Traversing the intact/fibrillated joint surface: a biomechanical interpretation

Abstract

Cartilage taken from the osteoarthritic bovine patellae was used to investigate the progression of change in the collagenous architecture associated with the development of fibrillated lesions. Differential interference contrast optical microscopy using fully hydrated radial sections revealed a continuity in the alteration of the fibrillar architecture in the general matrix consistent with the progressive destructuring of a native radial arrangement of fibrils repeatedly interconnected in the transverse direction via a non-entwinement-based linking mechanism. This destructuring is shown to occur in the still intact regions adjacent to the disrupted lesion thus rendering them more vulnerable to radial rupture. Two contrasting modes of surface rupture were observed and these are explained in terms of the absence or presence of a skewed structural weakening of the intermediate zone. A mechanism of surface rupture initiation based on simple bi-layer theory is proposed to account for the intensification of surface ruptures observed in the intact regions on advancing towards the fibrillation front. Focusing specifically on the primary collagen architecture in the cartilage matrix, this study proposes a pathway of change from intact to overt disruption within a unified structural framework.

Introduction

The mechanical properties of articular cartilage (AC) arise from the interplay of two primary components – the collagen fibrils and the proteoglycan (PG) component with its high swelling potential. The fibrils are arranged as a three-dimensional meshwork of ultrastructural dimensions that limits the swelling of the PGs thus conferring on the cartilage an intrinsic swelling stiffness. This stiffness, combined with that derived from the resistance to outflow of the matrix fluid through the ultralow-permeability structure, provides a level of deformability sufficient to spread the contact forces over an area of the joint surface large enough to maintain contact stresses transmitted into the bone within safe limits. Any loss of effective coupling between the collagen network and the PGs will lessen the ability of AC to perform its role as a stress-reducing layer.

Intact AC is zonally organized into surface, intermediate or transition, mid and deep zones (Meachim & Stockwell, 1979) all of which reflect a well-defined fibrillar architecture (Broom, 1984a,1986a; Jeffery et al. 1991; Clark & Simonian, 1997; Lewis & Johnson, 2001). Each of these zones possesses its own characteristic mechanical anisotropy as demonstrated using microrupture propagation studies (Broom, 1984a).

Osteoarthrtitis (OA) is a condition of the articulating joint involving extensive remodelling of the subchondral bone and permanent destruction of the covering of cartilage. Although the detailed mechanisms by which OA develops are still poorly understood it is now widely accepted that the chondrocytes are stimulated to produce enzymes that degrade both the PGs and the type II collagen (Silver et al. 2001; Squires et al. 2003). The now weakened matrix is thus rendered more vulnerable to the mechanical environment in which it is required to function (Dodge & Poole, 1989; Rizkalla et al. 1992; Hollander et al. 1994, 1995; Young et al. 2002).

Although a primary focus of OA research must be the early molecular mechanisms of AC degradation, structural alterations associated with degradation will be better understood if the interfibrillar relationships serving this tissue's complex fibrillar architecture are more accurately defined. To this end, Broom et al. (2001) recently proposed a structural model that seeks to unify a substantial body of microstructural and ultrastructural data obtained from both normal and degenerate cartilage matrices. Specifically, the model predicts that, if the cohesivity of the normal fibrillar network arises from a non-entwinement-based mode of interconnection, a major alteration in the normal fibrillar network will result when this interconnecting system is degraded.

Closely related to changes in the fibrillar architecture of AC are its bulk swelling properties. Cartilage swelling is known to be an important characteristic of human OA and there is general agreement that this swelling is linked with a weakened or altered fibrillar network (Maroudas et al. 1986; Muir, 1989; Hwang et al. 1992).Bank et al. (2000) showed that the degree of degradation of collagen molecules within the fibrils in AC is directly related to the degree of swelling of the OA matrix in hypotonic saline.

