure 3.3. At any anatomic Statistical analysis was accomplished using R (R location E* was significantly higher and d was Foundation for Statistical Computing, Austria). significantly lower in FI-mode compared to SS- mode. In cartilage and also most synthetic viscoelastic materials faster deformation rates result in RESULTS higher E* and lower d. Also, some significant differences in these parameters were found among In FI-mode, for both HC and SC specimens, the various locations on the swine knee for both both force and displacement vs. time traced a testing modes as follows. portion of an essentially sinusoidal curve but The E* in FI-mode on the LC and the with peak force and peak displacement not oc- MC was lower compared to the LP, the LPG and curring at the same time (Figure 3.2). This phase the MPG. In SS-mode fewer differences were difference was attributable to the essentially vis- found. The E* on the LC was lower than on the

Figure 3.2: Typical example of the force and displacement curves of a swine (line) and human (dotted line) sample in fast impact (left) and slow sinusoidal (right) mode.

55 LPG and the MPG and also lower on the MC specimens the normalized spread was between compared to the MPG. 5% and 19%. The overall normalized variation The d of the LC was lower compared to within the SC samples was 16% in FI-mode and the LP, LPG and MPG and lower on the MC 7% in SS-mode. The normalized spread ind was than on the LP in FI-mode. In SS-mode there lower in SS-mode compared to FI-mode. were no differences in d among the various locations. For both test modes there was no statisti- DISCUSSION cally significant difference in E* between human and swine LC (Figure 3.4). However, the d was Effect of specimen dimensions significantly higher in FI-mode and lower in SS- mode for the HC compared to the SC. To ensure that tissue specimen thickness would The median absolute deviation was cal- not influence results, indentation in both modes culated and divided by the median to normalize was limited to less than ~10% of tissue thick- and thus allow comparisons of the spreads within ness [5]. Further, Niederauer et al. [24] showed the data (Table 3.1). The normalized spread in E* that cartilage stiffness is not correlated with car- of the HC specimens was 26% for both FI- and tilage thickness. We also used a small indenter SS-mode. For the SC the normalized spread was diameter compared to the diameter of the osteo- between 10% and 52% depending on location chondral plugs, in order to minimize specimen and test mode. The overall normalized variabil- dimension effects in directions parallel to the ity within the SC samples was 36% for FI-mode specimen articular surface. and 34% for SS-mode. The normalized spread was lower for the d compared to the E*. For the Effects of test mode on force-time and dis- HC specimens the normalized spread was 21% placement-time data in FI-mode and 8% in SS-mode. In the SC

Figure 3.3: Box-and-Whisker plot of the dynamic modulus (up) and the loss angle (bottom) of the swine specimens at fast impact (left) and slow sinusoidal (right) mode. LC= lateral condyle; MC= medial condyle; LP= lateral patella; LPG= lateral patellar groove; MPG= medial patellar groove. Horizontal lines indicate a significant difference with p < 0.05.

56 is low, and at 0.1 Hz there is time for water to move through the structure in response to load in spite of the low permeability. The response of cartilage is then poroviscoelastic. In SS-mode --and thus under poroviscoelastic response conditions and at smaller displacements than in FI- mode-- discontinuities were observed in both the displacement-time and force-time curves. The discontinuities in the displacement-time curves were due to measurement artefacts. These were uniform throughout the measurements. The measurement system was unable to resolve the small changes in displacement with time occurring near the peak of the applied sinusoidal wave form. In contrast the non-sinusoidal variations in the force-time curves was seen only with cartilage specimens and not seen with essentially elastic synthetic materials (data not shown), indicating that this was not an artefact. For cartilage, near the peak of displacement, the change in the indentation depth with time was extreme- Figure 3.4: Box-and-Whisker plot of the dynamic modu- ly small. Therefore, the non-sinusoidal decline in lus (up) and the loss angle (bottom) of the human and force during this time can likely be attributed to swine lateral condyle (LC) specimens at fast impact (FI) energy loss of the cartilage. and slow sinusoidal (SS) mode. Horizontal lines indicate a significant difference with p < 0.05 Effects of cartilage test mode and anatomic location on E* and d As expected, since cartilage is partly a viscoelastic material, the maxima in force-time and As expected E* was higher and d was lower at all displacement-time data (Figure 3.2) occurred swine anatomic locations in FI-mode compared at different times in both FI- and SS-modes. to SS-mode (Figure 3.3). This is the normal re- This phase shift is due to energy losses result- sponse of materials exhibiting some viscoelastic ing from internal friction. In FI-mode the high behaviour when tested at a high strain rate (FI) strain rate coupled with the low permeability of vs. a low strain rate (SS). water through cartilage resulted in essentially There were also marked differences in viscoelastic rather than more complex porovis- E* as a function of anatomic location in both coelastic behaviour, and thus smooth changes FI- and SS-modes. However, there were more in both force and displacement with time were significant differences in E* as a function of ana- observed. In contrast in SS-mode, the strain rate tomic location for FI-mode than for SS-mode,

Table 3.1: Median absolute deviation divided by the median from the human and swine specimens on the various locations. LC= lateral condyle; MC= medial condyle; LP= lateral patella; LPG= lateral patellar groove; MPG= medial patellar groove Parameter Mode Human Swine LC LC MC LP LPG MPG All E* FI 26% 14% 21% 52% 16% 10% 36% SS 26% 24% 31% 43% 33% 14% 34% d FI 21% 11% 19% 18% 8% 19% 16% SS 8% 6% 12% 6% 5% 6% 7%

57 and only FI-mode detected any significant dif- in the superficial and upper mid zone [11]. Due ferences in d as a function of anatomic location. to the smaller variation in d and the changes oc- Thus FI-mode seems to provide a more sensitive curring in osteoarthritic cartilage, osteoarthritis measure of cartilage stiffness parameters. This might be easier to detect by looking at the d in- is somewhat surprising since as described above, stead of the E*. the high strain rate and short duration of the FI-mode limits the motion of water through Differences in E* and d between human and cartilage. As a result, one might expect that the swine LC specimens FI-method would be less sensitive to changes in the proteoglycan portion of the cartilage struc- The E* of the HC and SC were not signifi- ture and thus less sensitive overall. The fact that cantly different in either FI- or SS-mode. This this was not the case points up the importance of was encouraging as a purpose of this study was evaluating cartilage stiffness parameters at sub- to evaluate whether healthy SC can serve as a stantially different strain rates that are related to stand-in for healthy HC in stiffness parameter the spectrum of strain rates that occur in vivo. tests. However the trends were that SC was stiff- Also, as discussed later, there are other phenom- er than HC in FI-mode but less stiff in SS-mode. ena which are only revealed by using more than In contrast to the E* results, the d of SC was sig- one strain rate. nificantly (~35%) lower in FI-mode and (~12%) It is already known that the stiffness of higher in SS-mode compared to HC. Overall, cartilage varies at different locations in human the swine vs. human differences were not enor- [22] and sheep [1] knees. Lyyra et al. [22] found mous, and that is encouraging. However, the the highest stiffness in the load bearing areas of results do raise two questions. First, why were the condyles, where we found the highest E* in the results were somewhat different for HC and the lateral patellar groove and the lowest E* on SC, and second why were the swine vs. human the lateral and the medial condyles. This differ- relationships opposite for FI mode compared to ence might be due to a different loading pattern SS-mode? in the swine knee compared to the human knee. The relatively small swine vs. human dif- To quantify the normalized spread ferences in a given test (FI or SS) seem likely to within the different specimens, the median ab- be due to a difference in composition of the carti- solute deviation was divided by the median. The lage. Composition differences could be expected results showed a large spread between the vari- for two reasons. First, the swine specimens were ous samples (Table 3.1). The normalized spread from young, healthy animals (age ~10 months) in E* between the HC samples was 26%. Lyyra while the human specimens--although visually et al [22] determined the coefficient of variation judged to be healthy--were from humans far past which was 29% between individuals. The E* of skeletal maturity (age 42-69 years). In addition the HC specimens did not correlate with age the swine data for various anatomic locations in this study. The spread might be due to the show considerable variation, no doubt reflect- cartilage being exposed to another load pattern, ing load-related needs. In this light, it may be e.g. sports, body weight. Another reason could that differences in the stance and gait of swine vs. be differences in cartilage thickness, although humans result in different load-related needs on thickness and stiffness were not shown to be cor- the LC and thus differences in structure and me- related in the study of Niederauer et al [24]. chanical properties between swine and human The normalized spread between subjects LC cartilage. and between the different locations was smaller The second question remains: why were in d than in E* (Table 3.1). The spread was the the swine vs. human relationships for LC car- smallest in the d in the SS-mode. In the early tilage in both E* (trend) and d (significant) op- stage of osteoarthritis more proteoglycans are posite for FI-mode compared to SS-mode? These produced by the chondrocytes [7], which chang- opposite results suggest the following. es the capability of the cartilage to hold the wa- In FI-mode there is little or no movement ter. Besides the increase in proteoglycans during of water and thus the E* values may mostly re- osteoarthritis, cartilage collagen declines [20, flect mechanics of cross-linked collagen. Cross- 31] and damage occurs in the existing collagen linking of collagen or other polymers (e.g. latex)

58 generally causes them to behave more like rub- lus, the loss angle is an important dynamic stiff- ber--less viscously and more elastically (higher ness parameter of cartilage and should therefore E* lower d). Thus in FI-mode, the lower E* and be a part of every test set that is meant to define higher d for HC compared to SC might have the mechanical quality of cartilage or suitability been attributable to a relatively lower amount of of a cartilage repair material. cross linking of collagen in the (older) HC than in the (younger) SC. However, it is reported that collagen cross linking is likely to increase with CONCLUSIONS age [21]. Also it has been shown that the collagen fibre arrangement is similar in HC and SC Swine cartilage dynamic modulus and loss angle [19]. Thus the difference found is perhaps attrib- varied significantly with anatomic location with- utable to a species difference in the amount of in the swine knee. More significant differences collagen cross linking. were seen in FI (fast impact) mode than in SS Conversely, in SS-mode HC exhibited a (slow sinusoidal) mode. In fact, in SS-mode no higher E* (trend) and lower d (significant) com- significant differences were seen in loss angle as a pared to SC. As described above, a great part of function of anatomic location. This was surpris- the cartilage mechanical response during slow ing since one might expect that structure-related deformation (0.1 Hz in this case) is due to the differences in cartilage permeability might be nature of the motion of water in the proteogly- an expected difference as function of anatomic can portion of the cartilage structure. A rela- location. It seems that such a difference would tively lower amount of water in the proteogly- be more evident during SS deformation since cans could be expected to result in less viscous movement of water is possible in this mode. No behaviour--i.e. higher E* and lower d--as seen ready explanation was found for this apparently here for HC (older) specimens vs. SC (younger) anomalous result. cartilage specimens. Indeed, a decline in proteo- Differences in human-swine loss angle glycan water content of cartilage with age has trends for FI (fast impact) vs. SS (slow sinusoidal) been reported [34]. test modes support the need for using multiple The above-described contrasting, struc- function-related test modes to more completely turally-explainable results for E* and d for the understand cartilage mechanical behaviour. FI vs. the SS test modes supports the idea that Finally, keeping loss angle differences in cartilage mechanics should be evaluated over a mind, swine specimens could serve as a standard range of representative load deformation condi- of comparison for mechanical evaluation of e.g. tions. Otherwise the picture is incomplete and surgically repaired cartilage, engineered carti- misleading conclusions could be drawn. Results lage or synthetic repair materials. here are for two extremes. They suggest that even more information related to age and disease induced variations in cartilage structure might ACKNOWLEDGMENTS be revealed by a more thorough exploration of the effects of rate and amplitude of deformation. H.-J. Wyss for the funds donated to University The results also point up that besides the modu- Basel; Hardy & Otto Frey-Zünd Stiftung

