Dynamic viscoelastic characterisation of human osteochondral tissue: understanding the effect of the cartilage-bone interface

The cartilage-bone interface in articulating joints is key to moderating the transmission of tensile, compressive, and shear forces from the articular cartilage to the subchondral bone [1]. The complex organisation of collagen fibres within cartilage, in part, enables it to store and dissipate energy [2], and articular cartilage is considered to be a frequency-dependent viscoelastic structure [3‐6]. Studies that have analysed this interface have primarily focused on its structure and composition, characterising the calcified cartilage and underlying tidemark where collagen type I and II integrate [7‐10]. More recently, biological signalling between articular cartilage and subchondral bone have been identified through vascular microchannels that traverse the subchondral bone and calcified cartilage, allowing diffusion of small molecules [11].

The viscoelastic properties of isolated articular cartilage have been well-characterised. Edelsten et al. demonstrated that cartilage behaves non-linearly under high-speed loading [12]. More recently, Lawless et al. [13] focused on observations of isolated cartilage under load at frequencies across a physiological range (1–92 Hz [3]) and determined that the storage and loss moduli were frequency-dependent. Despite providing valuable results on the behaviour of articular cartilage under dynamic loading, both these studies looked at cartilage in isolation and therefore were not able to offer insight on the behaviour of the cartilage-bone system as a whole.

In the last decade, it has been shown that severe impacts to articulating joints are known to result in damage to the bone rather than the cartilage [14]. High-impact loading has been suggested as a major risk factor of osteoarthritis (OA), and subchondral bone has been identified as having a role in the progression of the disease [15‐17]. While it is known that both subchondral bone and loading frequency have a significant effect on cartilage viscoelasticity [13, 18], there is currently a gap in our understanding of subchondral bone’s response to loading, independent of cartilage. Recent work by Fell et al. endeavoured to rectify this by conducting an analysis of the mechanical properties of bovine cartilage and bone [19]. Interestingly, a linear correlation between bone mineral density (BMD) and cartilage viscoelasticity was identified, suggesting there is an important relationship between the two tissues at the interface. However, to date, there are no studies of this nature conducted in human tissue.

The mechanical properties of osteochondral tissues are evidently very complex. In addition to the studies previously described that have measured cartilage viscoelasticity directly, mathematical models have been developed to elucidate further information on how articular cartilage functions in vivo. For example, local temperature changes in cartilage under load have been assessed through the development of a model to determine temperature increase from cartilage intrinsic viscoelasticity [20]. Modelling cartilage allows for properties that are difficult to measure directly in vivo to be examined. However, there are disputes regarding some mathematical models of cartilage. For instance, as proposed by Huyghe et al. [21], triphasic theory may not respect the laws of thermodynamics. Given the complexity of both tissues, studying and modelling cartilage or bone in isolation does not provide a complete picture of either tissue’s capacity to store and dissipate energy [19]. Whilst there are studies that have identified the mechanical properties of animal cartilage, both on and off bone [13, 18], the role of the cartilage-bone interface in the dissipation of energy has not previously been investigated in human tissue, nor has it been incorporated into mathematical models of articular cartilage.

The aim of this study is to characterise the viscoelastic properties of human osteochondral tissues and assess the dissipation of energy by these tissues. More specifically, an approach that characterises viscoelastic behaviour of the osteochondral core and isolated tissues in a physiological frequency range has advantages in being able to assess the significance of the interactions between the two tissues. Therefore, energy dissipation has been analysed for osteochondral tissues. By using dynamic mechanical analysis (DMA), the viscoelastic properties of the human cartilage-bone unit were directly compared to the subchondral bone and articular cartilage. Furthermore, the bone mineral density (BMD) of the subchondral bone was determined, by micro-computed tomography (μ-CT), to identify any relationships with its mechanical properties, or the viscoelastic properties of cartilage.

Of the 23 specimens tested, the results of two were rejected since the cartilage thickness (2.6 and 2.7 mm) lay far outside the normal physiological range of 1–2 mm [32] and may be representative of degenerated cartilage [33]. Peirce’s criterion was used to eliminate the two values [34].