The bovine patella provides a large mammal tissue model suitable for investigating early structure-related changes in the joint cartilage with the ability to track the progression to a fully developed osteoarthritic lesion. Patellar cartilage from these mature animals (> 5 years old) consistently ranges from fully intact through to mild and severely fibrillated states. Our principal aim was to answer the following question: is the ostensibly healthy tissue adjacent to the disrupted region in fact already exhibiting more subtle signs of degeneration and thus on a downward spiral towards overt matrix breakdown? We have sought to answer this question using microstructural evidence obtained from both the articular surface layer (AS) and the underlying general matrix across the intact/fibrillated transition zone.

Materials and methods

Patellae were collected immediately following the slaughter of mature female bovine animals 5–7 years of age. Using the Outerbridge (1961) classification scheme patellae graded approximately as grade 1 or 2 were selected and then stored frozen at −20 °C until required. Prior to experimentation, the patellae were thawed in cold running water for 2 h. Their articular surfaces were then stained with Indian ink and a photographic record was taken of the extent of AS disruption.

Using a fine-toothed bone saw a block of AC-on-bone with en face dimensions of approximately 15 × 25 mm was sawn from each patella so as to incorporate the intact–fibrillated transition along the larger proximal–distal dimension. The blocks were equilibrated in 0.15 m saline and then fixed in 10% formalin with cetylpyridium chloride at room temperature for at least 24 h. Most tissue blocks were then further divided into subblocks again along their larger dimension. Using a double-bladed parallel cutting device (Broom, 1984a; Chen & Broom, 1999) full thickness cuts were made through the AC, more central regions of each block being chosen so as to be at least 3 mm from any sawn edge. The vertically cut layer of AC was then undercut along the osteochondral junction to yield full depth radial slices approximately 0.1–0.15 mm in section thickness.

The slices were then wet-mounted in 0.15 m saline on a glass slide and cover-slip and examined using Nomarski differential interference contrast (DIC) light microscopy. No staining or further preparation was required. Two points should be noted concerning the method of tissue storage and fixation used in the study. First, our laboratory's long-established practice of using cartilage harvested fresh and then frozen and later thawed once for experimentation has never revealed any evidence of structural alteration arising from such treatment when compared with similar tissue examined in the fresh state. Secondly, equally effective DIC imaging can be obtained from cartilage in its unfixed hydrated tissue. However, the relatively soft fibrillated regions in the unfixed state are considerably more difficult to slice. Thus, in the interests of consistency throughout this study we adopted the procedure of formalin-fixing on-bone all tissues prior to their examination.

At least two tissue slices were examined from each of the sub-blocks prepared from ten grade 1 and ten grade 2 patellae.

Results

Our rationale for examining tissue from patellae exhibiting only mild degrees of degeneration was to capture microscopically the progression of structural disruption without the complicating presence of structure resulting from attempted repair responses associated with advanced OA (Buckwalter & Mankin, 1997). Both groups of patellae provided extensive regions of intact and fibrillated surfaces that were free of eburnation, i.e. a substantial covering of the original cartilage remained.

Macroscopic appearance

Figure 1(A,B) show representative examples of AC-on-bone blocks sawn from grade 1 and grade 2 patellae with their focal lesions revealed with Indian ink staining. An AS largely largely free of ruptures can be seen to blend into the visibly disrupted region of the grade 1 patella, whereas on the grade 2 surface the intact region exhibits a network of cracks which increased in density as the fibrillated region was approached.

Fig. 1.

Cartilage-on-bone samples obtained from grade 1 (A) and grade 2 (B) bovine patellae showing the intact-fibrillated transition. In B note the increased density of cracks as the fibrillated region is approached.

Microscopic appearance

Microscopically the intact regions of the grade 1 patellae exhibited an entirely normal zonal differentiation of the matrix. The AS was smooth and largely free of any clefts (Fig. 2A). In the intermediate zone the extracellular matrix was typically textureless (Fig. 2B), i.e. without any obvious directional appearance. The mid and deep zone matrix exhibited normal chondrocyte column formation and a fine, barely resolved radial texture (Fig. 2C).

Fig. 2.

Micrographs showing zonal variation in matrix of intact region of a grade 1 patella adjacent to fibrillated region: (A) superficial zone; (B) intermediate or transition zone; (C) deep matrix. Arrows indicate radial direction.