59 REFERENCES 11. Hollander AP, Pidoux I, Reiner A, Rorabeck C, Bourne R, Poole AR. Damage to type II 1. Appleyard R. Topographical analysis of the collagen in aging and osteoarthritis starts at structural, biochemical and dynamic biome- the articular surface, originates around chon- chanical properties of cartilage in an ovine drocytes, and extends into the cartilage with model of osteoarthritis. Osteoarthritis and progressive degeneration. The Journal of Clin- Cartilage 11: 65-77, 2003. ical Investigation 96: 2859-2869, 1995. 2. Ardura Garcia H, Daniels AU, Wirz D. Du- 12. Hunziker EB. Articular cartilage repair: ba- al-mode dynamic functional stiffness of ar- sic science and clinical progress. A review of ticular cartilage. European Cells and Materi- the current status and prospects. Osteoarthri- als 16, 2008. tis and Cartilage 10: 432-463, 2001. 3. Arnold MP, Daniels AU, Ronken S, Ardura 13. Jurvelin J, Buschmann M, Hunziker E. Op- Garcia H, Friederich NF, Kurokawa T, Gong tical and mechanical determination of pois- JP, Wirz D. Acrylamide Polymer Double- son’s ratio of adult bovine humeral articular Network Hydrogels: Candidate Cartilage cartilage. Journal of Biomechanics 30: 235- Repair Materials with Cartilage-Like Dy- 241, 1997. namic Stiffness and Attractive Surgery-Re- 14. Jurvelin J, Kiviranta I, Säämänen AM, Tam- lated Attachment Mechanics. Cartilage (May mi M, Helminen HJ. Indentation stiffness of 20, 2011). doi: 10.1177/1947603511402320. young canine knee articular cartilage - Influ- 4. Athanasiou KA, Agarwal A, Dzida FJ. Com- ence of strenuous joint loading. Journal of parative study of the intrinsic mechanical Biomechanics 23: 1239-1246, 1990. properties of the human acetabular and fem- 15. Kleemann RU, Krocker D, Cedraro A, oral head cartilage. Journal of Orthopaedic Tuischer J, Duda GN. Altered cartilage me- Research 12: 340-349, 1994. chanics and histology in knee osteoarthritis: 5. Bückle H. The science of hardness testing relation to clinical assessment (ICRS Grade). and its research applications. American Soci- Osteoarthritis and Cartilage 13: 958-963, ety for Metals, Metals Park, OH, 1973. 2005. 6. Brown KL, Cruess RL. Bone and cartilage 16. Korhonen RK, Laasanen MS, Töyräs J, Ri- transplantation in orthopaedic surgery. A re- eppo J, Hirvonen J, Helminen HJ, Jurvelin view. The Journal of Bone and Joint Surgery J. Comparison of the equilibrium response 64: 270-279, 1982. of articular cartilage in unconfined compres- 7. Cs-Szabó G, Melching LI, Roughley PJ, sion, confined compression and indentation. Glant TT. Changes in messenger rna and Journal of Biomechanics 35: 903-909, 2002. protein levels of proteoglycans and link pro- 17. Korhonen RK, Wong M, Arokoski J, Lind- tein in human osteoarthritic cartilage sam- gren R, Helminen HJ, Hunziker EB, Jurv- ples. Arthritis and Rheumatism 40: 1037-1045, elin J. Importance of the superficial tissue 1997. layer for the indentation stiffness of articular 8. Ficklin T, Thomas G, Barthel JC, Asanbaeva cartilage. Medical Engineering & Physics 24: A, Thonar EJ, Masuda K, Chen AC, Sah RL, 99-108, 2002. Davol A, Klisch SM. Articular cartilage me- 18. Kren AP, Rudnitskii VA, Deikun IG. Deter- chanical and biochemical property relations mining the viscoelastic parameters of vulca- before and after in vitro growth. Journal of nisates by the dynamic indentation method Biomechanics 40: 3607-3614, 2007. using a non-linear deformation model. Inter- 9. Fortin PR, Penrod JR, Clarke AE, St-Pierre national Polymer Science and Technology 32: Y, Joseph L, Bélisle P, Liang MH, Ferland D, 19-23, 2005. Phillips CB, Mahomed N, Tanzer M, Sledge 19. Kääb MJ, Gwynn IA, Nötzli HP. Collagen C, Fossel AH, Katz JN. Timing of total fibre arrangement in the tibial plateau articu- joint replacement affects clinical outcomes lar cartilage of man and other mammalian among patients with osteoarthritis of the hip species. Journal of Anatomy 193: 23-34, 1998. or knee. Arthritis and Rheumatism 46: 3327- 20. Lippiello L, Hall D, Mankin HJ. Collagen 3330, 2002. synthesis in normal and osteoarthritic hu- 10. Franke O, Durst K, Maier V, Göken M, man cartilage. The Journal of Clinical Inves- Birkholz T, Schneider H, Hennig F, Gelse K. tigation 59: 593-600, 1977. Mechanical properties of hyaline and repair 21. Loeser RF. Aging cartilage and osteoarthri- cartilage studied by nanoindentation. Acta tis--what’s the link? Science of aging knowl- Biomaterialia 3: 873-881, 2007. edge environment: pe31, 2004.

60 22. Lyyra T, Kiviranta I, Väätäinen U, Helmin- 29. Park S, Hung C, Ateshian G. Mechanical en HJ, Jurvelin J. In vivo characterization of response of bovine articular cartilage under indentation stiffness of articular cartilage in dynamic unconfined compression loading at the normal human knee. Journal of Biomedi- physiological stress levels. Osteoarthritis and cal Materials Research 48: 482-487, 1999. Cartilage 12: 65-73, 2004. 23. Mow V, Gibbs M, Lai W, Zhu W, Athana- 30. Peterson L, Brittberg M, Kiviranta I, Åker- siou KA. Biphasic indentation of articular lund EL, Lindahl A. Autologous chondro- cartilage--II. A numerical algorithm and an cyte transplantation: Biomechanics and experimental study. Journal of Biomechanics long-term durability. The American Journal of 22: 853-861, 1989. Sports Medicine 30: 2-12, 2002. 24. Niederauer GG, Niederauer GM, Cullen LC, 31. Pullig O, Weseloh G, Swoboda B. Expression Athanasiou KA, Thomas JB, Niederauer MQ. of type VI collagen in normal and osteoar- Correlation of cartilage stiffness to thickness thritic human cartilage. Osteoarthritis and and level of degeneration using a handheld Cartilage 7: 191-202, 1999. indentation probe. Annals of Biomedical En- 32. Schinagl RM, Gurskis D, Chen AC, Sah gineering 32: 352-359, 2004. RL. Depth-dependent confined compression 25. Nieminen MT, Töyräs J, Laasanen MS, Sil- modulus of full-thickness bovine articular vennoinen J, Helminen HJ, Jurvelin J. Pre- cartilage. Journal of Orthopaedic Research 15: diction of biomechanical properties of ar- 499-506, 1997. ticular cartilage with quantitative magnetic 33. Stolz M, Raiteri R, Daniels AU, VanLand- resonance imaging. Journal of Biomechanics ingham MR, Baschong W, Aebi U. Dynamic 37: 321-328, 2004. elastic modulus of porcine articular cartilage 26. Nugent GE, Law AW, Wong EG, Temple determined at two different levels of tissue MM, Bae WC, Chen AC, Kawcak CE, Sah organization by indentation-type atomic RL. Site- and exercise-related variation in force microscopy. Biophysical Journal 86: structure and function of cartilage from 3269-3283, 2004. equine distal metacarpal condyle. Osteoar- 34. Wells T, Davidson C, Mörgelin M, Bird JLE, thritis and Cartilage 12: 826-833, 2004. Bayliss MT, Dudhia J. Age-related changes 27. Oakley SP, Lassere MN, Portek I, Szomor Z, in the composition, the molecular stoichiom- Ghosh P, Kirkham BW, Murrell GAC, Wulf etry and the stability of proteoglycan aggre- S, Appleyard RC. Biomechanical, histologic gates extracted from human articular carti- and macroscopic assessment of articular car- lage. Biochemical Journal 370: 69-79, 2003. tilage in a sheep model of osteoarthritis. Os- 35. Wirz D, Kohler K, Keller B, Göpfert B, Hu- teoarthritis and Cartilage 12: 667-679, 2004. detz D, Daniels AU. Dynamic stiffness of 28. O’Hara BP, Urban JPG, Maroudas A. Influ- articular cartilage by single impact micro- ence of cyclic loading articular cartilage. An- indentation (SIMI). Journal of Biomechanics nals of the Rheumatic Diseases 49: 536-539, 41: S172, 2008. 1990.

4 Acrylamide polymer double- network hydrogels: Candidate cartilage repair materials with cartilage-like dynamic stiffness and attractive surgery-related attachment mechanics

In focal repair of joint cartilage and meniscus, initial stiffness and strength of repairs are generally much less than surrounding tissue. This increases early failure potential. Secure primary fixation of the repair material is also a problem. Acrylamide polymer double-network (DN) hydrogels are candidate-improved repair materials. DN-gels have exceptional strength and toughness compared to ordinary gels. This stems from the double-network structure in which there is a high molar ratio of the second network to the first network, with the first network highly crosslinked and the second loosely crosslinked. Previous studies of acrylic PAMPS/ PDMAAm and PAMPS/PAAm DN-gels demonstrated physicochemical stability and tissue compatibility as well as the ability to foster cartilage formation. Mechanical properties related to surgical use were tested in 2 types of DN-gels. Results: Remarkably, these >90%-water DN-gels exhibited dynamic impact stiffness (E*) values (~1.1 and ~1.5 MPa) approaching swine meniscus (~2.9 MPa). Dy- namic impact energy-absorbing capability was much lower (median loss angles of ~2°) than swine meniscus (>10°), but it is intriguing that >90%-water materials can efficiently store energy. Also, fine 4/0 suture tear-out strength approached cartilage (~2.1 and ~7.1 N v. ~13.5 N). Initial strength of attachment of DN-gels to cartilage with acrylic tissue An adapted version of this chapter has been pub- adhesive was also high (~0.20 and ~0.15 N/ lished as: M.P. Arnold, A.U. Daniels, S. Ronken, mm2). DN-gel strength and toughness prop- H. Ardura Garcia, N.F. Friederich, T. Kurokawa, erties stem from optimized entanglement of J.P. Gong and D. Wirz. Acrylamide polymer the 2 network components. DN-gels thus Double-Network hydrogels: Candidate carti- have obvious structural parallels with cartilage repair materials with cartilage-like dynamic laginous tissues, and their surgical handling stiffness and attractive surgery-related attach- properties make them ideal candidates for ment mechanics. Cartilage, 2011 clinical use.

INTRODUCTION scaffold. In this case, the plugs must also be extremely durable (lasting years) in order to be of Repair of Cartilage Lesions clinical use. In either case (scaffold or plug), it is also necessary to have a means for securing Cartilage and meniscal lesions have limited po- the implant to the osteochondral bone and the tential for spontaneous repair [9, 17]. Specifical- surrounding intact cartilage, which can be par- ly, joint surface lesions with surface areas larger ticularly challenging in defects that are not well than 4 cm2 are now believed to inevitably lead contained. After initial surgical wound healing, to degenerative arthritis [11]. This is especially scaffolds or plugs with cartilagelike mechanical the case in the knee, with serious consequences properties would be able to immediately distrib- (debilitating pain and markedly restricted mobil- ute gait-related biomechanical impacts nondis- ity), often leading to the need for major surgery, ruptively to the surrounding natural cartilage that is, total joint arthroplasty. As a consequence, and also protect the sensitive subchondral bone. in recent years, much effort has been devoted to developing methods for repairing such lesions. Double-Network Hydrogels The pioneering work of Peterson et al. [14-16] showed that a suspension of the patient’s Double-network hydrogels (DN-gels) are a new own previously harvested and expanded chon- family of candidate materials for potential use drocytes, injected behind a periosteal flap, is able in the repair of skeletal system soft tissues. They to build up hyaline-like cartilage tissue. How- have been developed for these and other pur- ever, this method has its disadvantages, mostly poses by Gong et al. at Hokkaido University the time required for the tissue to form and a and reported in the literature starting in 2003 consequent long rehabilitation period, up to 1 [5]. Ordinary single-network hydrogels contain- year, until pain-free full weightbearing is pos- ing 85% to 95% water do not have cartilage-like sible and joint homeostasis is re-established. As compressive strength. For example, articular car- a result, alternative methods have been proposed tilage is reported to have a compressive fracture and pursued. The main approach has been to strength of approximately 36 MPa [7]. In con- place harvested and expanded chondrocytes in a trast, an example of a single-network 92%-wa- scaffold material and stimulate them in vitro to ter gel, based on an acrylamide polymer, poly(2- begin cartilage formation in the scaffold prior to acrylamido-2-methylpropanesulfonic acid), that implantation. The resultant “construct” is then is, PAMPS, which is highly crosslinked (~4 mol implanted [22]. Using this approach, the reha- %), has a compressive fracture stress of only 0.4 bilitation period can be shortened considerably. MPa. However, when a large molar ratio of a sec- However, such tissue-engineered con- ond acrylamide polymer, poly(acrylamide), that structs still do not have mechanical properties is, PAAm, is added to PAMPS and controlled to (e.g., stiffness, strength) at the time of implan- be lightly crosslinked (e.g., 0.1 mol %), the result tation that are even remotely similar to natural is a 90%-water PAMPS/PAAm DN-gel with a articular cartilage. As a result, the rehabilita- markedly higher compressive fracture stress, 17.2 tion period must still be on the order of several MPa, which is 43 times higher than the PAMPS months in order to establish repaired tissue ca- gel [5]. In addition, a PAMPS/PAAm hydrogel pable of bearing cyclic impact loads in the knee with a >90%-water content does not fail until of the magnitude and frequency associated with compressive strain is over 90%. For comparison, normal daily activity [9, 10, 11, 18]. a commercially available PVA hydrogel is report- In order to further shorten the rehabilita- ed to have a compressive fracture stress in the tion period needed after a cell-based cartilage re- range of only 1.4 to 2.0 MPa and fails at 47% to pair, a tissue-engineered scaffold with cartilage- 62% compressive strain [19]. The tough 90%-wa- like initial mechanical properties (and of course ter PAMPS/PAAm DN-gel studied here thus ap- the ability to foster cartilage formation) would proaches the compressive strength of articular be an attractive solution. Alternatively, for small cartilage. It should be noted that cartilage and repairs, one could also consider using plugs of other skeletal system tissues are also high water- a completely artificial solid material with car- content materials and employ crosslinking and a tilage-like mechanical properties rather than a double-network strategy (e.g., highly crosslinked

