Apparatus and Method for Rapid Detection of Acoustic Anisotropy in Cartilage.

Abstract

Purpose:

Articular cartilage is known to be mechanically anisotropic. In this paper, the acoustic anisotropy of bovine articular cartilage and the effects of freeze-thaw cycling on acoustic anisotropy were investigated.

Methods:

We developed apparatus and methods that use a magnetic L-shaped sample holder, which allowed minimal handling of a tissue, reduced the number of measurements compared to previous studies, and produced highly reproducible results.

Results:

SOS was greater in the direction perpendicular to the articular surface compared to the direction parallel to the articular surface (N=17, P = 0.00001). Average SOS was 1,758 ± 107 m/s perpendicular to the surface, and 1,617 ± 55 m/s parallel to it. The average percentage difference in SOS between the perpendicular and parallel directions was 8.2% (95% CI: 5.4% to 11%). Freeze-thaw cycling did not have a significant effect on SOS (P>0.4).

Conclusion:

Acoustic measurement of tissue properties is particularly attractive for work in our laboratory since it has the potential for nondestructive characterization of the properties of developing engineered cartilage. Our approach allowed us to observe acoustic anisotropy of articular cartilage rapidly and reproducibly. This property was not significantly affected by freeze-thawing of the tissue samples, making cryopreservation practical for these assays.

1. Introduction

Articular cartilage (AC) tissue is thin, typically 1 – 3 mm, and possesses a complex stratified structure that give rise to unique mechanical behaviors. These allow it to withstand the forces linked to body weight and joint activity [1–3]. Despite an uninteresting macroscopic appearance, AC is structurally inhomogeneous, and is known to exhibit non-linear anisotropic mechanical properties both in tension and compression [4–9]. This complex mechanical structure has been shown to control fluid flow in the tissue, and to regulate the metabolic activity of chondrocytes and the anabolic/catabolic state of the tissue [10].

Importantly, although numerous previous studies have measured the mechanical behavior of AC [11,10,12–17], the methods used in these studies are intrusive and destructive. For many investigations involving live tissue or in vivo evaluations, it would be useful to develop noncontact, nondestructive methods for measuring tissue properties. In particular, we predict that this will become important as efforts to generate tissue-engineered cartilage for repairing defects in AC evolve. Cartilage tissue engineering (TE) typically involves seeding a scaffold with cells (chondrocytes or stem cells) and the growing the resulting construct in a bioreactor. Thus far, however, stable, long-term repair of AC lesions with engineered cartilage has remained elusive. TE can yield constructs with chemical ECM compositions that approximate AC, however, they do not generally provide the structure and mechanical organization that characterize native cartilage [18]. Several groups have suggested that recapitulating the properties of AC will be required for success [19–23]. In particular, failure to restore the depth-dependent structural anisotropy of the ECM in engineered AC likely contributes to the failure of implanted engineered cartilage [24].

We have been pursuing methods for high-throughput evaluation of engineered cartilage destined for implantation, and suggest that mechanical release criteria for engineered cartilage should include an assessment of anisotropy as a metric for determining the similarity of tissue-engineered cartilage to native tissue [25,13,26]. Ultrasound allows for non-invasive/non-destructive evaluation of tissue morphology and defect detection, and it has been used to assess mechanical properties of the tissue [27–32]. Patil et al. previously demonstrated anisotropy in the speed of sound measured through bovine AC in directions tangential (parallel) and normal (perpendicular) to the articular surface [33]. In this paper, we demonstrate a simple apparatus and method, which will allow high-throughput screening of normal and TE cartilage specimens. A particular advantage of our approach is the use of custom-designed specimen holders, which allow rapid measurement of SOS in orthogonal directions on the same tissue specimen. Moreover, we investigated freeze/thaw cycling to see whether it affected the anisotropy and hence mechanical properties of native articular cartilage