The intact regions of the more severely fibrillated grade 2 patellae exhibited the same zonal structure as in the grade 1 patellae but with the AS fragmented by a mosaic of clefts (Fig. 1B). Microscopically, these clefts represented two distinct modes of rupture. The first mode was observed generally in regions of the intact regions more distant from the disrupted lesion and involved the downwards propagation of a vertical tear which penetrated to varying depths into the underlying matrix (Fig. 3A,B). The root of the relatively shallow cleft in Fig. 3(A) is shown at higher resolution in Fig. 3(C); from the chondrocyte alignment it is clear that the cleft has penetrated only to the lower part of the intermediate zone where there is no resolvable radial texture. The higher resolution image of the matrix in the vicinity of the much deeper cleft in Fig. 3(B) reveals a well-developed radial texture with fine in-phase crimp formation (Fig. 3D).

Fig. 3.

(A,B) Varying depths of penetration of radial propagating clefts in the intact region of a grade 2 patella. (C,D) Higher magnification views of the clefts in A and B, respectively. Arrows indicates radial direction.

The second common mode of rupture of the intact AS involved a process of delamination within the superficial layer itself and occurred commonly in regions closer to the fibrillated lesion. Figure 4(A) shows what is possibly an early stage in which a still unruptured uppermost layer has partially debonded. More commonly, superficial tears tracked approximately parallel to the surface for some distance before deflecting downwards and proximally, resulting in a pattern of curvilinear tears, again penetrating to varying depths (Fig. 4B, C). Still closer to the fibrillation front, i.e. in the transition region, delamination was increasingly common and more complex in its morphology (Fig. 4D), possibly a consequence of both multiple delamination effects and related loss of some of the superficial matrix. These disruptive features in the otherwise intact superficial layer of the bovine patella resemble those observed much earlier in macroscopically intact human cartilage (Freeman & Meachim, 1979).

Fig. 4.

(A) A partial delamination of the grade 2 articular surface without involving any tensile rupture in the plane of the surface. (B,C) Isolated curvilinear tears penetrating to different depths. (D) A more complex mode of delamination typical of the transition region. Arrows indicate radial direction.

Frequently seen in the mid and deep zones of the grade 2 patellae in their intact regions closer to the fibrillated region was a texture consisting of well-defined linear striations (Fig. 5A). Higher resolution imaging of this structural feature was achieved by carefully focusing on the upper surface of tissue slices of reduced thickness (≈10–20 µm) in the direction of the optical axis. Viewed at this higher resolution the striations are clearly continuous with a more distinctly braided aggregation of fibrous elements blending almost imperceptibly into an in-phase crimped fibrous texture (Fig. 5B).

Fig. 5.

(A) A distinct linear fibrous texture in mid and deep zones of grade 2 patella in the intact region close to the fibrillated region. (B) Higher resolution image of this same texture showing its finely crimped geometry (see small arrow). Long arrows indicate radial direction.

The actual fibrillated regions were marked by a complete loss of the original AS with the original underlying matrix (now exposed directly to the joint space) transversely fragmented by clefts penetrating to varying depths into the mid and deep zones. The ‘tufted’ surface resulting from this cleft-ridden formation tended to display a distinct flattening towards the distal aspect (Fig. 6). Cell proliferation in the form of multicellular clusters appeared to be confined mainly to the uppermost parts of the tufted structure (see arrow in Fig. 6A), with relatively normal chondrocyte columns characterizing the remaining depths of the cartilage (Fig. 6B).

Fig. 6.

(A) Deeply clefted and distally flattened exposed matrix in severely fibrillated region of grade 2 patella. (B) Deeper matrix contiguous with A. The original articular surface is absent. Long arrow indicates radial direction; short arrow indicates cell cluster.

At high magnification it was possible to track a continuous progression of change in matrix texture with respect to depth. Commencing in the deep zone, the general matrix generally exhibited only a faintly resolved radial texture (Fig. 7A). With decreasing depth this texture became increasingly easy to resolve as a structure distinctly aligned in the radial direction (Fig. 7B). With still decreasing depth this strongly aligned structure gradually transformed into a fine in-phase crimped morphology (Fig. 7C). Finally, on reaching the upper tufted zone of the cartilage this in-phase crimped morphology coarsened in both its periodicity and its amplitude (Fig. 7D).