67 collagen plus proteoglycan gel) to achieve their good idea to develop and use cartilage-like repair mechanical properties. materials that can be safely glued into the defect, Gong et al. have performed and reported thus avoiding the need for sutures. a variety of preliminary, promising biomechani- Because the DN-gels studied here are cal and biological studies of DN-gels over the based on acrylamide polymers, it seemed likely past few years [2, 5, 12, 13, 20, 26]. Recently, that an acrylic tissue adhesive might work well a study of the repair of induced osteochondral to bond them to other surfaces, which is why we defects in rabbit knees with a DN-gel composed chose to investigate whether this was the case. of a PAMPS first network and a PDMAAm, As a result of reviewing the intriguing work of that is, poly(N,N’-dimethylacrylamide), second Gong and her group, we were fortunate to estab- network was performed [25]. The PAMPS/PD- lish a collaboration with her to study further the MAAm DN-gel used had been shown previ- properties of DN-gels that might be of interest ously in a rabbit model to exhibit no decline in before their clinical use. stiffness, strength, or strain at failure at 6 weeks The aims of this study of two types of [2] and to elicit little inflammatory response [20]. acrylic polymer DN-gels were the following: For the repair study [25], the defects created in • to measure dynamic stiffness and the the patellofemoral groove of the femoral condyle ability to dissipate energy using function- were 4.3 mm in diameter and 15 mm deep, thus related mechanical techniques previously extending approximately 12 mm or more into established by the authors [4, 24]; osteochondral bone. The bony part of the de- • to devise and use methods to measure the fect was partially filled with a cylindrical plug surgical attachment strength that can be of the same diameter made from the PAMPS/ achieved with 1) sutures and 2) a surgical PDMAAm DN-gel, leaving the last 1.5 to 2.5 tissue adhesive; and mm of depth (relative to the original cartilage • to compare the results for the 2 DN-gels surface) empty. After 4 weeks, the empty space and compare the properties with those of (above the DN-gel plugs) had become complete- natural cartilage where applicable. ly filled with white, opaque tissue. It appears that hyaline-like cartilage was formed on top of the PAMPS/PDMAAm cylinders in the osteochon- MATERIALS dral defects. Tissue adhesives are increasingly being For this study, two different acrylamide poly- evaluated and used as an alternative to sutures mer DN-gels, known as PAMPS/PDMAAm for small-scale repairs. They offer the potential and PAMPS/PAAm, were provided. The pre- advantage of distributing the load over a much paring ratios and water content of the two DN- larger interfacial area than is possible with su- gels are given in Table 4.1. While the second- tures and thus markedly reducing the focal network component was different for the two stresses created by sutures. They also offer speed gels, (DMAAm v. AAm), the only difference in and simplicity compared to sutures, and the preparing ratios for the two gels was for the ul- repairs have been found to be sufficiently - du traviolet initiator for the second-network compo- rable to allow subsequent healing in many ap- nent (0.03 v. 0.01 mol%). As shown, both DN- plications. The inflammatory response to clini- gels contained more than 90% water. However, cally approved adhesives is acceptably low. A the difference in water content (94.0% v. 90.9%) recent orthopedic surgery-related in vitro study means that the PAMPS/PDMAAm gel con- compared a clinically approved tissue adhesive, tained only approximately 66% as much polymer Histoacryl (primary active ingredient of N-bu- as the PAMPS/PAAm. The methods for produc- tyl-2-cyanoacrylate; B. Braun Melsungen AG, ing the DN-gel structures from the polymeric Melsungen, Germany), with sutures in the re- components are described elsewhere [5, 23]. The pair of knee meniscal tears [1]. They found His- DN-gels were then placed in normal saline and toacryl (B. Braun Melsungen AG) significantly shipped to Basel by ordinary post, where they increased the force required to produce a 2 mm were kept in saline at 4 °C to 6 °C before testing gap in the repairs. Because of the technically at room temperature. Specimen dimensions for easier handling for the surgeon, it might be a each test mode are described later.

68 Table 4.1: Preparing ratios and water content of the two double-network (DN) polymer hydrogels Components: first network Components: second network DN-gel Monomer Cross- UVI Monomer Cross- UVI Watera linker linker PAMPS/ AMPS MBAA o.1 DMAAm MBAA 0.03 94.0% PDMAAm 1 mol/l 4 mol% mol% 2 mol/l 0.01 mol% mol% PAMPS/ AMPS MBAA 0.1 AAm MBAA 0.01 90.0% PAAm 1 mol/l 4 mol% mol% 2 mol/l 0.01 mol% mol%

Note: Prepared from the following: AMPS = 2-acrylamido-2-methylpropanesulfonic acid; DMAAm = N,N’-dimethyl acrylamide; AAm = acrylamide; MBAA = N,N’-methylenebisacrylamide; UVI = ultraviolet light initiator. aDN gel equilibrium water content in normal saline (0.91% w/v NaCl).

METHODS tured by the electromagnetic coil through which the indenter moves.In these tests, the mass- and Dynamic stiffness by millimeter-scale micro- gravity-produced acceleration of the indenter indentation results in nondestructive indentations having depths of 0.1 to 0.2 mm. From the dynamic mo- The Basel authors have previously developed and tion data and indenter mass and geometry, it is used microindentation methods for determining possible to calculate the same parameters as in the dynamic stiffness parameters (dynamic mod- cyclic loading tests: E* and d. Ten DN-gel spec- ulus [E*] and loss angle [d]) of cartilage and me- imens were tested; they were 3 mm thick and niscus [4, 24]. The methods were employed here about 10 × 20 mm in lateral dimensions. They on the DN-gel specimens. Briefly, the methods are designed to recognize two things: 1) the values of E* and d for poroviscoelastic materials are extremely dependent on deformation rate; and 2) articular cartilage must function in two extremely different loading regimes: the sudden transient deformations that occur during gait, and the slow quasicyclic deformations that cause fluid to move in and out of cartilage and thus provide a means for nutrition. Thus, both a “gait mode” and a “nutrition mode” testing procedure have been developed. In both tests, described briefly below, it is possible to measure the loss angle directly (Figure 1.10). In our gait mode, evaluation of dynamic stiffness is accomplished by fast impact (FI) microindentation (MI), using a modified version (Figure 4.1) of an instrument developed at the Minsk Institute of Physics [8]. FIMI does not precisely duplicate the complex impact loading/ unloading patterns seen in gait. However, the indenter velocity at impact is in the gait range: approximately 0.3 m/s. Briefly, the dynamic motion (distance v. time) of a falling microindenter (steel, 1.0 mm diameter spherical tip; 1.7 g mass Figure 4.1: Fast impact (FI) mode modulus and loss of indenter) is captured electromagnetically. The angle measurement device, mounted on a stable loading velocity at impact is among the parameters cap- frame, equipped with a load cell and laser positioning system.

69 saline during testing at room temperature. For a given specimen, 10 replicate slow sinusoidal tests were performed at the same spot at intervals of about 20 seconds, shown to be sufficient to -al low dimensional recovery. The resultant E* and d were averaged to produce E* and d values for a given specimen.

Suture tear out

There are clinical situations in which cartilage defects are not perfectly contained. In such cases, inserting a simple unsecured plug of repair material is not an adequate repair technique. It is thus an advantage if a repair material can be secured with sutures. DN-gels are known to be highly resistant to propagation of a preinduced slit in a standardized test of tear resistance [21]. Therefore, high suture tear-out forces could be expected. To test this hypothesis, again, the MTS Synergie 100 test instrument (MTS Sys- tems Corporation) was used. DN-gel specimen dimensions were 3 × 10 × 20 mm; 3 specimens of both types of DN-gel were tested. One end of a DN-gel specimen was fixed using acrylic adhesive in an aluminum fixture matching the Figure 4.2: Spherical steel indenter, 3.2 mm in diameter, mounted on a material testing system (MTS Synergie 100, thickness and exceeding the width of the DN- MTS Systems Corporation, Eden Prairie, MN), indent- gel specimens. A small-diameter (4/0 = 0.15 mm) ing a 3-mm-thick PAMPS/PDMAAm hydrogel speci- surgical suture was passed through the other end men. This configuration was used for the slow sinusoidal of the DN gel specimen, laterally centered and (SS) microindentation tests. approximately 3 mm from the end of the specimen, using the needle integrated with the suture by the manufacturer. The suture was Vicryl 4/0 were kept moist with saline during testing at (Ethicon, Johnson & Johnson Medical GmbH, room temperature. For a given specimen, 10 rep- Neuss, Germany), a type used in fine-scale soft licate impact tests were performed at the same tissue approximation. It is composed of a braided spot at intervals of approximately 20 seconds, bioabsorbable copolymer (Polyglactin 910 = gly- and the resultant E* and d were averaged to pro- colide-L-lactide random copolymer) coated with duce E* and d values for a given specimen. another bioabsorbable copolymer (Polyglactin In our nutrition mode, evaluation of dy- 370 = 65/35 mole ratio lactide-glycolide copoly- namic stiffness is accomplished by slow sinusoi- mer). The suture is further coated with calcium dal microindentation (SSMI). The SSMI tests are stearate to promote easy passage through tis- performed with a MTS Synergie 100 mechanical sues, precise knotting, and so on. After passage testing instrument (MTS Systems Corporation, through the gel, the suture was then tied to itself Eden Prairie, MN), programmed to perform a to form a loop. The loop was slipped through a series of single sinusoidal cycles at 0.1 Hz. The hook fixture attached to the test machine platen. microindenter (steel, ~3.2 mm diameter spheri- The aluminum fixture was attached to a 100 N cal tip) moves under sinusoidal displacement load cell attached to the crosshead of the test control with a maximum speed of 0.015 m/s to machine. Specimens were kept moist with saline a depth of approximately 0.1 mm (Figure 4.2). during testing at room temperature. The cross- The same 10 DN-gel specimens were tested as head was moved upward at a speed of 1 mm/s, in gait mode. They were again kept moist with while recording load cell force versus crosshead

70 motion, until the suture completely tore through surface secured to the test machine’s fixed plat- the DN-gel specimen. en. Test surfaces were either ordinary plate glass The MTS Synergie 100 test instrument (precleaned with ethanol) or articular cartilage (MTS Systems Corporation) was also used for in the form of osteochondral plugs 7.6 mm in the tissue adhesive tests. The adhesive used diameter, taken from the knees of 9 month old was Histoacryl (B. Braun Melsungen AG) (see swine, obtained from a retail meat vendor. These above). Three specimens of both types of DN- specimens were fresh frozen in 0.9% saline so- gel were tested; the dimensions were 3 × 10 × 20 lution and thawed before testing. The cartilage mm. One end of the specimen was secured with surface was used without any cleaning except acrylic glue in a slot opening in a small alumi- removal of surface moisture with a soft paper tis- num fixture with slot dimensions matching the sue. After allowing a minimum 60 seconds for thickness and exceeding the width of the DN- adhesion to become secure, the crosshead of the gel specimens. The fixture was attached to the test machine was raised at a speed of 1 mm/s load cell/crosshead of the test machine. A drop while recording force and crosshead displace- of Histoacryl (B. Braun Melsungen AG) was ment. The results were normalized to the appar- applied to the other end of the DN-gel speci- ent contact area between DN-gels and material men, which was then lowered to contact a test surface.

Table 4.2: Dynamic modulus E* of double-network hydrogels in fast impact (FI) and slow sinusoidal (SS) mode tests Dynamic mod- Mean Standard Median ulus E* [MPa] deviation PAMPS/ 1.52 0.07 1.53 PDMAAm FI PAMPS/ 1.07 0.04 1.05 PAAm FI PAMPS/ 0.85 0.09 0.87 PDMAAm SS PAMPS/ 0.5 0.14 0.5 PAAm SS Note: Number of replicate specimens (n = 10)

Table 4.3: Loss angle d of double-network hydrogels in fast impact (FI) and slow sinusoidal (SS) mode tests Loss Angle [d] Mean Standard Median deviation PAMPS/ 1.75 0.36 1.91 PDMAAm FI PAMPS/ 2.03 0.34 2.14 Figure 4.3: Comparison of stiffness properties of PAMPS PAAm FI doublenetwork hydrogels (DN-gels) with different second-network components (PDMAAm and PAAm). Dy- PAMPS/ 1.43 1.2 1.42 d namic modulus E* and loss angle measured in both fast PDMAAm SS impact (FI) and slow sinusoidal (SS) microindentation. See text for complete descriptions of test modes. For each PAMPS/ 5.84 7.3 3.04 DN-gel and mode, the left bar shows box and whisker PAAm SS plot showing median and quartiles, and the right bar shows mean and standard deviation. Note: Number of replicate specimens (n = 10)

71 Statistical Analysis

Wilcoxon rank–sum and signed–rank tests were performed on stiffness data (p < 0.05). Statisti- cal analysis was performed and created using R (http://www.R-project.org).

RESULTS

Stiffness

d Figure 4.4: Tear-out forces for 4/0 (0.15 mm diameter) For dynamic modulus E* and loss angle , the braided suture versus crosshead displacement. Three rep- two DN-gels were measured in two modes: licate tests for each 3 mm thick gel specimen. Median FIMI and SSMI (Tables 4.2 and 4.3; Figure 4.3). maximum force of PAMPS/ PDMAAm was 2.1 N, and In general, it is expected for viscoelastic materi- median maximum force of PAMPS/PAAm was 7.4 N. als to have a higher E* and a lower d at more rapid deformation rates. The E* for both DN-gels was significantly higher at the more rapid (FI) deformation rate. However, d did not change significantly between the two deformation rates employed. A paired Wilcoxon rank–sum test revealed a significant difference for each DN-gel between FI- and SS-mode values for E* but not for d. The differences between the two gels were significant for E* but not for d.

Suture Tear Out Figure 4.5: Pull-off force normalized by contact area versus crosshead displacement for double-network (DN) specimens attached to a glass plate using an acrylic tissue Recordings of resultant force versus test instru- adhesive. Nominal area of contact = 3 × 10 mm. Three ment crosshead position are shown in Figure replicate tests for each DN-gel. Median maximum pull- 4.4. In these exploratory experiments, only three off force of PAMPS/PDMAAm was 0.23 N/mm2, and replicate tests for each DN-gel were possible, so median maximum pull-off force of PAMPS/PAAm was 0.18 N/mm2. statistical comparison of results for the two gels was not feasible. The median maximum tear-out force for 94%-water PAMPS/PDMAAm was approximately 2.1 N, and the median maximum for 90.9%-water PAMPS/PAAm was approximately 7.1 N.