This study established the viscoelastic behaviour of isolated cartilage and bone specimens in compression, and compared it with the osteochondral core, an analogy not previously conducted with human tissue. Cartilage-bone cores, isolated cartilage, and subchondral bone specimens were determined to be viscoelastic across all frequencies tested. Significant differences in behaviour of the core compared with the isolated specimens for k’ and k” were identified. Most notably, cartilage-bone specimens displayed a frequency-independent trend for k’ unlike isolated specimens, which were frequency-dependent. Previous research has focused on cartilage or cartilage-bone specimens alone [4, 6, 13, 18] and has not looked at the independent properties of human subchondral bone. DMA testing of human subchondral bone was conducted in the present work and the results highlight a significant difference in the viscoelastic properties of bone in comparison to cartilage; E” was shown to increase with increasing frequency for cartilage yet decrease with increasing frequency for bone. This data emphasises the complex mechanical properties of osteochondral tissues and in particular demonstrates the significance of the cartilage-bone interface.

A previous study by Temple et al. characterised the viscoelastic properties of human articular cartilage and revealed that this tissue behaves in a frequency-dependent manner, fitting a logarithmic function, thus supporting the findings of this study [4]. They reported storage and loss moduli for articular cartilage from human femoral heads to be 31.9–43.3 MPa and 5.3–8.5 MPa respectively. This is in line with reported mean storage and loss moduli in the present analysis of 34.4 ± 3.35 MPa and 6.17 ± 0.48 MPa respectively. In addition, Jeffrey & Aspden [35] calculated a ‘dynamic’ modulus for human articular cartilage from femoral heads by applying an impact load. The resulting dynamic modulus of human articular cartilage at stresses of 10 MPa and 23 MPa was reported to be 64 ± 13 MPa and 85.1 ± 4.9 MPa respectively. This is higher than the values reported for cartilage storage modulus in this study, although it is important to note that the induced stresses applied by Jeffrey & Aspden [35] were much higher, and higher stress is known to increase the storage modulus [13]. The present study has demonstrated that both storage and loss modulus for cartilage increase with increasing frequency. Notably, more energy is stored by both cartilage and bone tissues than is dissipated, with the effect most apparent at higher frequencies. It was determined that E” was an order of magnitude lower than E’ for bone across all frequencies tested. Similarly, E” for cartilage was lower than E’ between 1 and 90 Hz, as confirmed by other authors [4]. A small subset of the population is known to have high heel-strike rise times during gait, which induce forces at high frequencies (as high as 92 Hz) on the lower limbs [3]. Initially hypothesised by Radin et al. [36], high-frequency loading on the lower limbs has been identified as a marker of early onset osteoarthritis (OA), a degenerative disease that results in cartilage loss [37]. This study demonstrated increased energy storage for both cartilage and bone at higher frequencies, with energy dissipation determined to be an order of magnitude lower than energy storage. This behaviour could be a contributing factor to disease progression due to damaging stress concentrations. This is significant, as subchondral bone tissue in particular is likely to incur damage due to its increased ability to store energy yet decreased ability to dissipate it. However, as bone has a higher regenerative capacity than cartilage [38], this may be a mechanism for preventing cartilage damage, as the bone is more readily able to heal.

Bovine articular cartilage is considered a good model for human articular cartilage, as it displays similar trends in viscoelastic properties [4]. A recent study by Fell et al. [19] determined the viscoelastic properties of bovine cartilage and subchondral bone. They reported mean storage and loss moduli of bovine cartilage to be 45 MPa and 5.5 MPa respectively, whilst isolated bovine subchondral bone was found to have a mean storage modulus of 110 MPa and a mean loss modulus of 5 MPa. The storage modulus for bovine cartilage was therefore 1.3 times higher than isolated human cartilage in this study, which had a mean of 34.4 MPa. However, the bovine storage moduli for bone and loss moduli for cartilage reported by Fell et al. [19] were found to be lower than the respective results for the isolated human specimens in this study, though in the same order of magnitude. A possible reason for this is that the bovine tissues were obtained from the knee joint, whereas the human specimens were from the hip. Therefore, anatomical region is important when investigating and comparing the mechanical properties of osteochondral tissues.