2. Materials and Methods

2.1. Test sample preparation

Knees from two normal bovine calves were obtained from a local butcher shop (Fig. 1). Osteochondral cylinders (Samples 1–9 from animal #1 and samples 10–17 from animal #2) were cut from the condyles using a 6-mm diameter coring tool under continuous irrigation with phosphate-buffered saline (PBS). The cylinders were then cut across their diameter using a razor blade and frozen at −20 °C until use. For testing, the samples were thawed at room temperature in PBS for 30 minutes prior to the test [34]. Using a razor blade, the bone layer was carefully removed from the cartilage, and a 1–2 mm wide strip of cartilage was then cut perpendicular to the articular surface (parallel to the diameter cut). Surfaces were kept as parallel and perpendicular to each other as reasonably possible. This resulted in a full-thickness strip of cartilage about 6 mm long (Fig. 1), with orthogonal faces tangential (parallel) and normal (perpendicular) to the articular surface. Similar to Jurvelin et al., the orientation of the strip was not controlled with respect to split lines [35]. The samples were kept moist with PBS during the entire preparation process.

Fig. 1.

Tissue sample preparation: Bovine osteochondral samples were cored using a 6 mm diameter diamond coring tool. The cylinders were then cut across their diameter to make semi-cylinders. Finally, the adherent bone was removed resulting in a strip of cartilage with two orthogonal faces tangential and normal to the articular surface

2.2. Sample holder

L-shaped sample holders were machined from magnetic stainless steel (416 Alloy, McMaster-Carr, Cleveland, OH, Fig. 2A&B). The two flat polished orthogonal inside surfaces of the holder also serve as ultrasound reflectors. The cartilage strip was placed lengthwise in the inside angle of the sample holder (Fig. 2C), then the ends were glued to the holder (cyanoacrylate, Loctite 4011, McMaster-Carr, Cleveland, OH, Fig. 2D), leaving a glue-free testing region while holding the sample tightly against the holder’s orthogonal faces. To distinguish directions during the test, the sample and the holder side-wall were marked with ink to identify the articular surface (Fig. 2D).

Fig. 2.

Sample holder design. A) Two dimensional CAD drawing of the L-shaped sample holder. B) A 3D representation of the sample holder. Dimensions are in millimeters. C) Front view of finished device with cartilage sample glued on the edges to the holder. D) Side view of the sample. Articular surface is marked in blue. Ruler indices are 500 μm apart

All samples were photographed in two orthogonal directions using a stereomicroscope (EZ4W, Leica Microsystems, Buffalo Grove, IL, Fig. 2C&D) and the thickness in each direction were measured using imageJ [36].

2.3. Testing device

This L shaped holder and cartilage sample assembly was then placed in the ultrasound testing device (Fig. 3). This is a variant of the device we have reported previously [37,38], modified to include rare-earth magnets for positioning and securing the assembly in the ultrasound device. This system allows the assembly to be repositioned so that SOS measurements can easily be made in either of two perpendicular directions. The US testing device is also equipped with an X-Y stage (Edmund Optics Inc., Barrington, NJ) to position the transducer with respect to the sample (Fig. 3). This allowed us to record echoes from the metal sample holder and the cartilage surfaces by moving the US transducer. Another positioning stage (M4004-DM, Parker, Cleveland, OH) was used to position the sample holder in the Z direction (Fig. 3).

Fig. 3.

Ultrasound rig incorporating a magnetic sample holder and a focused ultrasound transducer

Non-contact ultrasound evaluation of the samples was performed using a focused transducer (Olympus, 30 MHz, model V375-SU, nominal 19 mm focal length, beam width 150 μm, Fig. 3). Focal length and depth of field (4 mm) were confirmed experimentally for this transducer, using ASTM E1065/E1065M [39]. A pulser-receiver operating in transmit/receive mode provided excitation and received reflections from the sample (DPR500, JSR Ultrasonics, Imaginant, Pittsford, NY). Pulser-receiver parameters were: transmit/receive, gain: 0 dB, low pass filter: 300 MHz, high pass filter: 5 MHz, pulse repeat frequency: 200 Hz, excitation: 275V, and damping: 44Ω. Signals were sampled at 5 Gs/s using a digital oscilloscope (Keysight MSOX3054T).

2.4. Procedure

In all cases, round trip time of flight (TOF) between the US transducer and a surface of interest were measured using the oscilloscope cursors.

