Surface damage of bovine articular cartilage-off-bone: the effect of variations in underlying substrate and frequency

Twelve bovine humeral heads were obtained from a supplier (Dissect Supplies, Kings Heath, Birmingham, UK) from animals of maximum 30 months old at slaughter. Bovine articular cartilage was selected based on the previously established relationship of the frequency-dependent viscoelastic properties, consistent with that of human articular cartilage [25]. The humeral heads were covered in tissue paper, coated with Ringer’s solution prepared to a full strength mass concentration by dissolving 4.83 g of Ringer’s tablets (Oxoid Ltd., Hampshire, UK), per 500 ml of distilled water, and separately stored in double heat-sealed plastic at − 40 °C [13, 23]. Bovine humeral heads were thawed at room temperature for testing preparation [25]. The freeze-thaw process does not affect the mechanical properties of articular cartilage [26]. India ink (Daler-Rowney, Bracknell, UK) was used to ensure that humeral head cartilage surfaces were free from lesions ahead of testing [13, 21, 27]. For this procedure, the entire humeral head surface was covered with India ink so that on removal (rinsing) any defects could be identified; a commonly employed technique [13, 21, 27]. This allowed the coring procedure to take place at intact regions only. Further, on harvesting of each cartilage core, India ink was applied to each specimen, rinsed off, and an image captured for comparing damage post-testing. A representative image of a specimen captured before testing (but following application/removal of India ink) is displayed in Fig. 1.

To prepare off-bone cartilage cores, a hand cork-borer with an on face diameter of 10.5 mm was used to create circular dents on the surface of the articular cartilage, and through to the underlying bone. Following on from confirmation with the use of India ink of undamaged regions for coring, caution was taken during the harvesting process to attain undamaged cartilage specimens. Further, the absence of damage was confirmed via observation following the use of India ink post-harvesting. The locations selected were identical for all specimens, namely at the central region, ensuring the cartilage surface was complete and undamaged [25]. A surgical scalpel with blade size 10A (Swann-Morton, Sheffield, UK) was used to isolate the cartilage core from the underlying subchondral bone [12, 13, 17, 25, 28]. On harvesting of the cartilage cores from the underlying bone, they were immediately immersed in Ringer’s solution to prevent dehydration. Furthermore, the specimens were extracted on the day of testing; curling was not observed on any tissue harvested. To measure thickness, the cartilage core was held with a Vernier calliper at its centre; care was taken to avoid compression of the cartilage specimen. The thickness of the cartilage specimens measured prior to testing was 0.99 ± 0.004 mm (mean ± standard deviation), and the diameter was 10.9 ± 0.210 mm, as measured using a Vernier calliper (Draper Tools Ltd., Hampshire, UK). Six off-bone bovine articular cartilage cores were removed from each of twelve humeral heads, resulting in 72 individual specimens of off-bone cartilage. The specimens were immersed in Ringer’s solution for 30 min [29] prior to testing [25].

Specimens were positioned on the substrate blocks (Sawbones, Washington, USA) for testing. Four Sawbone densities were tested, which were 0.1556, 0.3222, 0.5667 and 0.6000 g/cm³. The substrate with the lowest density, substrate one, was used as an osteoporotic representation of bone [30, 31] whereas the highest density, substrate four, was closer to that of cancellous bone [32]. The Sawbone was prepared with a vertical bandsaw (Startrite Volant 24, UK) into blocks with dimensions 30 mm × 30 mm × 10 mm. The Young’s modulus of each substrate was obtained from compression tests using a Bose Electroforce 3300 testing machine run using Bose WinTest software (Bose Corporation, ElectroForce Systems Group, Minnesota, USA; updated to: TA Instruments, New Castle, DE, USA), derived from the slope of the stress-strain curve calculated from the force-displacement plot, at 0.02 mm/s with compression from a metal plate, 81 mm in diameter onto the substrate block (Table 1).