Attachment to Surfaces Using a Tissue Adhe- sive

Figure 4.5 shows the force versus crosshead position for the adhesion of DN-gels to a glass plate using Histoacryl (B. Braun Melsungen Figure 4.6: Pull-off force normalized by contact area AG) tissue adhesive. Also in these exploratory versus crosshead displacement for double-network hy- experiments, only three replicate tests for each drogel (DN-gel) specimens attached to porcine articular DN-gel were possible, so statistical comparison cartilage using an acrylic tissue adhesive. Three replicate tests for each DN-gel. Median maximum pull-off force of of results for the two gels was not feasible. The PAMPS/PDMAAm was 0.20 N/mm2, and median maxi- median maximum pull-off force of PAMPS/PD- mum pull-off force of PAMPS/PAAm was 0.15 N/mm2.

72 MAAm was 0.23 N/mm2 (range, 0.21-0.59 N/ mode. However, the PAMPS/PDMAAm gel was mm2), and median maximum pull-off force of approximately 50% as stiff as swine meniscus in PAMPS/PAAm was 0.18 N/mm2 (range, 0.09- both FI-mode and SS-mode. It seems likely that 0.23 N/mm2). To the naked eye, pull off always PAMPS/PDMAAm DN-gels with lower water occurred at the interface between the gel and the content, closer to cartilage (e.g., 65%-80% rather glass surface rather than in the gel substance. than >90%), may have E* values more closely ap- Similarly, Figure 4.6 shows force versus proaching that of the articular cartilage surface. crosshead position for the adhesion of DN-gels The two DN-gels exhibit some charac- to porcine articular cartilage using Histoacryl (B. teristics of a viscoelastic material because their Braun Melsungen AG) tissue adhesive. Again, moduli decline when the deformation rate is de- only three replicate tests for each DN-gel were creased (i.e., from FI-mode to SS-mode). How- possible. The median maximum pull-off force ever, one measure of viscoelasticity, that is, the of PAMPS/PDMAAm was 0.20 N/mm2 (range, loss angle, of the two gels did not differ signifi- 0.15-0.23 N/mm2), and median maximum pull- cantly, and both DN-gels were relatively “rub- off force of PAMPS/PAAm was 0.15 N/mm2 bery,” that is, more energy storing, in that the (range, 0.14-0.24 N/mm2). To the naked eye, loss angle d was low even at 0.1 Hz in the slow, pull off always occurred at the interface between cyclic N-mode tests. In both modes, median val- cartilage surface and the DN-gels rather than in ues were below 4° for both gels. In general terms, either the cartilage or gel substance. It is interest- the low loss angles of these DN-gels in slow ing to note that the forces required to produce (0.1 Hz) cyclic deformation can be attributed adhesion failures approached the forces required to the fact that their structures are chemically to produce single-suture tear out. crosslinked. Therefore, deformation does not en- tail much sliding between polymer chains, and there is thus little of the attendant frictional dis- DISCUSSION sipation of energy that is known to result from such sliding. This less viscoelastic, more energy- The DN-gels showed very promising results, be- storing character of these acrylamide DN-gels is ing almost as stiff as normal cartilage and al- particularly evident in comparison to swine ar- lowing for a safe fixation to the surrounding ticular cartilage and meniscus [25]. The FI-mode tissue, either by suturing or gluing. The results loss angle for both these tissues is approximately of a given test of fixation (e.g., suture tear out 12°, and the N-mode loss angle is approximately or attachment of specimens to surfaces with tis- 37° for cartilage and approximately 26° for me- sue adhesive) showed a wide spread. This was niscus [25]. Thus, the tissues, in spite of being due to the fact that it was intended to simulate stiffer, can be viewed as better in absorbing and the variable situation in the operation room. For dissipating energy than the two DN-gels. What example, for the attachment with cyanoacrylat- difference this might make in implant use and eglue to cartilage, we dried the surface with a whether it would be possible to produce a DN- paper towel, added a drop of glue, and placed the gel structure comparable to either of these tis- gels by hand on the cartilage surface. sues in both dynamic stiffness and viscoelasticity have not been explored yet. Stiffness Parameters It is interesting to note that the PAMPS/ PDMAAm gel was stiffer than the PAMPS/ The PAMPS/PDMAAm gel was significantly PAAm gel in spite of having a higher water (~41%) stiffer (higher E*) than the PAMPS/ content (94% v. 90.9%) and correspondingly PAAm gel under fast (FI-mode) deformation. containing only approximately 66% as much PAMPS/PDMAAm was also stiffer (~12%) in polymer. This suggests that the gel stiffness -pa slow deformation (SS-mode), but the difference rameters are more strongly influenced by differ- was not significant. In a separate study [20, 25], ences in structure related to the use of differ- we have evaluated the stiffness of swine knee ar- ent second-network components (PDMAAm v. ticular cartilage and meniscus in the same two PAAm) than by gel water content. test modes. The DN-gel E* values were only about 10% of cartilage values in either FI- or SS-

73 Surgical Fixation Stability These preliminary results for the adhesion of DN-gels to cartilage with tissue adhesive As mentioned in the Introduction, there are are certainly promising. The force-to-failure val- clinical situations in which cartilage defects are ues are of the same order of magnitude as for not perfectly contained. It is thus an advantage single-suture tear out. On the one hand, mul- if a repair material can be secured with sutures. tiple sutures might provide even greater tear-out The suture tear-out strength experiments showed strength, although practically speaking, each that both of these DN-gels were tear resistant suture has the disadvantage of damaging the enough to be secured with fine surgical sutures. normal surrounding cartilage [6]. On the other The suture tear out of PAMPS/PDMAAm was hand, no effort was made here to optimize the comparable to nasal cartilage pull out of sutures. adhesion of gel to cartilage; for example, the Farhadi et al. [3] showed for nasal cartilage a actual area of cartilage-gel apposition may have suture pull-out force of 4.5 N/mm normalized been less than the apparent area, and no effort to the thickness of the specimen. The value for was made to prepare either surface in any special PAMPS/PDMAAm normalized to the thickness way. It should also be noted that attachment was is 3.5 N/mm. Perhaps in the suture tear-out tests to a mechanically cut gel surface. Thus, one can reported here, the somewhat lower water content concludethat this pull-off strength in this test (higher polymer content) of the PAMPS/PAAm was due to bonding with the internal gel bulk gel also contributed to increased toughness com- structure, not just to a bond with as-prepared pared to the PAMPS/PDMAAm gel. The tear- surface structures. However, the asprepared sur- out strength for this DN-gel should be seen as face of DN-gels is known to be covered with the truly remarkable considering its 90.9%-water second-network component, and thus, adhesion content. The high suture tearout strength stems to this surface might well be different. However, from the exceptional fracture energy of acryl- it should be noted that retention of water in these amide DN-gels, as high as 103 J/m2. In studies of DN-gels does not depend on a special surface PAMPS/PAAm, it has been shown that achiev- structure. The gels are not observed to leak water ing such fracture energies depends critically on either when they are cut or subjected to substan- several factors, and it is believed that the mo- tial mechanical deformation for short times. It is lecular weight of the second-network component certainly a weakness of our study that no more is the most important one. If it is above a certain than three measurements per DN-gel type were value, the increase in chain entanglement greatly recorded, but we felt that this was sufficient data increases the work to fracture [23]. to give a good impression that DN-gels can ei- The tissue adhesive pull-off strength tests ther be glued or sutured to surrounding cartilage also showed promising high results for both of and bone. the two DN-gels to either inorganic (silicate Concerning the future of PAMPS/PAAm glass) or tissue cartilage) surfaces. The results hydrogels for clinical cartilage repair, it may be thus suggest that pull-off strength is more close- possible to alter their structure to render them ly related to bonding phenomena between the anisotropic and thus further mimic the structure primary DN-gel component (PAMPS in both and properties of articular cartilage. Also, as yet cases) and the glass or cartilage surface than unpublished research suggests that cell infiltra- to any effects related to the second component. tion is possible. In addition, PAMPS/PAAm hy- Overall, bonding strength of the gels to glass drogels can be produced in desired shapes and/ was somewhat higher than bonding to cartilage or trimmed with surgical instruments in the (0.23 and 0.18 N/mm2 v. 0.20 and 0.15 N/mm2). operating room. Finally, sterilization can be ac- The values were variable but of the same order of complished by conventional methods without al- magnitude. The variability of the results may be tering structure and properties. These statements explained by the relatively big impact of small are based on works in progress, and the details changes of the contact areas between the DN- of methods and results are reserved for future gels and the surface they were glued to. Also, a reports. small amount of not perfect perpendicularity between glass or cartilage and the DN-gel might have influenced the forces measured.

74 CONCLUSIONS single-suture tear-out strength. Finally, the double-network structure has The previous work of Gong and her colleagues obvious parallels to the double-network strate- has already shown that acrylamide-based DN- gies employed by the body in creating cartilage gels have intriguing and promising potential for and other load-bearing soft tissues. Further labo- use in the repair of skeletal system soft tissues. ratory mechanical property studies are underway This study was performed to further investigate with acrylamide DN-gels of much lower water several mechanical properties related to clinical content, similar to articular cartilage. implant use. The results further support the potential of acrylamidebased DN-gels for such use. In spite of their very high water content, >90%, ACKNOWLEDGMENTS AND FUNDING the gels studied exhibited stiffness (E*) values approaching that of swine meniscal tissue. Funding was provided by Deutsche Arthrose- Suture tear-out strength values ap- Hilfe eingetragener Verein (Frankfurt, Germa- proached those for natural cartilage, again in ny); Hardy u. Otto Frey-Zünd Stiftung (Basel, spite of the extremely high water content. Equal- Switzerland); and a Grant-in-Aid for Specially ly intriguing was the finding that the strength of Promoted Research (no. 18002002) from the attachment of a cut gel surface to natural carti- Japanese Ministry of Education, Science, Sports lage with an acrylic tissue adhesive approached and Culture.

75 REFERENCES Macromolecules 42: 2184-2189, 2009. 13. Nakayama A, Kakugo A, Gong JP, Osada Y, 1. Ardura Garcia H, Daniels AU, Wirz D. Du- Takai M, Erata T, Kawano S. High Mechani- al-mode dynamic functional stiffness of ar- cal Strength Double-Network Hydrogel with ticular cartilage. European Cells and Materi- Bacterial Cellulose. Advanced Functional Ma- als 16, 2008. terials 14: 1124-1128, 2004. 2. Ayan I, Colak M, Comelekoglu U, Milcan A, 14. Peterson L, Brittberg M, Kiviranta I, Ak- Ogenler O, Oztuna V, Kuyurtar F. Histoac- erlund EL, Lindahl A. Autologous chon- ryl glue in meniscal repairs (a biomechanical drocyte transplantation. Biomechanics and study). International Orthopaedics 31: 241- long-term durability. The American Journal of 246, 2007. Sports Medicine 30: 2-12, 2002. 3. Azuma C, Yasuda K, Tanabe Y, Taniguro H, 15. Peterson L, Minas T, Brittberg M, Nilsson Kanaya F, Nakayama A, Chen YM, Gong JP, A, Sjögren-Jansson E, Lindahl A. Two- to Osada Y. Biodegradation of high-toughness 9-year outcome after autologous chondrocyte double network hydrogels as potential ma- transplantation of the knee. Clinical Ortho- terials for artificial cartilage. Journal of Bio- paedics and Related Research : 212-234, 2000. medical Materials Research. Part A 81: 373- 16. Peterson L. Articular cartilage injuries treat- 380, 2007. ed with autologous chondrocyte transplanta- 4. Farhadi J, Fulco I, Miot S, Wirz D, Haug M, tion in the human knee. Acta Orthopaedica Dickinson SC, Hollander AP, Daniels AU, Belgica 62 Suppl 1: 196-200, 1996. Pierer G, Heberer M, Martin I. Precultiva- 17. Saris DBF, Dhert WJA, Verbout AJ. Joint ho- tion of engineered human nasal cartilage en- meostasis: the discrepancy between old and hances the mechanical properties relevant for fresh defects in cartilage repair. The Journal of use in facial reconstructive surgery. Annals of Bone and Joint Surgery 85: 1067-1076, 2003. surgery 244: 978-985, 2006. 18. Saris DBF, Vanlauwe J, Victor J, Haspl M, 5. Gong JP, Katsuyama Y, Kurokawa T, Osada Bohnsack M, Fortems Y, Vandekerckhove B, Y. Double-Network Hydrogels with Extreme- Almqvist KF, Claes T, Handelberg F, Lagae ly High Mechanical Strength. Advanced Ma- K, van der Bauwhede J, Vandenneucker H, terials 15: 1155-1158, 2003. Yang KGA, Jelic M, Verdonk R, Veulemans 6. Hunziker EB, Stähli A. Surgical suturing N, Bellemans J, Luyten FP. Characterized of articular cartilage induces osteoarthritis- chondrocyte implantation results in better like changes. Osteoarthritis and Cartilage 16: structural repair when treating symptomatic 1067-1073, 2008. cartilage defects of the knee in a randomized 7. Kerin AJ, Wisnom MR, Adams MA. The controlled trial versus microfracture. The compressive strength of articular cartilage. American Journal of Sports Medicine 36: 235- Proceedings of the Institution of Mechanical 246, 2008. Engineers. Part H 212: 273-280, 1998. 19. Stammen JA, Williams S, Ku DN, Guldberg 8. Kren AP, Rudnitskii VA, Deikun IG. Deter- RE. Mechanical properties of a novel PVA mining the viscoelastic parameters of vulca- hydrogel in shear and unconfined compres- nisates by the dynamic indentation method sion. Biomaterials 22: 799-806, 2001. using a non-linear deformation model. Inter- 20. Tanabe Y, Yasuda K, Azuma C, Taniguro national Polymer Science and Technology 32: H, Onodera S, Suzuki A, Chen YM, Gong 19-23, 2005. JP, Osada Y. Biological responses of novel 9. Mandelbaum B, Browne J, Fu F. Articular high-toughness double network hydrogels in cartilage lesions of the knee. American Jour- muscle and the subcutaneous tissues. Journal nal of Sports Medicine 26: 853-861, 1998. of Materials Science: Materials in Medicine 19: 10. Minas T. The role of cartilage repair tech- 1379-1387, 2008. niques, including chondrocyte transplanta- 21. Tanaka Y, Kuwabara R, Na Y-H, Kurokawa tion, in focal chondral knee damage. Instruc- T, Gong JP, Osada Y. Determination of frac- tional Course Lectures 48: 629-643, 1999. ture energy of high strength double network 11. Mithoefer K, Scopp JM, Mandelbaum BR. hydrogels. The Journal of Physical Chemistry. Articular cartilage repair in athletes. Instruc- B 109: 11559-11562, 2005. tional Course Lectures 56: 457-468, 2007. 22. Temenoff JS, Mikos AG. Review: tissue engi- 12. Nakajima T, Furukawa H, Tanaka Y, Ku- neering for regeneration of articular cartilage. rokawa T, Osada Y, Gong JP. True Chemi- Biomaterials 21: 431-440, 2000. cal Structure of Double Network Hydrogels. 23. Tsukeshiba H, Huang M, Na YH, Kuro-