The holder and sample were first placed, submerged in PBS, at the focal distance of the transducer (~19mm), but outside of the US transducer beam, then, by using the X-Y stage, the US transducer was moved laterally until the metal surface of the holder was detected (Fig. 4 – T₄).

Fig. 4.

TOF measurements, implementing similar method used in previous studies. [33,30] Panel A: T₁ time from front face of transducer to top surface of cartilage. T₂ time to bottom surface of cartilage. Panel B: determination of SOS through PBS T₃ time to holder through PBS. T₄, time to holder after displacing the sample holder by 2 mm (Green Arrow). a, b, and c: planes of the transducer face, and the front and back of the cartilage sample, respectively

We then identified the sample edges and centered the US beam on the width of the cartilage sample. Z-position was fine adjusted to maximize the amplitude of the first reflection off the cartilage. At this position, two echoes, one from the surface of the cartilage (Fig. 4 – T₁), the other from the surface of the metal holder after passing through the cartilage strip (Fig. 4 – T₂) were recorded. Fig. 5 shows an example of the reflected signal from the front and back surfaces of the cartilage sample (red line) and from the surface of the metal holder without cartilage.

Fig. 5.

Example of the reflected signal waveform from the front (A) and back (B) surfaces of the cartilage sample (red line, corresponds to positions b and c in Fig. 4) and from the surface of the metal holder without cartilage (black line). Note leftward shift in red signal trace from the back of the sample/metal holder owing to the higher SOS in cartilage compared to PBS.

After these measurements, the sample holder and sample were rotated 90 degrees, and the procedure was repeated to measure the same TOFs in this direction.

To compute the SOS in cartilage, the SOS in PBS (C_PBS) was needed. We did this by first measuring the time T₄, which is the TOF of the echo from the back of the sample holder. The holder was then displaced 2 mm down and the time T₃ was measured.

$$C_{PBS}=\frac{2\left(2\text{\hspace{0.17em}mm}\right)}{T_3-T_4}$$ (1)

The thickness of a sample D_bc was then found from

$$D_{bc}=\frac{C_{PBS}\left(T_3-T_1\right)}{2}$$ (2)

Using equations 1 and 2, the SOS in a sample C_S is then

$$C_s=\frac{2D_{bc}}{T_2-T_1}=\frac{T_3-T_1}{T_2-T_1}{C}_{PBS}$$ (3)

Note that it is not actually necessary to compute the thickness to get the SOS in AC.

2.5. Freeze-thaw experiments

Because freeze-thaw cycling of the cartilage has been shown to alter some mechanical properties of AC, we subjected a subset of the samples (N=9) to 10 freeze-thaw cycles to see if changes in acoustic anisotropy could be detected. The freezing was done with the tissue mounted on the holder, alternating between LN₂ for 30 seconds and RT PBS for 90 seconds repeated 10 times. After this, the measurement described above were repeated.

2.6. Statistical analysis:

Comparisons of SOS in normal versus tangential directions were made using Wilcoxon signed rank tests for paired data. The percent difference between SOS in X = normal and Y = tangential directions was calculated for each sample as 100*(X-Y)/(mean(X,Y)) and a 95% confidence interval for the mean percent difference was calculated. Thickness measurements obtained from stereomicroscope and ultrasound were also compared to those measured using US using the Wilcoxon signed rank tests.

3. Results

The speed of sound in PBS was measured in each test (Table 1), and was used in the calculation for SOS and thickness for each sample. Lab temperature was maintained at 21 ± 1.5° C across experiments, however within each sample, all measurements were completed in a 30 minute period during which the temperature remained constant. In each of the samples, SOS was greater (average 1,758.3 ± 106.7 m/s) in the direction perpendicular to the articular surface compared to the direction parallel to the articular surface (average 1,617 ± 55.1 m/s P = 0.00001, Wilcoxon signed rank test, Table 1). The average percentage difference in SOS between the normal and tangential directions was 8.2% (95% CI: 5.4% to 11%).

Table 1:

Anisotropy test results. C₁ and C₂ are SOS normal and tangential to the articular surface, respectively.