Similar trends to the results in the present study were displayed by Fell et al. [19] for cartilage and bone storage and loss moduli (1–90 Hz). Of note, the loss modulus trends found in bovine tissue were also evident in human tissue (Fig. 3b). Whilst the trends were similar for bovine and human osteochondral tissue, Fell et al. evidenced a ‘crossover’ of the cartilage and bone loss modulus values [19]. At 1 Hz, bone had a higher loss modulus than cartilage, but at frequencies above 8 Hz this was reversed with cartilage samples exhibiting a higher loss modulus than bone. The authors suggested this might be a mechanism to prevent cartilage damage, as cartilage will dissipate more energy than bone under high-frequency loading. However, this trend was not noted in human osteochondral tissue, where bone had a higher loss modulus than cartilage at all frequencies tested (Fig. 3b). This data raises the hypothesis that cartilage damage in OA patients could be due to an inability of cartilage to dissipate energy into the bone under high-frequency loading, but further experimental work would be required to evidence this. A comparison of the trends seen here with Fell et al. demonstrates the importance of studying human osteochondral tissue since it is clear that animal models do not always display the same trends [4]. The present study is the first to report viscoelastic properties of human articular cartilage, subchondral bone, and the osteochondral core using Dynamic Mechanical Analysis.

As well as investigating isolated tissues, this research aimed to better understand the osteochondral core as a whole system. Storage stiffness for the cartilage-bone system was logarithmically frequency dependent and lower than cartilage and bone for all frequencies tested (Fig. 4). Loss stiffness for the cartilage-bone system was independent of frequency and lower than isolated specimens across the range of frequencies tested. These results are in line with previous work, which looked solely at bovine cartilage-bone cores and found loss stiffness to be frequency-independent [13, 18]. The difference in behaviour of cartilage isolated from and attached to subchondral bone is emphasised here and has demonstrated that cartilage should not be considered in isolation when determining properties representative of in vivo behaviour. The data obtained in our study supports the development and testing of whole tissue-replacement systems as opposed to cartilage replacement materials in isolation.

Prior studies of bovine cartilage both on- and off-bone found the loss modulus of on-bone cartilage to be frequency-independent, whereas cartilage off-bone has a frequency-dependent modulus [13, 18]. Lawless et al. [13] found that there was no dependency of the storage stiffness on the presence or absence of the underlying subchondral bone, and therefore proposed that on-bone cartilage may be more predisposed to failure than off-bone cartilage due to the storage/loss ratio being higher for cartilage on-bone. The findings of the present study report the same frequency-independence for cartilage on-bone loss modulus, with isolated cartilage displaying a frequency-dependent trend. Thus, findings reported in prior studies support the current results, although it should be noted that the aim of the present study was focused on the viscoelasticity of subchondral bone and the role of the cartilage-bone interface, rather than the cartilage itself. Hence, a more detailed discussion on the viscoelastic properties of cartilage both on- and off-bone is provided elsewhere [13].

In order to explain the difference in behaviour between isolated cartilage and the cartilage-bone-system system, Edelsten et al. [12] suggested that cartilage attached to subchondral bone is more constrained in its deformation, and this may lead to it appearing stiffer and more elastic than when in isolation. Experimentally, this result has been verified as k” becomes frequency-independent under load [12, 13]. However, the trends in E’ and E” in isolated bone identified in this study (Fig. 3) may infer an additional explanation. Bone was found to be positively frequency-dependent for storage and negatively frequency-dependent for loss moduli. Therefore, the constraining effects of the bone could not be the sole reason for the frequency independence of k” for the cartilage-bone-system. A decrease in loss stiffness indicates a decrease in the energy dissipated, suggesting that bone does not dissipate as much energy to the surrounding tissue at higher frequencies. This may prevent the load energy being returned to the cartilage, which could be a further mechanism to prevent cartilage damage. While observing equine osteochondral cores under high-impact, Malekipour et al. identified that bone can absorb a much higher amount of impact energy than cartilage [39]. Furthermore, it is often bone that breaks during high-impact loading, as the main mechanism by which it absorbs energy is through trabeculae fracture [39]. Although this may be desirable, as bone has a greater propensity to heal than cartilage, this may put the joint at risk of long-term damage. Clearly, the interface between the two tissues plays an essential role in transferring the load energy through the cartilage and into the bone.

A significant difference was identified between loss stiffness values for the cartilage-bone system and cartilage specimens. In contrast, when comparing the cartilage-bone system and bone specimens this was not observed, signifying that the loss stiffness of the osteochondral core is more closely aligned to the loss stiffness of bone than of isolated cartilage. This suggests a key property to consider is the energy transfer, including dissipation of energy, through the osteochondral junction. In order to characterise this in the present study, the total energy dissipated from the cartilage at each frequency during DMA was calculated. The benefit of this approach is the entire system is considered as a whole by determining absolute values rather than material properties. The results demonstrated that cartilage dissipated higher total energy than bone and the cartilage-bone system across all frequencies tested (Fig. 5). While expected, as one of the key roles of articular cartilage is to dissipate energy into the underlying bone, to the best of the authors knowledge this is the first report of such a relationship from human tissue analysis.