76 kawa T, Kuwabara R, Tanaka Y, Furukawa Kwon HJ, Onodera S, Chen YM, Kurokawa H, Osada Y, Gong JP. Effect of polymer T, Kanaya F, Ohmiya Y, Osada Y. A novel entanglement on the toughening of double double-network hydrogel induces spontane- network hydrogels. The Journal of Physical ous articular cartilage regeneration in vivo in Chemistry. B 109: 16304-16309, 2005. a large osteochondral defect. Macromolecular 24. Wirz D, Kohler C, Keller K, Göpfert B, Hu- Bioscience 9: 307-316, 2009. detz D, Daniels AU. Dynamic Stiffness of 26. Yasuda K, Ping Gong J, Katsuyama Y, Na- Articular Cartilage By Single Impact Micro- kayama A, Tanabe Y, Kondo E, Ueno M, Indentation (SIMI). Journal of Biomechanics Osada Y. Biomechanical properties of high- 41: S172, 2008. toughness double network hydrogels. Bioma- 25. Yasuda K, Kitamura N, Gong JP, Arakaki K, terials 26: 4468-4475, 2005.

5 Double network acrylamide hydrogel compositions adapted to achieve cartilage-like dynamic stiffness

Since articular cartilage has a limited potential for spontaneous healing, various techniques are employed to repair cartilage lesions. Acrylate-based double-network (DN) hydrogels containing ~90% water have shown promising properties as repair materials for skeletal system soft tissues. Although their mechanical properties approach those of native cartilage, the critical factor -stiffness- of DN-gels does not equal the stiffness of articular cartilage. This study investigated whether revised PAMPS/PAAm compositions with lower water content result in stiffness parameters closer to cartilage. DN-gels containing 61, 86 and 90% water were evaluated using two non-destructive, mm-scale indentation test-modes: fast-impact (FI) and slow- sinusoidal (SS) deformation. Deformation resistance (dynamic modulus) and energy handling (loss angle) were determined. The dynamic modulus increased with decreasing water content in both testing modes. In the 61% water DN-gel the modulus resembled that of cartilage (FI-mode: DN-gel=12, cartilage = 17; SS-mode: DN-gel =4, cartilage = 1.7 MPa). Loss angle increased with decreasing water content in fast-impact, but not in slow- sinusoidal deformation. However, loss angle An adapted version of this chapter has been ac- was still much lower than cartilage (FI: DN- cepted for publication as: S. Ronken, D. Wirz, gel = 5, cartilage = 11; SS: DN-gel = 10, car- A.U. Daniels, T. Kurokawa, J.P. Gong and M.P. tilage = 32°), indicating somewhat less ability Arnold, Double network hydrogel composition to dissipate energy. Overall, results show that adapted to achieve cartilage-like dynamic stiff- it is possible to adapt DN-gel composition to ness. Biomechanics and Modeling in Mechanobi- produce dynamic stiffness properties close to ology normal articular cartilage.

INTRODUCTION acrylamido-2-methylpropane sulfonic acid) with a second slightly crosslinked polymer, PAAm Articular cartilage is an extraordinary tissue -- i.e. poly(acrylamide) then created within the because of its ability to tolerate a tremendous first structure. PAMPS/PAAm is a tough, tear amount of intensive and repetitive physical resistant, non cytotoxic, non-absorbable DN-gel. stress, and to frequently do so for a lifetime [17]. The surprising result is a material with impact However, the greatest limitation of articular and tear resistance approaching an elastomer in cartilage is its poor capacity to heal spontane- spite of the high water-content. Structurally, they ously [5, 9]. Consequently, unhealed damage to also resemble cartilage and other skeletal system articular cartilage in the knee is a common clini- soft tissues, which are also high water-content cal problem. Widuchowski et al. [27] examined materials or structures with a double-network 25,124 knees from patients with acute knee in- strategy to achieve their mechanical properties. juries or unexplained knee pain and dysfunction Cartilage for instance can be seen as a high wa- from 1989 to 2004. Arthroscopy revealed that ter content double network material or structure 60% of the patients had chondral lesions. Also comprised of highly crosslinked collagen fibres after anterior cruciate ligament injuries, the risk interspersed with proteoglycan gel. of cartilage lesions is very high [24]. These un- Importantly, a closely related formula- healed injuries can lead to chronic pain, mark- tion, PAMPS/PDMAAm DN-gel, has previ- edly restricted mobility and further degeneration ously been shown to support cartilage formation. of the articular cartilage, such as secondary os- In a rabbit study [30] a DN-gel plug inserted in teoarthritis. In many cases the situation must fi- a large osteochondral defect induced substan- nally be ameliorated by major surgery - i.e. total tial spontaneous cartilage formation in vivo joint arthroplasty. by 4 weeks. In contrast, this was rarely found In recent years several techniques have for empty defects or defects filled with either a been devised and used to repair cartilage lesions polyvinylacrylate gel or an ultra-high molecular and stave off further degeneration. These include weight polyethylene plug. microfracture, autologous chondrocyte trans- It also has been previously shown [1] that plantation, mosaicplasty and most recently tis- a 90%-water PAMPS/PAAm DN-gel approach- sue-engineered constructs [15, 19, 20, 22]. These es the stiffness of articular cartilage. It also has repair techniques focus on repairing the cartilage attractive surgery-related attachment proper- structure. Unfortunately most of these methods ties—e.g. it can be sutured or can be attached do not result in cartilage with initial local me- to tissue with cyanoacrylate tissue adhesives. chanical properties (e.g., stiffness, strength) at However, this 90% water PAMPS/PAAm DN- the time of implantation that are even remotely gel is still not as stiff as native articular cartilage. similar to normal articular cartilage. Another Accordingly, to improve the possibility of using drawback of most of these methods is that the DN-gels as a cartilage repair material, PAMPS/ rehabilitation period takes several months up to PAAm DN-gel water content was adapted in 1-2 years in order to establish repaired tissue ca- this study to approach dynamic stiffness proper- pable of bearing cyclic impact loads of the knee ties of normal articular cartilage. of the magnitude and frequency associated with normal daily activity [9, 11, 12, 23]. From a patient’s point of view, these repairing techniques MATERIALS still need improvement. Therefore our research focuses on the PAMPS/PAAm DN-gels with lower water content: replacement of cartilage function so energy In order to fine-tune the biomechanical proper- distribution and dissipation in the surround- ties of the DN-gels, the molecular ratios of both ing cartilage will remain similar and further the first and the second network had to be varied degeneration of cartilage is avoided. The repair accordingly in order to produce gels with differ- material studied here is an example of a double- ent water contents. The three PAMPS/PAAm network hydrogel (DN-gels) developed by Gong DN-gels produced were determined to have water et al. [2, 4, 13, 14, 29]. It is based on a highly contents of 90.9%, 86.5%, and 61.3% (see Table crosslinked first polymer, PAMPS—i.e. poly(2- 5.1). The dimensions of the DN-gel specimens

83 Table 5.1: Preparing ratios and water content of the three double-network hydrogels Components: first network Components: second network DN-gel Monomer Cross- UVI Monomer Cross- UVI Water linker linker PAMPS/ AMPS MBAA o.1 AAm MBAA 0.03 90.9% PAAm 90% 1 mol/l 4 mol% mol% 2 mol/l 0.01 mol% mol% PAMPS/ AMPS MBAA 0.6 AAm - 0.01 86.5% PAAm 86% 1 mol/l 4 mol% mol% 2 mol/l 0.01 mol% mol% PAMPS/ AMPS MBAA 0.1 AAm MBAA 0.01 90.0% PAAm 61% 1 mol/l 4 mol% mol% 2 mol/l 0.01 mol% mol% Note: Prepared from the following: AMPS: 2-acrylamido-2-methylpropanesulfonic acid; AAM: acrylamide; MBBA: N,N`-methylenebisacrylamide; UVI: ultraviolet light initiator

were 20x10x3 mm. The method used to produce als. The dynamic stiffness parameters of porovis- these various DN-gels is described elsewhere [4]. coelastic materials, e.g. cartilage, are even more After producing the DN-gels, they were shipped strain rate dependent since the extent to which to Basel in normal saline and stored at 4 to 6 °C water is forced from the material also depends before testing at room temperature. on strain rate and time. This is seen in the behaviour of cartilage under two different loading Swine cartilage specimens: Cylindrical osteochon- regimes—(a) the sudden transient deformations dral plugs of 7.6 mm in diameter were harvested which occur during gait and are too brief to from the knee of 10-month-old swine using a force water out of the tissue due to its low per- standard diamond core-drill designed for mosa- meability, and (b) the slow quasi-cyclic deforma- icplasty (Synthes, Oberdorf, Switzerland). Plugs tions which cause fluid to move in and out of were harvested from the lateral condyles and cartilage and thus provide a means for nutrition. kept wet with phosphate-buffered saline (PBS) Therefore, both a Fast Impact Mode and a Slow prior to and during testing. Sinusoidal Mode test method were developed and are used by the authors working in Basel. For both test modes, the dynamic modulus can METHODS be calculated as described by Wirz et al. [28] and Kren et al. [7]. The loss angle can be calculated Mechanical testing directly from the time lag of the displacement curve relative to the load curve. Two micro-indentation methods were used as previously described [1, 21] to determine the dy- Fast impact (FI) mode namic stiffness parameters (dynamic modulus E* To simulate the impact velocity in normal hu- and loss angle d) of cartilage, meniscus and pos- man gait, a fast impact micro-indentation in- sible implant materials. The dynamic modulus is strument was used. This is a modified version of a measure of the deformation resistance of a ma- an instrument developed at the Minsk Institute terial. The loss angle is a measure of the energy of Physics [7]. A pendulum-mounted spherical dissipation. If a cyclic load is applied to a vis- indenter (diameter: 1.0 mm; 1.9 g) falls down coelastic material, the time to maximum strain on the specimen under gravitational force. The will lag the time to maximum stress [8]. motion of the indenter is captured electromag- Articular cartilage is a structure, but as a netically during indentation and rebound. The first approximation it can be treated as a material duration of impact was 1-2 ms and the initial for purposes of assessing mechanical properties impact velocity ~0.3 m/s. On each specimen, and comparing them with those of true materi- 10 replicate FI measurements were performed

84 on the same spot at ~20 s time intervals. Resul- tant E* and d were calculated for each impact and then each set was averaged to get one set of specimen values.

Slow sinusoidal (SS) mode To simulate more static loading patterns of human cartilage, a Synergie 100 MTS mechanical testing instrument was used to perform slow sinusoidal micro-indentations. A spherical indenter (diameter 1.0 mm) was moved sinusoidally with a frequency of 0.1Hz under computer software control of displacement. Indentation was performed to a depth of ~0.05 mm (SS-0.05) and ~0.1 mm (SS-0.1), with a maximum speed of ~0.015 and ~0.03 m/s. The same specimens were measured as in FI mode. On each specimen, 3 replicate SS measurements were performed at intervals of ~1 min on the same spot. Resultant E* and d were averaged to get one set of specimen values.

Statistics

A Lilliefors test was used to determine whether the E* and d data sets were normally distributed. If data were normally distributed, a two-sample t-test was performed with a=0.05, otherwise a Wilcoxon rank sum test would be performed. Statistical analysis was accomplished using R (R Foundation for Statistical Computing, Austria).

Figure 5.2: Calculated dynamic modulus (E*; up) and loss angle (d; bottom) of swine cartilage, PAMPS/PAAm61%, PAMPS/PAAm87% and PAMPS/PAAm91% in the different test modes (FI, SS-0.1 and SS-0.05). Dashed horizontal lines are average swine cartilage data values added for comparison. Solid horizontal lines indicate a significant difference with p<0.05. E* is significantly higher in FI-mode compared to both SS-modes in all gels. In all gels there is a significant difference between all test modes in d.