Sample #	C1 (m/s)	C2 (m/s)	PBS SOS (m/s)
1	1774.35	1754.44	1504.43
2	1883.40	1606.00	1514.39
3	1818.00	1568.80	1498.12
4	1939.01	1656.40	1498.12
5	1731.50	1576.06	1514.39
6	1673.23	1641.23	1513.43
7	1954.19	1632.42	1520.51
8	1903.03	1681.81	1522.43
9	1687.40	1656.00	1528
10	1722.00	1591.00	1525
11	1633.60	1608.80	1506
12	1793.80	1572.70	1495
13	1655.70	1518.50	1497.20
14	1626.02	1588.03	1491.19
15	1723.10	1590.4	1512.6
16	1686.10	1582.4	1499.9
17	1686.17	1664.86	1491.42
Mean ± SD	1758.27 ± 106.72	1617.05 ± 55.09	1507.78 ± 12

Ultrasound-computed thicknesses were within 4.9% of optical thickness measurements using a stereomicroscope with no statistically significant difference found (N=8, sign test p=0.28, and signed ranked test p=0.10).

For the subset of 9 samples which were subjected to freeze-thaw cycles, we observed small changes in SOS and thickness of the samples before and after the process. Specifically, the thickness changes before and after the freeze-thaw cycling were −5.4% normal to the articular surface, and 4.5% tangential to it.

SOS changes before and after freeze-thaw cycling were not statistically significant in both directions (P_normal = 0.73, P_tangential = 0.43). However, the cycling affected the thickness differently. As such, the thickness significantly changed in the direction, parallel to the articular surface (P = 0.008), while the thickness variations remained insignificant normal to the articular surface (P = 0.054). Consistent with this result, the SOS perpendicular to the AC surface was still greater than and significantly different from that parallel to the surface after ten freeze-thaw cycles (P = 0.008). Results are summarized in Table 2.

Table 2:

SOS in cartilage in 2 orthogonal directions (C₁ normal to articular surface and C₂ tangential to articular surface) before and after 10 freeze-thaw cycles

Sample #	C1 before (m/s)	C1 after (m/s)	C2 before (m/s)	C2 after (m/s)
9	1687.4	1758	1656	1613
10	1722	1669	1591	1589
11	1633.6	1634.3	1608.8	1603.8
12	1793.8	1688.4	1572.7	1579.4
13	1655.7	1688.8	1518.5	1505.5
14	1626.02	1678.8	1588.03	1591.1
15	1723.1	1737.9	1590.4	1667.2
16	1686.1	1647.9	1582.4	1648.1
17	1686.17	1765.2	1664.86	1711.1
Mean ± SD	1690.43 ± 51.69	1696.48 ± 46.96	1596.97 ± 43.8	1612.02 ± 58.75

4. Discussion

Failure to recapitulate the structural anisotropy of native cartilage may be a contributing factor in the failures of engineered cartilage [24]. This investigation is part of a series of studies [27,40,37,29] to implement non-destructive evaluation of the mechanical properties of cartilage using ultrasound. In this investigation, we sought to develop and validate a rapid technique for assessing the anisotropy of acoustic properties of cartilage using normal native bovine tissue. The goal is to then use this anisotropy as a high-throughput metric for assessing the progression of maturing engineered tissue toward the properties of native AC.

Ultrasound techniques have been demonstrated for noninvasive evaluation of tissue morphology and elastography and for biomechanical assessment [29,30,41–45]. It is known that mechanical and acoustic properties of tissues are not entirely independent: for a plane wave, SOS is a function of a material’s modulus and density [46]. Walker et al. showed a positive correlation between the aggregate modulus (H) derived from ultrasound measurements and Young’s modulus (E) derived from a one-dimensional, mechanical, unconfined compression test in hydrogels [37,40].

For this study, we developed a simple experimental apparatus and method for measuring SOS. Key goals were to simplify and accelerate the test as a screening tool, and to establish values for the SOS in normal AC to be targeted for engineered tissues. The approach we developed, including the L-shaped sample holder (Fig. 2A&B) allowed us to minimize handling of the tissue, and, by simply turning the L-shaped holder 90 degrees (Fig. 2) the SOS could easily be measured in two orthogonal directions on the same tissue sample.