A square aluminium test rig was designed with dimensions 41 mm × 41 mm × 16 mm into which the Sawbone substrate was placed. A stainless-steel flat circular faced indenter, with a 0.5 mm bevelled edge to avoid stress concentration at its edges, 5.2 mm in diameter [2] was fixed to the actuator of the testing machine and used to induce damage through indentation to the surface of the cartilage-off-bone specimen. This procedure was similar to a previous study for on-bone cartilage [2]. A Bose ElectroForce 3200 testing machine controlled via the Bose WinTest 4.1 software (Bose Corporation, ElectroForce Systems Group, Minnesota, USA; updated to: TA Instruments, New Castle, DE, USA) was used for testing. The stainless-steel indenter was descended onto the cartilage specimen at the point of testing (Fig. 2).

Six cartilage specimens were tested per substrate, at a given frequency. The cartilage specimens were manually positioned on the underlying substrate, without the use of adhesive treatment prior to testing. Specimens were not observed to move during testing. This placement replicates the effect of the soft-on-hard construct but may not replicate the restriction of collagen at the cartilage calcified zone, due to the removal of the underlying bone [12]. A sinusoidally varying force under unconfined compression between 5 and 50 N was applied to the specimens for 10,000 cycles [2]. Repeated loading has previously been found to induce disruption to the articular cartilage surface [1, 2, 33]. Frequencies of 1, 10 and 50 Hz were applied to the specimens, for a completion of 72 individual tests. At 5000 cycles, the halfway point during testing, the articular cartilage-off-bone specimens were irrigated with Ringer’s solution [20, 21], to ensure hydration of the cartilage. Further, recent work has also confirmed the absence of a significant difference on the dynamic viscoelastic behaviour of articular cartilage, when tested for short periods in either air or Ringer’s solution [13].

Pictures were taken of each specimen before and after testing for clear observation of the damage induced after testing. India ink was used to identify alterations to the cartilage-off-bone specimen following testing, in addition to the sample prior to testing for damage absence confirmation [2, 27, 33]. For evaluation of the damage present, ImageJ software (version 1.48, Rasband, W.S., U.S. National Institutes of Health, Bethesda, Maryland, USA) was used with calibration of the image via a scale bar. The area of indentation (mm²) and crack length (mm) were analysed as two separate conditions for each specimen. Image measurements were repeated twice per sample and the mean value reported. Firstly, the area indented following testing with the indenter was highlighted (i.e. a non-damaged measurement), with use of the free-hand tool on the ImageJ software. Secondly, the length of cracks was measured using an existing method, and this was considered as a damage measurement [2].

Representative images of cartilage specimens after testing are shown for 1, 10 and 50 Hz, in Fig. 3, Fig. 4 and Fig. 5, respectively, at all four substrates investigated. The damage observed as crack formation is clearly indicated with a black outlined ellipse. Figure. 6 and Fig. 7 display the relationship between substrate density and the mean crack length and mean indented area, respectively, at all three frequencies investigated.

Crack length was significantly correlated to substrate density at 10 Hz (p < 0.05) (Fig. 6, Table 2), however, it was not significantly associated with substrate density at 1 or 50 Hz (p > 0.05) (Fig. 6, Table 2). The combined variables of frequency at 10 Hz with substrate density led to an increase in crack length with density (p < 0.05) (Fig. 6, Table 2). This relationship between the effects of frequency and substrate density, on the crack length of off-bone articular cartilage is best described by a linear curve as represented by eq. (1):

where c is the crack length mean total, A is the gradient of the slope, B is substrate density and D is the intercept; D and A are empirically derived constants. The associated details are summarised in Table 2, at all three frequencies tested.

Frequency of loading in combination with substrate density had no effect on the area of indentation when samples were tested at 1, 10 and 50 Hz (p > 0.05) (Fig. 7). The effect of frequency alone, on the area of indentation, however, did demonstrate a negative correlation between an increase in frequency from 1 to 50 Hz, and the indented area (Fig. 7). The post-test recovery following the testing of the cartilage specimens varied for each frequency. Testing at 1 Hz of frequency resulted in the least recovery of the articular cartilage specimen; with a larger indented area observed. This is represented by the observations where the indented area, as clearly highlighted by the black circular staining of India ink, extended to the majority of the cartilage specimen surface at 1 Hz (Fig. 3), in comparison to the lesser surface indented at 10 (Fig. 4) and 50 Hz (Fig. 5).