RESULTS

In FI-mode, force rose more rapidly with displacement than in SS-mode (Figure 5.1). This indicates a higher stiffness of the DN-gels in FI- mode. The force-displacement slopes in SS-0.05 Figure 5.1: Typical Force-displacement curves of the and SS-0.1 modes were comparable, which in- PAMPS/PAAm 90.9% in FI-mode (black), in SS-0.1- dicates a similar stiffness in these test methods. mode (dark grey) and in SS-0.05-mode (light grey) The data sets were all normally distrib-

85 uted and therefore a two-sample t-test was used. Since the cross-linked polymer structures are The calculated E* was significantly higher in FI- themselves viscoelastic, a higher concentration mode compared to SS-mode for all three DN- of polymer could be expected to dissipate more gels tested. PAMPS/PAAm 61% had a higher E* energy. However, the ratio between the two in all test modes compared to PAMPS/PAAm polymers was necessarily different among the 86% and PAMPS/PAAm 90%. In FI-mode three DN-gels, and this perhaps makes trying PAMPS/PAAm 86% had a higher E* than to explain the results on a basis of water content PAMPS/PAAm 90% (Figure 5.2). alone too simplistic. The d was significantly higher in SS- mode in all gels compared to FI-mode. In SS- Dynamic stiffness of DN-gels compared to 0.1-mode the d was lower than in SS-0.05-mode cartilage in all gels. In SS-mode no difference was found in d among the three different gels. However, in The results show that it was possible to bring the FI-mode the d was higher in PAMPS/PAAm 61% dynamic stiffness of the PAMPS/PAAm DN- compared to PAMPS/PAAm 86% and PAMPS/ gels closer to normal cartilage by modifying the PAAm 90%, and higher in PAMPS/PAAm 86% structures in a way which allowed lower water than in PAMPS/PAAm 90%. content. As shown in Figure 5.2, the PAMPS/ Cartilage had a higher E* in FI-mode PAAm 61% was about 1.5-2 times stiffer than compared to SS-mode. In FI-mode E* values of cartilage in SS-mode, but ~30% less stiff in FI- cartilage were higher compared to all DN-gels mode. Compared to tissue engineered constructs, (Figure 5.2). In SS-mode E* of cartilage was sig- which are only up to 10% of cartilage stiffness nificantly higher compared to PAMPS/PAAm [22] and autologous chondrocyte transplanta- 86% and PAMPS/PAAm 90% and lower than tion, which is about 60% of cartilage stiffness PAMPS/PAAm 61%. a year after surgery [19] initial repair stiffness is The d of cartilage was lower in FI-mode closer to native cartilage. On the other hand, the compared to SS-mode. In SS-0.1-mode the d of PAMPS/PAAm 61% in FI-mode was ~60% d was lower than in SS-0.05-mode. In all test lower than that of cartilage and ~70% lower in modes the d of cartilage was higher than all DN- SS-mode. PAMPS/PAAm 86% and 90% had a gel tested. lower E* and a lower d compared to cartilage in all test modes. The crucial dynamic mechanical differ- DISCUSSION ence between all three of the PAMPS/PAAm DN-gels and normal cartilage is that all three DN-gel water content effects had much lower loss angles (d) than cartilage in both test modes. This means that compared The dynamic modulus, E*, increased as the water to cartilage, these gels are less able to dissipate content decreased in all test modes. A possible ex- energy. Also, due to its higher d, the E* of car- planation is that if the concentration of polymer tilage is more strain rate dependent than that of is higher, there is more structure per unit volume DN-gels. Therefore, by adjusting water content to resist deformation. Conversely, the DN-gel d in the manner done here, the low loss angles did not change as a function of water content of PAMPS/PAAm DN-gels means that their E* in SS mode. In our previous work we suggest values could not be made similar to cartilage that the d in SS-mode is mainly due to water at both strain rates—i.e. during both fast im- movement within the structure [21]. The results pact (FI) and slow sinusoidal (SS) testing. For presented here imply that the water movement is example, if one “tunes” the DN-gel value of E* similar in all PAMPS/PAAm DN-gels and the to be similar to cartilage in FI, one is left with a polymer-water-ratio does not change the abil- gel which has a higher E* in SS mode. The con- ity of a given deformation to move water within sequence of this difference in mechanical prop- the structure. But in addition, in FI-mode, the erties with the surrounding tissue can only be d increases with decreasing water content. This speculated upon. However, the difference would means that an increase in polymer concentra- be much lower compared to the same properties tion increases the d at high deformation rates. produced using tissue repair techniques already

86 in use. [30] non-porous plugs of a highly similar DN- One possible structural reason that the d gel have been shown to foster cartilage forma- of all the PAMPS/PAAm DN-gels is low com- tion in a rabbit osteochondral defect model. pared to cartilage may be because both compo- Although these DN-gels look promising nents of the polymer structure are highly chemi- as a cartilage repair material, in this study only cally crosslinked. This crosslinking reduces the their dynamic stiffness was investigated. Be- possibility of sliding between the polymer chains fore these DN-gels can be used in clinic other during deformation and thus reduces the fric- aspects should be investigated mainly focussed tional dissipation of energy. Another possible on the biocompatibility, such as immunological cause of lower energy dissipation compared to reactions, absorption and integration to the sur- cartilage might be less movement of water ei- rounding tissues. ther within the DN-gel structure or out of the structure during deformation. Water can be forced out of cartilage by static loads [10] but CONCLUSIONS similar loads do not result in forcing water out of PAMPS/PAAm DN-gels. Gong et al. [4] showed In all three of the PAMPS/PAAm DN-gels, the that after deformation to ~20% of the original d increases with decreasing deformation rate in thickness; still no water is squeezed out of the SS mode compared to FI mode. This is what is structure. In other words, the water within the expected for viscoelastic materials [8, 18]. Al- DN-gel-structure is more highly trapped com- though the DN-gels thus show normal visco- pared to cartilage. elastic behaviour with respect strain rate and d, These results show that cartilage-like dy- they do not do so with respect to E*. For normal namic stiffness can be achieved by these com- viscoelastic materials, E* increases with increas- positional changes; however, it is unknown how ing strain rate [8, 18] - and is thus higher in FI- it affects other important mechanical properties. mode compared to SS-mode. However, for the Therefore the authors plan to investigate other DN-gels no difference in E* was found between mechanical properties, such as strength, fatigue the two SS-modes, even though the deformation and tear resistance in further study. These DN- rate was doubled for SS-0.1 compared to SS-0.05 gels have already been shown to be superior to (~ 0.03 vs. 0.015 m/s). conventional gels in simulated use friction and Biomechanically these DN-gels look wear tests [29]. promising as potential cartilage repair materials. These DN-gels have potential for clini- However, other properties, such as fixation sta- cal use. They are easy to sterilise since autoclav- bility and mechanical performance in vivo have ing has been shown not to affect their structures. to be explored. They can be trimmed or produced in desired shapes and the surgical fixation, e.g. with sutures or tissue adhesive, capability has proven to ACKNOWLEDGEMENTS be substantial. Besides this, if the gel is created with pore size are large enough, cell infiltration Hardy & Otto Frey-Zünd Stiftung for the sup- is likely possible, to assure integration to the sur- port and H.J. Wyss for the funds donated to rounding tissue. Also, as previously mentioned University Basel.

87 REFERENCES tional Course Lectures 56: 457-468, 2007. 13. Nakajima T, Furukawa H, Tanaka Y, Ku- 1. Arnold MP, Daniels AU, Ronken S, Ardura rokawa T, Osada Y, Gong JP. True Chemi- Garcia H, Friederich NF, Kurokawa T, Gong cal Structure of Double Network Hydrogels. JP, Wirz D. Acrylamide Polymer Double- Macromolecules 42(6):2184-2189, 2009. Network Hydrogels: Candidate Cartilage 14. Nakayama A, Kakugo A, Gong JP, Osada Y, Repair Materials with Cartilage-Like Dy- Takai M, Erata T, Kawano S. High Mechani- namic Stiffness and Attractive Surgery-Re- cal Strength Double-Network Hydrogel with lated Attachment Mechanics. Cartilage, 2011 Bacterial Cellulose. Advanced Functional Ma- 2. Azuma C, Yasuda K, Tanabe Y, Taniguro H, terials 14(11): 1124-1128, 2004. Kanaya F, Nakayama A, Chen YM, Gong 15. Negrin L, Kutscha-Lissberg F, Gartlehner G, JP, Osada Y (2006) Biodegradation of high- Vecsei V. Clinical outcome after microfrac- toughness double network hydrogels as po- ture of the knee: a meta-analysis of before/ tential materials for artificial cartilage. Jour- after-data of controlled studies. International nal of Biomedical Material Research Part A 81: Orthopeadics, 2011. 373-380, 2006. 16. Nehrer S, Minas T. Treatment of Articular 3. Buckwalter J, Mankin H. Articular carti- Cartilage Defects. Investigative Radiology lage repair and transplantation. Arthritis and 35(10): 639-646, 2000. Rheumatology 41(8): 1331–1342, 1998. 17. Newman AP. Articular cartilage repair. The 4. Gong JP, Katsuyama Y, Kurokawa T, Osada American Journal of Sports Medicine 26(2): Y. Double-Network Hydrogels with Extreme- 309-324, 1998. ly High Mechanical Strength. Advanced Ma- 18. Park S, Hung C, Ateshian G. Mechanical terials 15(14): 1155-1158, 2003. response of bovine articular cartilage under 5. Hunter W. Of the structure and diseases of dynamic unconfined compression loading at articulating cartilages. Philosophical Transac- physiological stress levels. Osteoarthritis and tions of The Royal Society 42: 514-521, 1743. Cartilage 12: 65-73, 2004. 6. Hunziker EB. Articular cartilage repair: ba- 19. Peterson L, Brittberg M, Kiviranta I, Åker- sic science and clinical progress. A review of lund EL, Lindahl A. Autologous chondro- the current status and prospects. Osteoarthri- cyte transplantation: Biomechanics and tis and Cartilage 10(6): 432-463, 2002. long-term durability. The American Journal of 7. Kren AP, Rudnitskii VA, Deikun IG. Deter- Sports Medicine 30(1): 2-12, 2002. mining the viscoelastic parameters of vulca- 20. Robert H (2011) Chondral repair of the knee nisates by the dynamic indentation method joint using mosaicplasty. Orthopeadics and using a non-linear deformation model. In- Traumatology, Surgery and Research 97(4): ternational Polymer Science and Technol ogy 418-29, 2011. 32(7): 19-23, 2005. 21. Ronken S, Arnold MP, Ardura García H, 8. Lakes RS. Viscoelastic solids. CRC Press Jeger A, Daniels AU, Wirz D. A comparison LLC, Boca Raton, 1999. of healthy human and swine articular car- 9. Mandelbaum BR, Browne JE, Fu F, Micheli tilage dynamic indentation mechanics. Bio- L, Mosely JB, Erggelet C, Minas T, Peterson mechanics and Modeling in Mechanobiology, L. Articular Cartilage Lesions of the Knee. 2011. The American Journal of Sports Medicine 22. Santoro R, Olivares AL, Brans G, Wirz D, 26(6): 853-861, 1998. Longinotti C, Lacroix D, Martin I, Wendt D. 10. Mankin H, Mow V, Buckwalter JA, Iannotti Bioreactor based engineering of large-scale J, Ratcliffe A. Articular cartilage structure, human cartilage grafts for joint resurfacing. composition, and function. In: Buckwalter Biomaterials 31(34): 8946-8952, 2010. JA, Orthopaedic basic science: biology and 23. Saris DBF, Vanlauwe J, Victor J, Haspl M, biomechanics of the musculoskeletal system, Bohnsack M, Fortems Y, Vandekerckhove B, American Academy of Orthopaedic Surgeons, Almqvist KF, Claes T, Handelberg F, Lagae Rosemont, IL, 443–470, 2000. K, van der Bauwhede J, Vandenneucker H, 11. Minas T. The role of cartilage repair tech- Yang KG, Jelic M, Verdonk R, Veulemans niques, including chondrocyte transplanta- N, Bellemans J, Luyten FP. Characterized tion, in focal chondral knee damage. Instruc- chondrocyte implantation results in better tional Course Lectures 48: 629-643, 1999. structural repair when treating symptomatic 12. Mithoefer K, Scopp JM, Mandelbaum BR. cartilage defects of the knee in a randomized Articular cartilage repair in athletes. Instruc- controlled trial versus microfracture. The

88 American Journal of Sports Medicine 36(2): 2007. 235-246, 2008. 28. Wirz D, Kohler K, Keller B, Göpfert B, Hu- 24. Slauterbeck JR, Kousa P, Clifton BC, Naud detz D, Daniels AU. Dynamic stiffness of S, Tourville TW, Johnson RJ, Beynnon BD. articular cartilage by single impact micro- Geographic mapping of meniscus and carti- indentation (SIMI). Journal of Biomechanics lage lesions associated with anterior cruciate 41(Suppl 1) S172, 2008. ligament injuries. The Journal of Bone and 29. Yasuda K, Gong JP, Katsuyama Y, Nakayama Joint Surgery 91: 2094-2103, 2009. A, Tanabe Y, Kondo E, Ueno M, Osada Y. 25. Smith GD, Knutsen G, Richardson JB. A Biomechanical properties of high-toughness clinical review of cartilage repair techniques. double network hydrogels. Biomaterials The Journal of Bone and Joint Surgery 87(4): 26(21): 4468-4475, 2005. 445-449, 2005. 30. Yasuda K, Kitamura N, Gong JP, Arakaki K, 26. Temenoff JS, Mikos AG. Review: tissue engi- Kwon HJ, Onodera S, Chen YM, Kurokawa neering for regeneration of articular cartilage. T, Kanaya F, Ohmiya Y, Osada Y. A novel Biomaterials 21(5):431-440, 2000. double-network hydrogel induces spontane- 27. Widuchowski W, Widuchowski J, Trzaska T. ous articular cartilage regeneration in vivo in Articular cartilage defects: study of 25,124 a large osteochondral defect. Macromolecular knee arthroscopies. Knee 14(3): 177-182, Bioscience 9(4): 307-316, 2009.