It has been hypothesised that high heel-strike rise times during gait causes a predisposition to the onset of OA [6, 36]. This high-impact loading, combined with a reduced ability to dissipate energy at high frequencies, is likely to cause damage to the bone, leading to subchondral bone remodelling and therefore a higher BMD. Prior to work by Fell et al. [19], no previous studies had attempted to link bone histomorphology with viscoelastic properties of osteochondral tissues. The study on bovine osteochondral cores identified a significant relationship between subchondral bone BMD and cartilage loss stiffness [19]. Correlations between cartilage thickness with BMD and subchondral bone plate thickness were also unearthed. This suggests that cartilage health is interrelated with the histomorphological properties of the subchondral bone. The results of the present study demonstrated a statistically significant (p < 0.05) link at all frequencies except at 90 Hz (p = 0.09) between the total energy dissipated by isolated cartilage specimens and BMD (Fig. 6b). This supports the findings by Fell et al. and validates the hypothesis that the dissipation of energy through to the subchondral bone provides mechanical signals, which can alter the structure of the tissue, for example a remodelling of the trabeculae [40]. It should be noted that although the goodness of fit is low between the BMD and total energy dissipated from the cartilage, they are comparable to R and R² values presented in other BMD studies [19, 41, 42].

There are a small number of limitations within the present study. One of the key challenges with human tissue is that specimens can have wide variation due to factors such as age, weight, and gender, which are known to have an effect on the health of these tissues. In addition, the patients who donated their tissue for this study are of an advanced age and therefore may not be representative of younger adults. Furthermore, it has been previously reported that bovine cartilage stiffens with age [43]. Whilst the tissues in the present study are obtained from older adults, the joints showed no sign of OA joint damage upon chondropathy assessment and are representative of cartilage that has maintained health throughout its lifespan. The high variation between specimens, combined with reduced specimen numbers due to tissue availability, is likely to be the reason that a significant link between BMD and cartilage viscoelasticity was not established in the present study, despite being demonstrated in bovine tissue [19]. It should also be noted that the cartilage surface in the sample blocks exhibited a slight curvature (Fig. 1b). To minimise the impact of this on mechanical test data, cores were taken from flatter regions of the block, and hence it was not accounted for. There are advantages and disadvantages to using stiffness measurements, as reported in this study. The calculation of storage and loss moduli is dependent on thickness, and therefore any limitations associated with thickness measurements are included in moduli, but are not included in stiffness [44], however, direct comparison of stiffness can only be made for the same shape and size of specimens. As moduli and stiffness are fundamentally different measurements, it limits the comparability across literature, although their trends with frequency of loading can be compared because the modulus is simply the stiffness divided by a constant (i.e. a shape factor) [13]. The primary reason for choosing to report stiffness in this study is that the osteochondral cores were a structure consisting of both soft and hard tissues. A separate limitation during DMA testing of biological tissues is damping if these are tested in a fluid medium. To avoid potential damping due to fluid in this study, the samples were not submerged in fluid and, therefore, the dissipation of energy reported was solely due to the cartilage tissue. Damping has been previously shown to have little effect between samples tested in air and in Ringer’s solution [13].

The complexity of the frequency-dependent viscoelastic properties of cartilage and bone described in this study demonstrate that there is a sophisticated interaction between these two tissues in regard to storage and dissipation of energy. High-frequency loading results in an increased storage of energy in the subchondral bone, a likely mechanism to prevent cartilage damage, which is a factor that should be further investigated in relation to the progression of OA. The results obtained in this study provide details of native tissue behaviour in vitro at physiologically relevant frequency ranges, the first reported values in human osteochondral tissue under dynamic loading.

This study demonstrates that subchondral bone is viscoelastic over a physiological frequency range (1–90 Hz). Storage and loss moduli for cartilage and storage modulus for bone increase with increasing frequency, while the loss modulus of bone decreases. Loss stiffness for the cartilage-bone-system is frequency-independent yet for isolated specimens it was shown to be frequency-dependent. Finally, a statistically significant link has been identified between cartilage energy dissipated and bone histomorphology. From these results, the importance of characterising the properties of articular cartilage both isolated from, and attached to, subchondral bone is clearly highlighted. Overall, this work demonstrates the significance of the bone-cartilage interface, in particular for energy storage and dissipation.