Sample images displayed at 1 Hz (Fig. 3) illustrate cracks that were of a single-line configuration, notably through the specimen periphery, parallel to the specimen circumference, observed at a maximum mean of 3.01 mm (Fig. 6; Table 3). As the applied frequency at testing increased to 10 Hz, mean crack length at this frequency increased to its maximum by 7.94 mm (Fig. 6; Table 3).

Representative images of 10 Hz in applied frequency at each substrate (Fig. 4), illustrate crack formation predominantly observed as multiple parallel straight-line arrangements, of various lengths, similar to previous studies [1, 2], commonly observed at the periphery of the cartilage specimen. At 50 Hz, the maximum mean crack length reduced by 5.30 mm from 10 Hz (Fig. 6; Table 3), where crack formation was primarily located across the diameter of the sample of single-line conformations, of various lengths, parallel to the specimen circumference (Fig. 5).

It is worth highlighting the potential limitation of testing off-bone cartilage, due to the absence of the restrictive attachment provided by the underlying subchondral bone that is found in the natural environment in-vivo [14]. The removal of the subchondral bone creates an alteration in the load transfer properties of the cartilage-off-bone specimen, notably the absence of the calcified cartilage layer with a stiffness in-between that of articular cartilage and the subchondral bone [15]. Despite this, however, the substrates used in this study have acted as the underlying bone of a controlled density, to allow for evaluation of the effects of the density of the underlying substrate alone, on associated cartilage damage.

While the results for off-bone cartilage failure in this study are larger than for on-bone data in the literature [2], there are limitations in directly comparing the on- and off-bone cartilage results from this study and the previous study [2]. Although identical joint locations i.e. the bovine humeral head have been assessed, load ranges applied during testing have differed, as well as specimen geometries [2]. In addition, in this study bovine articular cartilage cores have been removed from the adjacent regions of cartilage, therefore, weakening the cartilage due to the disruption of its extracellular matrix. The previous study [2], tested a select area of cartilage on a large joint sample with an undisrupted matrix [58]. However, the removal of the articular cartilage in this way was kept consistent throughout the investigation, thus specifically concerning the unknown relationship between underlying substrate density and cartilage, off-bone, at a varied frequency.

This study has used freeze-thaw cycles when preparing samples for testing. Although this process may have limitations, a recent study has concluded “multiple freeze-thaw cycles cannot be explicitly or statistically linked to mechanical changes within the cartilage” [59]. Previous studies have utilised bovine humeral heads [22], bovine knee joints [23], as well as bovine femoral heads [25]; each study referring to the absence of a freeze-thaw effect on the mechanical properties of cartilage [26]. Additionally, the storage of tissue at − 40 °C is a previously established protocol utilised by several studies [20‐25], and therefore was the approach taken for tissue storage in this study. Regardless, the results and conclusions obtained from this study on the effect of frequency and substrate on surface failure of cartilage are based upon a controlled testing protocol, and so findings are unlikely to be biased due to the freeze-thaw process used.

The use of an indentation test with articular cartilage is an established method previously developed to closely represent the physiological loading conditions of articular cartilage in vivo [60], as well as for damage inducing to the surface of articular cartilage [2]; having the advantage of being highly repeatable. Despite the hardness as well as the smaller diameter of the indenter in comparison to the cartilage specimen, a 0.5 mm radius bevelled edge was used to prevent artificial damage induced through stress concentrations at the edge. Further, the use of an indenter with a diameter smaller than the cartilage specimen, enables the deformation behaviour by the collagen matrix of the surrounding cartilage specimen [61] outside of the indentation area. Ultimately, this protocol has enabled a controlled evaluation of the effect of substrate density on cartilage failure.