Fig. 7.

Micrographs showing progressive change in matrix texture with depth in the fibrillated region of a grade 2 patella. The faintly resolved radial texture of the deep zone (A) gives way to an increasingly visible radial texture and crimp as the exposed fibrillated surface is approached (B–D). Arrows indicate radial direction.

The transition region between the intact and grossly fibrillated regions (Fig. 8) was strongly clefted but still retained in many places a near full thickness of cartilage, i.e. there was evidence of the original superficial zone of cartilage (see arrowed site in Fig. 8). As in the fibrillated regions some clefts penetrated into the mid and deep zones and with decreasing depth there was a similar progression of increasing prominence of radial texture and in-phase crimp.

Fig. 8.

Transition region of a grade 2 patella with a near full thickness of the original cartilage but containing deep radial clefts. Short arrow indicates original articular surface. Long arrow shows radial direction.

Discussion

Other investigators have used birefringence of both stained and unstained histological sections under polarizing microscopy to infer changes in collagen fibril orientation associated with degeneration (Williams et al. 1996; Panula et al. 1998; Xia et al. 2001; Han et al. 2002). However, apart from the inherent problem of artefacts associated with histological preparations, this structural tool provides only an overall indication of collagen orientation. Our use of DIC optical imaging had several major advantages. First, it provided a view of the fully hydrated radial sections at high resolution within minutes of their removal from the formalin-fixed bulk sample. This facilitated a more immediate correlation between the on-bone macroscopic state of the tissue and its microstructural appearance. Secondly, the relatively large thickness of hydrated section that is able to be imaged at high resolution with DIC microscopy (up to ≈150 µm) combined with a relatively small depth of focus yielded excellent image clarity within any given focal plane within the section thickness and greatly enhanced the visual interpretation of the fibrillar structures with little or no risk of structural distortion.

Osteoarthroses in the bovine stifle joint are commonly thought to result from instability associated with prior joint trauma and possibly overweight (Walker, 1971). Occasional cartilage defects in the bovine patella have also been reported (Bartels, 1975). The generally consistent pattern of cartilage disruption in the patella lesions described in the present investigation presumably reflects the biomechanical environment within which this organ functions in the animal.

Clearly it is impossible to reconstruct the exact sequence of structural events that have led to the patterns of disruption described in the present study. However, there are two structural features present in all of the individual slices of tissue studied which suggest that a plausible schema for the gradual breakdown of the functionally important collagenous framework can be reconstructed.

First, in the overtly disrupted regions there was no detectable dissonance between what was clearly original hyaline cartilage and those regions that contained a clearly resolvable fibrous texture. Both regions appeared to be linked by matrix exhibiting a gradual blending of structural texture from almost amorphous hyaline to strongly textured, i.e. in all of the samples there was a clear implication of structural continuity. Secondly, this continuity of textural change was always expressed in the radial direction, i.e. it was a structural change that gave every indication of being related to the primary architecture of the original healthy cartilage matrix (for a contrasting histological picture of the discontinuities associated with reparative responses in end-stage OA see Mainil-Varlet et al. 2003).

A recent study (Broom & Flachsmann, 2003) investigating specifically the relationship between general matrix swelling and microscopic texture of intact cartilage adjacent to regions of fibrillation in grade 1 patellae indicated that much higher levels of transverse swelling of the general matrix were associated with a distinct, microscopically visible radial texture. Related ultrastructural evidence has shown that this texture arises from an aggregating tendency of the once spatially discrete arrangement of fibrils making up the fibrillar architecture of the healthy general matrix (Chen & Broom, 1999).