6 Discussion and Outlook

DISCUSSION a benefit for the quality of life for many people around the world. Economic and social impact of OA Mathematical models Last decade was chosen by the World Health Or- ganisation (WHO) as the Bone and Joint decade Different types of models are used in order to de- to get more attention for musculoskeletal describe cartilage behaviour. None of these models ceases. One of the most common forms of joint describe cartilage perfectly; they all have their diseases is osteoarthritis (OA). As the risk of get- advantages and disadvantages. All of them are ting OA increases with age (70% of >65 year old partly based on (wrong) assumptions and only people suffer from OA) and the age of the world an approach of real cartilage. One should be population increases the prevalence of OA will aware of the disadvantages of the model used so rise in the next years [4, 10, 21]. no wrong conclusions are drawn when using a Large untreated cartilage lesions, e.g. certain model. However, when a perfect model due to traumatic accidents, will inevitably lead for cartilage would exist, the question remains to secondary OA [3]. Therefore an optimal treat- how valuable it would be. Cartilage is a living ment is needed to prevent or delay OA. Several tissue, its structure and ECM components vary techniques to treat cartilage lesions are already between different locations and needs of the hu- applied in clinic [7, 14, 17, 22]. However, there is man body. The model would most probably not no satisfying treatment found yet. If a large por- be appropriate anymore when testing OA carti- tion of the cartilage is damaged, these cartilage lage, since in OA cartilage the quantity of ECM repair techniques are insufficient and the only components and the structure have changed. possibility is to implant a total or partial joint Also replacement materials do not necessarily replacement prosthesis. Such an implant costs a have a similar structure and thus a model de- lot of money, with the operation and planning, scribing cartilage will most likely not apply to the stay in the hospital and rehabilitation after- these materials. Here the question rises whether wards. However, if those people would not get a the model still can be used and, if not, whether prosthesis and therefore would be disabled and results from two different models can still be would not be able to live independently due to compared. their disability, the costs of health care, help at Mathematical models are developed in home or living in an (elderly) home would after a order to being able to calculate cartilage stiffness few months already be more expensive than the out of indentation data. This is important for the joint replacement. understanding of cartilage behaviour in different Due to the higher demand of people who situations and for evaluation of stiffness distribu- want to be able to be active without suffering tion on a joint. To do so there is the need for a from pain also when they are getting older, the relative stiffness value. Thus a model which can number of people getting a prosthesis increases discriminate between cartilage stiffness on dif- and the average age when they get it decreases. ferent locations and between healthy and degen- The life time of a prosthesis is about 10-15 years, erated cartilage will be prefered. depending on the joint in which it is implanted. As the age of people getting a prosthesis decreas- Cartilage stiffness and energy dissipation es, more people will possibly need a replacement. People suffering from early OA get informed Although cartilage has a low capacity for self- about risk factors and go a physical therapist. If repair it has the capability to adapt to its needs. the OA is worsened nothing can be done until it When cartilage is unused, i.e. no loading is pres- has become so bad that total or partial joint re- ent, the chondrocytes will not get nutrition any- placement is the only option. If a good cartilage more and eventually the cartilage will degenerate repair or treatment of OA would be available or and disappear. But it also works the other way possible, the prevalence of a prosthesis and thus around: at locations where a lot of (shear) load- also a replacement prosthesis would decrease and ing is present, a cartilage layer can form. This is the age when people get one would increase. This seen in people with a sesamoid bone, where a would influence health costs positively and have piece of bone with a cartilage layer is formed in

95 a tendon. rates compared to fast deformations, since water Articular cartilage has varying thick- movement is more apparent in slow deformation. ness and stiffness, even within one joint. It is Besides this, since the energy dissipation of the adapted to the need of a specific location and cartilage is more uniform, small changes might has developed into the ultimate design for a spe- be detected earlier when looking at energy dis- cific person. At locations with high compressive sipation. forces the cartilage layer can, and maybe has to, Once, when we are able to diagnose OA be thicker compared to locations where mainly in early stages, we will be able to apply tissue shear forces act, since fluid flow is more appar- engineered cartilage methods like autologous ent in compressive loading situations. The thick- chondrocyte transplantation [5, 13] in earlier ness, stiffness and energy dissipation of the carti- stages and better result may be expected. With lage plays a major role in stress distribution and the possibility of early detection of OA we will transmission to the bone. Besides this, also the learn more about the natural history of OA. We strain rate-dependency defines how much ener- also will be able to treat persons, where early OA gy is stored or dissipated. In chapter 3, we have is detected accidentally during an arthroscopic shown that differences in mechanical behaviour intervention e.g. because of a little meniscal tear. might be more apparent in certain loading fre- Of course, it is not feasible to test everyone ev- quencies and that not only stiffness of cartilage ery few years in order to be able to detect early is important but also energy dissipation. In order OA, but it may be expected that the results of to fully understand cartilage behaviour it should non-interventional examinations like CT or always be tested in more than one loading fre- MRI will show a correlation to mechanically quency and not only stiffness, but also the en- detected early OA [16]. A positive correlation ergy loss should be determined. is found in bone mineralization measured with CT and bone strength [12, 24]. In a preliminary Osteoarthritis diagnosis study we found an inverse correlation between The mechanical properties of cartilage change cartilage stiffness and bone strength [18]. Fur- when it degrades. In an advanced degradation ther, it is known that there is a strong correla- stage this can be measured, whereas in an early tion between OA and trauma in the knee and stage these differences may lie within the nor- ankle joint and obesity [2, 6, 11]. It also has been mal variation of mechanical properties. Brown shown that with MRI it is possible to predict et al. [1] found a decrease in indentation stiffness how high the chances are to develop severe OA for degraded cartilage, but only 17% lay outside if early OA is already diagnosed [15]. When it is normal variation and therefore it cannot be used known which people are developing OA, treat- for detection. Stolz et al. [20] found a difference ment should become possible to slow down or in nano-stiffness between different grades of OA. stop cartilage degradation or preferably even However, due to the small number of patients recover the damages which already took place. and the increase in stiffness with age and de- Of course this works vice versa as well: does it crease in OA it is uncertain whether this could make sense to find a treatment for (early) OA if be applied in the clinic. The advantage of nano- it is not possible to detect it before it is too late? stiffness measurements is that the properties of Therefore it is crucial that early detection of OA the collagen matrix and the proteoglycan gel can is made possible. be determined separately [9]. A change in one of the two components (collagen matrix or proteo- Replacement materials glycan gel) might be detected earlier when not determined simultaneously. Several cartilage repair strategies are being used The proteoglycan content increases in or developed to treat cartilage lesions. The over- OA, which changes the capability of the cartilage all goal of these strategies is to restore the func- to hold water. These (small) differences might be tion of the joint, in order to make pain-free and detected in certain loading frequencies and not non-restricted movement possible. To achieve in others. When the capacity of holding the wa- normal joint function, different strategies are ap- ter in the collagen changes, this is more likely plied. Most of them are based on restoring the detected earlier in tests using slow deformation cartilage structure with its original components.

96 In osteochondral allografting and mosaicplasty, mind is cartilage nutrition. Cartilage receives its plugs are inserted in the defects ensuring that nutrition from fluid flowing through the tissue at least part of the defect is filled with cartilage. and this should still be possible also when part In chondrocyte transplantation and microfrac- of the cartilage is replaced. Otherwise the chon- turing, the defect is filled with cells and liquid, drocytes in the cartilage will not survive and the which should regenerate a cartilage layer. How- cartilage will eventually degrade. ever, a hyaline cartilage layer is not always regen- Patients heterogeneity also plays a role erated. Also the approach of tissue-engineered in cartilage repair. It has been seen that micro- constructs is based on making a tissue which has fracture leads to better results in young patients cartilage-like structure and components. compared to older patients [19]. Besides this, The strategies applied in the clinic are the demand of the patients might vary. Other based on filling of the defect with blood, cells or qualifications have to be met in young patients osteochondral plugs. However, the question rises who want to continue playing sports, compared whether filling of the defect is necessary. Yasuda to (older) patients who want to stay mobile for et al. [23] showed that it was possible to regener- some years. It might then be worth waiting a ate cartilage on top of a DN-gel implanted in few more months if the result afterwards is more an osteochondral defect in rabbits. However, the satisfactory. A possible solution could be that result was dependent on how much space was young patients with still highly active cells get a left on top of the implanted DN-gel. No hyaline cell-based repair, where cartilage regeneration is cartilage was regenerated when the vacant space still possible, while older patients get an instant was deeper than 3.5 mm and the best results ready construct or material implanted, in order were found when the vacant space was 1.5 - 2.4 to decrease the time that they are not mobile and mm deep. delay or prevent the need of a prosthesis. What about the mechanical properties of In the end, the question stays: how good the repair? Isn’t it more important that the func- should the repair be? The optimal solution might tion of cartilage is restored? Cartilage absorbs be a personalized patient-specific repair. In and distributes loads which reduce peak loads young, healthy patients with cartilage lesions the in the bone. If the mechanical properties of the best long-term treatment might be a cell-based repair are different, loads will be distributed dif- or tissue engineered construct using the patient’s ferently in the joint. If the material is too stiff, own cells. In older patients, where it is less likely the load will become focused in a small region, that their cells can regenerate a cartilage layer whereas when stiffness is too low, load will be with sufficient mechanical properties, DN-gels transmitted to the bone. If the energy dissipa- might be the best treatment to prevent or delay tion is too low, too much energy will be stored, joint replacement. which might increase peak loads, whereas when too much energy is dissipated, recovery is slower which might increase loads in the surrounding CONCLUSION cartilage. So the mechanical properties are an important aspect of a possible cartilage repair. Cartilage is an extraordinary tissue with a com- However, up to now it is not known how similar plex structure. How it behaves under different the mechanics of a repair should be compared to loading conditions can be described by deter- cartilage, especially since there is large variation mining its stiffness and energy dissipation. Due between individuals and even within one joint. to the complexity of the cartilage, finding a re- The replacement material should not change placement material is a challenge. DN-gels have loading too much in the surrounding cartilage a high potential of becoming a cartilage repair, nor in the underlying bone, since that might in- especially in older patients, whereas in young duce cartilage degradation, bone loss, or bone patients with cartilage defects tissue engineered fracture. constructs might be the future. Another aspect which should be bore in

97 OUTLOOK it should be determined what key parameters are to mimick, i.e. structure, stiffness, content, fluid In this thesis it is shown that different inden- flow, energy dissipation and how close to native tation protocols lead to different results of me- these parameters should be mimicked. Is half as chanical properties of articular cartilage. Besides stiff, stiff enough to prevent further damage? Or that, it is shown that not only the stiffness is a would half as much collagen be enough to bear parameter, but also the energy dissipation is a key loading? Do all off the cartilage ECM contents parameter which indicates cartilage behaviour. have to be replaced in the same proportions? In order to understand cartilage behaviour and Most probably the answers on these questions to make early diagnosis of osteoarthritis possible, are patient specific. Young patients have a high- various aspects have to be taken into account in er demand and want the repair to hold longer future investigations. First of all one should be and withstand higher loads compared to older able to measure the mechanical properties in an patients. The minimum requirements to prevent intact joint e.g. during an arthroscopy. So results further damage should be investigated to lead to from the laboratory can be compared with results an as good as possible result. from the operation rooms and vice versa. Sec- Not only cartilage has the ability to re- ond, research on the changes in cartilage in OA model, also bone is a dynamic tissue which adapts should not only focus on cartilage stiffness, but to the needs of the body. If the bone has to bear also energy dissipation to ensure that not only high stresses it will strengthen and if it is unload- one aspect of cartilage behaviour is examined. ed, the result is bone loss. An extreme example Cartilage pieces with the same stiffness, but a of bone loss due to a decrease in stress is seen in different capacity to dissipate energy will behave astronauts who went to outer space. Depending differently under the same loading conditions. on cartilage thickness, stiffness and energy loss This will not only affect load distribution within more or less load is transferred to the underlying the cartilage, but also the load transmission to subchondral bone. A preliminary study showed the bone. Third, cartilage stiffness and energy an inverse correlation between cartilage stiffness dissipation and their changes in OA should be and subchondral bone strength in three human determined in several loading frequencies, since patellae [18]. No correlation was found between cartilage behaviour is extremely strain-rate-de- bone strength and cartilage energy dissipation, pendent. Changes in OA cartilage might be more nor between cartilage stiffness and energy dis- apparent in certain loading frequencies. Fourth, sipation. More human patellae will be measured it should be determined whether changes in OA and also cartilage thickness and grade of OA will can be detected more easily by measuring just be determined in order to see how these factors the collagen network or the proteoglycan matrix influence subchondral bone strength. It has been with nano-measurements. This would increase shown that subchondral bone remodels in OA; understanding of cartilage behaviour and the bone resorption increases in early OA and bone changes occurring in OA. accretes in later stages of OA [8]. To decrease the In this thesis it is shown that double- numbers of factors which play a major role in network hydrogels (DN-gels) are a promising the old human patellae, research should also be material for cartilage repair. They can be pro- performed on young patellae to be able to rule duced with different single networks to vary out the influence of cartilage degradation. Since their properties, such as biocompatibility and young human patellae are not commonly avail- bio-absorbability. It is shown here that the dy- able, animal patellae can be used for this study. namic stiffness values of those DN-gels can Of course it would also be important to know be tuned to approach those of native cartilage. whether the correlations found are also apparent However, to develop a cartilage repair material, on other locations and other joints.