In all cases in this study, the SOS was significantly higher in the direction normal to the articular surface than tangential to the articular surface (Table 1). These results are consistent with those found by Patil et al. [33]. Based on our and others’ SOS measurements, and linear models for wave propagation in solids, it would seem reasonable to conclude that the modulus of AC should be higher in the direction normal to the articular surface than tangential to the surface. However, mechanical measurements of aggregate modulus would appear to contradict this conclusion [35]. Nevertheless, direct comparisons between bulk mechanical properties and those implied from US may be possible. It should be noted that, where Jurvelin’s measurements are averaged over cylinders that were 1.7 mm in diameter and 1 mm thick, our measurements are at a 150 μm region near the center of the thickness. It is known that, in AC, Young’s modulus increases as a function of depth from the surface [47–51]. Poisson’s ratio, on the other hand peaks in the middle region in bovine AC [48]. Similarly, there are indications that tissue density varies rapidly as a function of depth [52]. Taken together, these findings suggest that there is not a simple relationship between acoustic and mechanical anisotropy, and factors such as structure may need to be incorporated into models that relate acoustic and mechanical properties of AC.

In the approach employed by Patil et al. SOS was measured in sublayers cut parallel to the articular surface, not in full thickness samples as done in this investigation. While evaluating the SOS in sublayers can demonstrate depth-dependent acoustic properties, a consequence of slicing the tissue is that these properties tangential to the surface are measured on separate samples from those used to measure the normal properties. As a screening tool, measuring the SOS in two orthogonal directions on the same sample at the same cross section has advantages, notably that less material is required and less preparation time is involved. Further, in our system, two of the tissue sample’s surfaces were in direct contact with the L-shaped holder; thus, there was no acoustic path between the sample and reflector. This resulted in one less TOF measurement than required in the approach used by Patil.

Cartilage can be stored briefly at 4°C [53], but cryostorage of tissues may be used to preserve them for subsequent implantation, or simply for logistical reasons when working with large numbers of samples. It seems plausible that freezing affects the properties of cartilage, and a number of studies have addressed its effects on, e.g., biomechanical, dimensional, electromechanical, MRI properties. Previous studies of freezing of cartilage have produced results that are inconsistent [54] and vary from “no change” [55–59] to “detectable changes” [60,61,54,62,63] depending on the properties measured, with a variety of freezing/thawing and cryostorage protocols likely contributing to the variable outcome. To determine whether acoustic properties were altered by freeze-thawing, we subjected a subset of our samples to a set of ten freeze-thaw cycles in LN₂. In our study (N=9), there does not appear to be a significant change in SOS before and after 10 freeze-thaw cycles in either direction. Further, the measured SOS were within the range measured in other studies [64,33]. Thus, temporary frozen storage does not appear to affect cartilage acoustic parameters; as a result, samples could be mounted on the holders and stored for subsequent measurements.

5. Limitations

Limitations of the approach are as follows.

• Since the surface of the cartilage strip may not be flat, it is, therefore, sometimes necessary to scan to find a good surface for analysis

• We believe it is important to ensure that the tissue is in intimate contact with reflector. If there were two signals, suggesting a gap between the sample and the holder, the sample removed andre-attached to the sample holder.

• Although thicknesses might not have been measured at the exact same location using US and microscopy overall, a comparison of the acoustic and optical thickness measurements did not reveal any significant differences.

• The orientation of the strip of cartilage was not controlled with respect to split lines [35].

• Similar to other studies, we removed the cartilage from its subchondral bone which can cause swelling and curling deformation in AC samples [65,66]. This affects the structure of cartilage and may require further study in the future.

6. Conclusions

A rapid method for measuring SOS in two orthogonal directions was developed. A key component is an L shaped sample holder, which allowed us to have a fast test procedure and repeatable results. Fewer measurements were required than suggested by previous studies, which reduces sample preparation time and the stacking of experimental error and resulted in a simple and rapid assessment of tissue anisotropy. Using this high-throughput approach, anisotropy of SOS in cartilage was found to be qualitatively the same as in previous work and to be largely unaffected by freeze-thawing of the tissue.