The increased swelling tendency of the more aggregated architecture would result from a loss of the overall interconnectivity, a property crucial to the maintenance of a spatially discrete architecture. In addition, this loss of interconnectivity would reflect a loss of transverse strength in the fibrillar framework, an interpretation supported by much earlier microrupture propagation experiments conducted by Broom (1984a,b) which demonstrated that radial propagation of a tear was achieved at substantially lower tear opening forces in those matrices that were strongly textured compared with the more amorphous matrices.

Assuming that the currently observed patterns of disruption have occurred within a coherent structural framework as summarized above it is now possible to propose a likely sequence of degradative changes in those fundamental structural relationships responsible for giving the healthy cartilage matrix its remarkable durability. We begin with a spatially discrete arrangement of repeatedly interconnected fibrils that have a ‘native’ radial arrangement in the general matrix (Fig. 9A), an architecture that would optimally constrain the high swelling PG components to achieve an adequate level of intrinsic stiffness, but also one that confers on the matrix a high transverse strength and thus resistance to radial rupture. Optically this ‘ideal’ structure would not be resolvable, a fact that is consistent with the general lack of directional texture in the general matrix of healthy cartilage (Broom & Flachsmann, 2003).

Fig. 9.

Two-dimensional schema depicting progressive breakdown and re-arrangement of a network of fibrils in the general matrix based on a mediated, non-entwined mode of transverse linkage. Note how differing levels of elimination of the interconnecting elements lead to a range of destructured configurations.

Concerning the mechanism of interconnection, we have shown in earlier ultrastructural studies that fibrils are occasionally seen to entwine physically in the general matrix (Broom & Silyn-Roberts, 1989; Chen & Broom, 1998) but all the evidence suggests that this is not a major linking mechanism. Furthermore, physical entwinement fails to account for the remarkable degree of transformation that is possible in the cartilage general matrix induced either by repeated mechanical impact (Broom, 1986b) or by limited attack with collagenase (Broom, 1988).

Broom et al. (2001) proposed that a more probable interconnecting mechanism will involve some mediating molecule or agent that is able repeatedly to link neighbouring fibrils at localized sites along their axes without involving any significant amount of physical entwinement. Such a mechanism is consistent with earlier published ultrastructural evidence showing that fibrils in the general matrix do associate closely and often in a transient manner (Broom & Marra, 1986). A major dismantling of this form of linkage, and thus the transverse interconnectivity of the fibrillar network as a whole, is then readily achieved simply by attacking the mediating agent.

Although acknowledging that some interfibril entwinement will operate in the matrix, the schema of structural decline from the ‘ideal’ arrangement (Fig. 9A) that we propose is based on the simplifying assumption of a mediated, non-entwining form of linkage that keeps the fibrils spatially separated in an optimally constrained configuration. Figure 9(B) therefore represents some limited destruction of the interfibril linkages, thus enabling some fibrils to aggregate into less constrained parallel configurations over limited distances along their lengths. This would account for the short-range radial texture that is readily resolved in Figs 2(C) and Figs 7(A). With an increasing level of destruction of the linkages and the increased level of aggregation that is now possible over larger distances along the fibril length the extent of radial texture will increase. We suggest that the strong linear striations (Figs 5A and Figs 7B) and the extensive in-phase crimped textures (Figs 5B and 7C) can be broadly interpreted in terms of the schematics shown in Fig. 9(C, D), respectively. These striations are also consistent with the strong DIC-imaged radial textures that were shown previously to correlate with the extensive, near-parallel bundles of fibrils imaged with transmission electron microscopy in malacic cartilage (Chen & Broom, 1998, 1999). This micro/ultramicroscopic correlation serves to emphasize the importance of using DIC light microscopy as a convenient and more immediate tool for investigating major structural changes in cartilage.

Finally, with an almost complete dismantling of the interconnectivity, and with the added freedom of relative movement between neighbouring fibrils that is now possible, the fine crimp morphology (Fig. 7C) will progressively coarsen to yield the deep convoluted arrays shown in Fig. 7(D). This final level of de-structuring is consistent with these arrays being found predominantly in the upper tufted structure of the fibrillated region, a region of the matrix in which the clefted structure, now exposed directly to loading, would undergo a large amount of repeated lateral shearing. Such repeated mechanical agitation of the fibrils would presumably loosen further any remaining interfibril interactions, thus allowing them to adopt a minimally constrained configuration.