98 REFERENCES 14: 12-17, 2010. 11. Manek NJ, Hart D, Spector TD, MacGregor 1. Brown CP, Crawford RW, Oloyede A. In- AJ. The association of body mass index and dentation stiffness does not discriminate osteoarthritis of the knee joint: an examina- between normal and degraded articular car- tion of genetic and environmental influenc- tilage. Clinical Biomechanics 22: 843-848, es. Arthritis and Rheumatism 48: 1024-1029, 2007. 2003. 2. Dawson J, Juszczak E, Thorogood M, Marks 12. Müller-Gerbl M, Putz R, Kenn R. Demon- S, Dodd C, Fitzpatrick R. An investigation stration of subchondral bone density patterns of risk factors for symptomatic osteoarthritis by three-dimensional CT osteoabsorptiom- of the knee in women using a life course ap- etry as a noninvasive method for in vivo as- proach. Journal of Epidemiology and Commu- sessment of individual long-term stresses in nity Health 57: 823–830, 2003. joints. Journal of Bone and Mineral Research 7 3. Gaissmaier C, Fritz J, Schewe B, Weise K, Suppl 2: S411-418, 1992. Mollenhauer J, Aicher W. Cartilage Defects: 13. Peterson L, Minas T, Brittberg M, Lindahl A. Epidemiology and Natural History. Osteosyn- Treatment of osteochondritis dissecans of the thesis and Trauma Care 14: 188-194, 2006. knee with autologous chondrocyte transplan- 4. Horan FT. The bone and joint decade 2000 tation: results at two to ten years. The Journal to 2010. The Journal of bone and joint surgery of Bone and Joint Surgery 85-A Suppl: 17-24, 93: 143-144, 2011. 2003. 5. Jakob RP. AMIC technique for cartilage 14. Peterson L, Minas T, Brittberg M, Nilsson repair, a single-step surgical intervention as A, Sjögren-Jansson E, Lindahl A. Two- to compared to other methods. European Cells 9-year outcome after autologous chondrocyte and Materials 12, 2006. transplantation of the knee. Clinical Ortho- 6. Jordan JM, Helmick CG, Renner JB, Luta G, paedics and Related Research: 212-234, 2000. Dragomir AD, Woodard J, Fang F, Schwartz 15. Raynauld JF, Martel-Pelletier J, Berthiaume TA, Abbate LM, Callahan LF, Kalsbeek WD, MJ, Beaudoin G, Choquette D, Haraoui B, Hochberg MC. Prevalence of knee symptoms Tannenbaum H, Meyer JM, Beary JF, Cline and radiographic and symptomatic knee os- GA, Pelletier JF. Long term evaluation of teoarthritis in African Americans and Cau- disease progression through the quantitative casians: the Johnston County Osteoarthri- magnetic resonance imaging of symptomatic tis Project. The Journal of Rheumatology 34: knee osteoarthritis patients: correlation with 172–180, 2007. clinical symptoms and radiographic changes. 7. Knutsen G, Engebretsen L, Ludvigsen TC, Arthritis Research and Therapy 8(1): R21 Drogset JO, Grøntvedt T, Solheim E, Strand 16. Recht MP, Goodwin DW, Winalski CS, T, Roberts S, Isaksen V, Johansen O. Autolo- White LM. MRI of articular cartilage: re- gous chondrocyte implantation compared visiting current status and future directions. with microfracture in the knee. A random- American Journal of Roentgenology 185: 899- ized trial. The Journal of Bone and Joint Sur- 914, 2005. gery 86-A: 455-464, 2004. 17. Robert H. Chondral repair of the knee joint 8. Kwan Tat S, Lajeunesse D, Pelletier JP, Mar- using mosaicplasty. Orthopaedics and Trau- tel-Pelletier J. Targeting subchondral bone for matology, Surgery and Research 97: 418-429, treating osteoarthritis: what is the evidence? 2011. Best Practice and Research Clinical Rheumatol- 18. Ronken S, Hoechel S, Wirz D, Müller-Gerbl ogy 24: 51-70, 2010 M. Cartilage stiffness and subchondral bone 9. Loparic M, Wirz D, Daniels AU, Raiteri plate strength of the human patella. 18th R, Vanlandingham MR, Guex G, Martin Congress of the European Society of Biome- I, Aebi U, Stolz M. Micro- and nanome- chanics, Lisbon, Portugal, July 1 - 4. 2012. chanical analysis of articular cartilage by 19. Steadman JR, Briggs KK, Rodrigo JJ, Kocher indentation-type atomic force microscopy: MS, Gill TJ, Rodkey WG. Outcomes of mi- validation with a gel-microfiber composite. crofracture for traumatic chondral defects of Biophysical Journal 98: 2731-2740, 2010. the knee: average 11-year follow-up. Arthros- 10. Mandzuk LL, McMillan DE, Bohm ER. copy 19: 477-484, 2003. The Bone and Joint Decade in Canada: A 20. Stolz M, Gottardi R, Raiteri R, Miot S, Mar- look back and a look forward. International tin I, Imer R, Staufer U, Raducanu A, Düg- Journal of Orthopaedic and Trauma Nursing gelin M, Baschong W, Daniels AU, Fried-

99 erich N, Aszodi A, Aebi U. Early detection 23. Yasuda K, Kitamura N, Gong JP, Arakaki K, of aging cartilage and osteoarthritis in mice Kwon HJ, Onodera S, Chen YM, Kurokawa and patient samples using atomic force mi- T, Kanaya F, Ohmiya Y, Osada Y. A novel croscopy. Nature Nanotechnology 4: 186–192, double-network hydrogel induces spontane- 2009. ous articular cartilage regeneration in vivo in 21. WHO Scientific Group. The burden of mus- a large osteochondral defect. Macromolecular culoskeletal conditions at the start of the Bioscience 9: 307-316, 2009. new millennium: report of a WHO scientific 24. Zumstein V, Kraljević M, Wirz D, Hügli R, group. WHO, 2003. Müller-Gerbl M. Correlation between min- 22. Williams RJ, Harnly HW. Microfracture: eralization and mechanical strength of the indications, technique, and results. Instruc- subchondral bone plate of the humeral head. tional Course Lectures 56: 419-428, 2007. Journal of Shoulder and Elbow Surgery

100

Acknowledgements

Well, so this is it... A couple of years of hard work I needed them. They made it possible for me to and this is the result. But of course without the do what I wanted most and follow my dreams. help of others I would not have succeeded. Also my sisters, Manon, Esmee and Jolijn, were always there for me when I needed them. Even First of all I would like to thank Dan, Dieter now when I am not living close by anymore, I and Markus for their help. We have spent many think our bond became only stronger. We do not hours over the last few years discussing results, see each other so often anymore, but we enjoy abstracts, manuscripts and much more. Their the moments together even more. knowledge, assistance and enthusiasm motivat- I also want to thank my friends from all ed me to continue. Also the support of Niklaus over the world. Thank you for supporting me and Magdalena was crucial to being able to fin- and listening to me when I wanted to complain ish this thesis. about something. After complaining I felt re- Thanks to Niklaus and Urs I got involved lieved and had some energy to proceed study- in a very interesting project in collaboration with ing. I want to especially thank Yvette, who was Nanosurf AG. This motivated me and gave me there for me when I needed her most. Thank all the courage to move further. So hereby I would of you as well for the fun part of being a friend: like to thank my colleagues at Nanosurf and es- going for a dinner, drink or chat and always hav- pecially to those who made all of this possible. ing time for a good laugh. My volleyball teams I also would like to thank my other col- were also of great help to take my mind off of my leagues who listened to me when I was frustrated thesis. I really could (and had to) think about to blow of steam. But also the non-work-related something else and afterwards I was motivated hours we have spent during lunch-breaks, Rhine to continue work again. swims, hikes or concerts. It was great fun having Special thanks go to Henk-Joost. Last you as colleagues and without you it would be year was a very tough year for both of us, with much harder to get to work every day. some wonderful moments to never forget, but Thanks also go to the WIN organising also some stressful moments, which we would team. It was a great opportunity to take a look like to forget as soon as possible, but will stay in a large company and explore some of the pos- forever. It made us stronger. He was always there sibilities after my PhD. I especially would like to support me, also when he had a tough day at to thank Leila, who was my mentor during this work. It wasn’t easy to live with me, especially on period. I have learned a lot and was able to de- days that things did not went the way I wanted. velop myself. Also thanks to the other mentees. The (weekend) trips, dinners, and other things We have shared all our experiences and learned a we did together were always lovely and gave me lot from each other. new energy to work on my thesis. Furthermore, I want to thank him for the challenge to go Last, but not least, I want to thank my family abroad. I’m still really happy with our choice go- and friends. They supported me when I had a ing to Switzerland together. Sometimes it was a hard time and made sure I had some distraction, hard time, but it opened my eyes to the world. also when I thought that I did not need it or did not had time for it. The space was limited to write a personal note to I want to thank my parents, who always all of you, but that does not mean that I do not supported me and always were there for me when value your help, support or friendship.

103

List Of Publications

Journal Publications Conference Abstracts

D. Wirz, S. Ronken, A.P. Kren, A.U. Daniels, Oral Presentations: Experimental verification of a non-linear model for computing cartilage modulus from micro- S. Ronken, M.P. Arnold, S. Hoechel, A.U. Dan- indentation data, submitted to Computational iels, M. Müller-Gerbl, D. Wirz, Mapping of dy- and Mathematical Methods in Medicine namic stiffness properties of cartilage on the human patella, 18th Congress of the European S. Ronken, M.P. Arnold, H. Ardura García, A. Society of Biomechanics, Lisbon, Portugal, July Jeger, A.U. Daniels, D. Wirz, A comparison of 1 - 4, 2012 healthy human and swine joint cartilage dynamic compression behaviour, Biomechanics S. Ronken, S. Hoechel, D. Wirz, M. Müller- and Modeling in Mechanobiology, 2011, DOI Gerbl, Cartilage stiffness and subchondral bone 10.1007/s10237-011-0338-7 strength of the human patella, 18th Congress of the European Society of Biomechanics, Lisbon, M.P. Arnold, A.U. Daniels, S. Ronken, H. Ar- Portugal, July 1 - 4, 2012 dura García, N.F. Friederich, T. Kurokawa, J.P. Gong, D. Wirz, Acrylamide polymer double- S. Ronken, M. P. Arnold, A.U. Daniels, D. Wirz, network hydrogels: candidate cartilage repair A comparison of healthy human and swine joint materials with cartilage-like dynamic stiffness cartilage dynamic compression behaviour in and attractive surgery-related attachment me- modes emulating joint function, International chanics, accepted for publication in Cartilage Society of Biomechanics, Brussels, Belgium, July 2011 3 - 7, 2011

S. Ronken, D. Wirz, A.U. Daniels, T. Kuro- S. Ronken, M. P. Arnold, A.U. Daniels, D. kawa, J.P. Gong, M.P. Arnold, Double network Wirz, Swine joint cartilage as an ex-vivo stan- acrylamide hydrogel compositions adapted to dard of comparison for human articular carti- achieve cartilage-like dynamic stiffness, Biome- lage, 7. Jahrestagung der Deutschen Gesellschaft chanics and Modeling in Mechanobiology, 2012 für Biomechanik, Murnau, Deutschland, May 19 - 21, 2011, nomination for the Young Investiga- I. Argatov, A.U. Daniels, G. Mishuris, S. Ronk- tor Award en, D. Wirz, Accounting for the thickness effect in dynamic spherical indentation of a viscoelas- S. Ronken, D. Wirz, M. Stolz, A.P. Kren, A.U. tic layer: Application to non-destructive testing Daniels, Experimental verification of a viscoelas- of articular cartilage, submitted to European tic model for computing cartilage modulus from Journal of Mechanics - A microindentation data, 17th Congress of the European Society of Biomechanics, Edinburgh, United Kingdom, July 5 - 8, 2010

105 J. Frère, S. Ronken, A.U. Daniels, B. Göpfert, S. Ronken, M. P. Arnold, A.U. Daniels, D. D. Wirz, Dynamic modulus and loss angle of Wirz, Can swine joint cartilage serve as an ex- articular cartilage in gait- and nutrition-related vivo substitute for human articular cartilage in test modes, 18th Conference of the European mechanical tests?, 5th Basel International Knee Orthopaedic Research Society, Davos, Switzer- Congress and Instructional Course, Basel, Swit- land, June 30 - July 2, 2010 zerland, May 9 - 10, 2011, best poster award

S. Ronken, M.P. Arnold, A.U. Daniels, D. Wirz, Poster presentations: Mechanical properties of fresh and Thiel’s -em balmed cartilage, 18th Conference of the Eu- S. Ronken, D. Wirz, A.U. Daniels, M. P. Arnold, ropean Orthopaedic Research Society, Davos, Diameter of the spherical indenter markedly af- Switzerland, June 30 - July 2, 2010 fects dynamic indention modulus of articular cartilage, International Society of Biomechanics, S. Ronken, D. Wirz, A.U. Daniels, Mechanical Brussels, Belgium, July 3 - 7, 2011 properties of fresh and Thiel’s embalmed cartilage, Bernd-Spiessl-Symposium, Basel, Switzer- S. Ronken, D. Wirz, A.U. Daniels, M. P. Ar- land, June 17 - 19, 2010 nold, Dynamic indentation modulus of articular cartilage changes with spherical indenter diam- S. Ronken, Mechanical properties of fresh and eter, 7. Jahrestagung der Deutschen Gesellschaft Thiel’s embalmed cartilage, 6th Instructional für Biomechanik, Murnau, Deutschland, May Course in Biomechanics, Pontresina, Switzer- 19 - 21, 2011 land, January 13 - 16, 2010, best poster award

106 Curriculum Vitae

Personal Data

Surname Ronken Given names Sarah Date of birth 28/01/1984 Place of birth Nederweert, the Netherlands

Education

Apr 2009 - May 2012 Ph.D. thesis, University of Basel. Supervisors: M.P. Arnold, D. Wirz and A.U. Daniels

Sep 2006 - March 2009 Master of Science in BioMedical Engineering Eindhoven Univesity of Technology, Eindhoven, the Netherlands

Aug 2002 - Aug 2007 Bachelor of Science in BioMedical Engineering Eindhoven University of Technology, Eindhoven, the Netherlands

Aug 1996 - Jun 2002 A-level education Philips van Horne SG, Weert, the Netherlands

107