The observed patterns of rupture on the original AS may also provide additional biomechanical insights. The penetrating tears associated with the delamination mode of rupture all tended to follow a curvilinear path downwards and proximally through the intermediate zone (Fig. 4B, C). This distinctive morphology is strongly reminiscent of the ‘leaved’ layers of fibrillar meshwork proposed by Jeffery et al. (1991) and Lewis & Johnson (2001), although it is a structural interpretation disputed by others (Silver et al. 2001). Interestingly, this distinctive pattern of rupture observed in the intact regions was not seen in all patellae comprising the present study. The straight radial tears (Fig. 3A, B) are more consistent with an absence of lateral bias in the intermediate zone and thus would seem to be in conflict with the Jeffery et al. structural model.

The two contrasting modes of surface rupture observed in the present study might be better explained in terms of the interaction between the mechanical environment and the fibrillar architecture of the intermediate zone. It is generally accepted that the fibrils in this zone are in transition between the deeper radial (i.e. overall) configuration and the surface tangential configuration but not necessarily with the preferred unidirectional lateral bias as is embodied in the model of Jeffery et al. If a loading bias was then imposed on this directionally ‘neutral’ transitional structure, such that it was sheared more intensely in one direction than in the other, the interconnectivity of the matrix might then be weakened in an asymmetrical manner. A surface tear, once initiated, would then track down through this transitional zone along these weakened pathways to produce the curvilinear ruptures. It should also be noted that the exposed tufted morphology characterizing all fibrillated regions was also deflected laterally and somewhat flattened in the same orientation bias as the observed curvilinear tears in the intact regions (see Fig. 6), consistent with a skewed pattern of loading.

At the macroscopic level the intensifying degree of surface cracking of the intact cartilage as the fibrillated region is approached (Fig. 1B) may also offer some insight into the possible mechanisms leading to overt disruption. Two quite different biomechanical scenarios are suggested, both of which assume the patella is subjected to a relatively narrow band of localized loading that sweeps over its medial and lateral surfaces in the proximal–distal plane during function.

The most straightforward explanation is that the consistent location of focal lesions is directly related to a repeating high-stress site on that part of the joint surface. The decreasing intensity of surface rupture with distance from the focal lesion would simply reflect a corresponding pattern of reduced intensity of loading.

The second scenario invokes simple bi-layer theory to account for the pattern of surface rupture and exploits the fact that the AS, by virtue of its in-plane arrangement of fibrils, has far greater strain-limiting properties than its underlying intermediate and general matrix. In healthy cartilage the general matrix has been shown to stretch transversely up to 80–100% before rupturing compared with the ∼20% extensivity that is possible in its AS (Broom, 1984a; Flachsmann et al. 2001). In addition, because the AS and its underlying matrix are structurally linked, the AS can rupture under direct compression as a consequence of secondary tensile forces developed in the plane of the surface (Flachsmann et al. 2001). Applied earlier to arteries containing an arteriovenous fistula (Broom et al. 1993) the bi-layer model can be generalized to predict, at least qualitatively, the rupture behaviour of cartilage with its AS and underlying matrix represented as two bonded layers with contrasting strain-limiting and strength properties.

Assuming a constant shear stress (proportional to the slope θ) along the bond for a given level of tensile load (P), the bi-layer model predicts a progressive rise in tensile stress in the more strain-limiting AS layer with increasing distance in from the free ends (Fig. 10A). When the tensile stress in the AS layer reaches its intrinsic failure strength (σ_f), rupture will occur at the site of highest stress (Fig. 10B). Once the first rupture has occurred the length of interface bond which transfers the shear stress is now halved. Because the maximum stress in the AS layer depends on the interface bond length this has the effect of relieving the stress in the AS (Fig. 10B). The now sequestrated AS will undergo further rupture if the load level P is increased such that again σ_f is reached in the sequestrated portions (Fig. 10C).

Fig. 10.

Schematic illustrating mode of rupture in a simple bi-layer model of cartilage in which there is structural bonding between two layers of contrasting strain-limiting properties. (Adapted from Broom et al. 1993.)

Two important predictions arise from the application of this model. First, the spacing of tears in the more strain-limiting layer will be inversely related to its intrinsic failure strength σ_f. Thus, the gradual intensification of splitting of the AS as the fibrillated region is approached (Fig. 1B) could be interpreted as indicating a pattern of progressive weakening of the still intact AS. It would also imply that structural weakening of the intact surface is taking place well before actual disruption occurs. In a recent study of the mechanics of rupture of the AS of cartilage-on-bone samples, R. Flachsmann et al. (unpublished observations) showed that the in-plane secondary tensile stresses resulting from direct compression produce patterns of AS rupture that were dependent on the condition of the tissue. Whereas in the healthy cartilage only single rupture morphologies were observed, in the mildly degenerate tissues a pattern of multiple rupture was common. This difference in rupture behaviour is again consistent with that predicted by the bi-layer model.

Secondly, if the interface shear stress reaches that of the intrinsic shear strength of the interface bond then failure by delamination would be predicted. This is illustrated in Fig. 10(C) where, with the repeated rupturing of the strain-limiting layer induced by increasing levels of loading, there is an associated increase in the slope of the stress distribution curve for the more strain-limiting AS. Such a mechanism might account for the more complex patterns of delamination reported in the present study (Fig. 4A). Furthermore, earlier ultrastructural studies of the AS by Kamalanathan & Broom (1993) hinted strongly at its multiply/multilayered architecture, as have the scanning electron microscopy studies of both Clark & Simonian (1997) and Lewis & Johnson (2001). Any generalized degradation of the collagen network would presumably reduce both the in-plane tensile strength of the specialized fibrillar architecture in the superficial layer as well as the structural bonding between its constituent sublayers (see ultrastructural evidence for this possible interlayer bonding in Lewis & Johnson, 2001) and thus its ability to resist interface shear stresses. This would in turn render it more susceptible to a delamination mode of failure.

Apart from the complete loss of its original AS layer, the overtly fibrillated matrix bears few microstructural features that are not already hinted at in the adjacent unfibrillated regions. The radial texture is intensified in the fibrillated matrix but is certainly prefigured in the intact matrix. Both the intact and the fibrillated regions contain deep penetrating clefts and an associated strong radial texture consistent with a loss of transverse interconnectivity and thus an increased vulnerability to radial rupture. These observations contrast markedly with the rupture behaviour of healthy cartilage-on-bone samples when subjected to intense levels of compressive impact. The clefts penetrated down to the intermediate layer but never into the middle and deep zones (Silyn-Roberts & Broom, 1990). This limited depth of propagation of stress-induced radial clefts combined with the general absence of a pronounced radial texture in the healthy tissue is consistent with the view that surface rupture in vivo is closely associated with and probably preceded by a generalized weakening of the collagen framework in the transverse direction.

Conclusions

By tracking microscopically across the intact–fibrillated zones of the mildly degenerate joint cartilage we have shown that there is a continuity in the pattern of structural change in the fibrillar architecture that is consistent with the progressive destructuring of a native radial arrangement of fibrils repeatedly interconnected in the transverse direction via a non-entwinement-based linking mechanism. Evidence suggests that this destructuring takes place in the still intact regions of the mildly degenerate joint and that it renders the general matrix more vulnerable to radial rupture.

Two contrasting modes of surface rupture were observed in the intact regions adjacent to the fibrillated regions and these are explained in terms of the absence or presence of a skewed structural weakening of the intermediate zone. Mechanisms of surface rupture initiation based on simple bi-layer theory are proposed to account for the intensification of surface rupture observed in the intact regions on advancing towards the fibrillation lesion, and for the observed patterns of delamination.

By focusing specifically on the primary fibrillar architecture of the cartilage matrix this study adds substantially to the histological assessment study of Mainil-Varlet et al. (2003) and offers a unifying structural framework that enhances considerably our understanding of the relationship between degeneration and cartilage disruption.