Assessment of subchondral bone microdamage quantification using contrast‐enhanced imaging techniques

Abstract

Bone microdamage is common at subchondral bone (SCB) sites subjected to repeated high rate and magnitude of loading in the limbs of athletic animals and humans. Microdamage can affect the biomechanical behaviour of bone under physiological loading conditions. To understand the effects of microdamage on the mechanical properties of SCB, it is important to be able to quantify it. The extent of SCB microdamage had been previously estimated qualitatively using plain microcomputed tomography (μCT) and a radiocontrast quantification method has been used for trabecular bone but this method may not be directly applicable to SCB due to differences in bone structure. In the current study, SCB microdamage detection using lead uranyl acetate (LUA) and quantification by contrast‐enhanced μCT and backscattered scanning electron microscopy (SEM) imaging techniques were assessed to determine the specificity of the labels to microdamage and the accuracy of damaged bone volume metrices. SCB specimens from the metacarpus of racehorses, with the hyaline articular cartilage (HAC) removed, were grouped into two with one group subjected to ex vivo uniaxial compression loading to create experimental bone damage. The other group was not loaded to preserve the pre‐existing in vivo propagated bone microdamage. A subset of each group was stained with LUA using an established or a modified protocol to determine label penetration into SCB. The μCT and SEM images of stained specimens showed that penetration of LUA into the SCB was better using the modified protocol, and this protocol was repeated in SCB specimens with intact hyaline articular cartilage. The percentage of total label localised to bone microdamage was determined on SEM images, and the estimated labelled bone volume determined by μCT in SCB groups was compared. Label was present around diffuse and linear microdamage as well as oblique linear microcracks present at the articular surface, except in microcracks with high‐density mineral infills. Bone surfaces lining pores with recent mineralisation were also labelled. Labelled bone volume fraction (LV/BV) estimated by μCT was higher in the absence of HAC. At least 50% of total labels were localised to bone microdamage when the bone area fraction (B.Ar/T.Ar) of the SCB was greater than 0.85 but less than 30% when B.Ar/T.Ar of the SCB was less than 0.85. To adjust for LUA labels on bone surfaces, a measure of the LV/BV corrected for bone surface area (LV/BV BS⁻¹) was used to quantify damaged SCB. In conclusion, removal of HAC and using a modified labelling protocol effectively stained damaged SCB of the metacarpus of racehorses and represents a technique useful for quantifying microdamage in SCB. This method can facilitate future investigations of the effects of microdamage on joint physiology.

Microdamage can affect the biomechanical behaviour of bone under physiological loading conditions but to understand the effects of microdamage on the mechanical properties of subchondral bone (SCB), it is important to be able to quantify it. The μCT and SEM images of subchondral bone specimens stained with lead uranyl acetate showed localisation of label to bone microdamage and bone surfaces where there were no damage. Accounting for the presence of label on bone surfaces is important when using contrast‐enhanced imaging method to quantify pre‐existing microdamage in SCB.

1. INTRODUCTION

Bone microdamage is common at subchondral bone (SCB) sites subjected to high rate and magnitude of loading (Coughlin & Kennedy, 2016; Muir et al., 1999; Turley et al., 2014; Winters et al., 2019). Under normal loading conditions, bone microdamage is repaired by bone remodelling during which the damaged bone matrix is removed by osteoclasts and replaced by a new matrix produced by osteoblasts (Boyde, 2003; Burr et al., 1985). When SCB is subjected to repetitive high magnitude loading during high‐impact activities, bone remodelling is suppressed resulting in an accumulation of microdamage at focal sites (Boyde, 2003; Holmes et al., 2014). This can lead to the propagation of fatigue fracture or the collapse of the articular surface and serious joint diseases such as osteoarthritis (Norrdin & Stover, 2006; Winters et al., 2019). In the metacarpophalangeal joints of racehorses, the palmar aspect of the distal condyles of the third metacarpal bone and the articular surface of the proximal sesamoid bones are common sites of bone microdamage accumulation due to repetitive extreme forces applied during high‐speed exercise (Ayodele et al., 2021; Harrison et al., 2014; Muir et al., 2008; Whitton et al., 2018). These bone sites are also commonly prone to fatigue bone injuries some of which can be fatal. Subsequent failure of the SCB suggests that the accumulation of bone microdamage is an important change associated with the loss of SCB's mechanical integrity.

The mechanical properties of bone, that is, bone's ability to withstand load, are primarily determined by bone microstructure (porosity and bone volume), and bone material composition (the mineral content, collagen crosslinks and the non‐collagenous components) (Bailey et al., 2004; Seeman & Delmas, 2006). In addition, the presence of pre‐existing damage can influence the mechanical properties (stiffness and energy loss) of bone as shown in trabecular and cortical bone (Burr et al., 1998; Hernandez et al., 2014; Karim & Vashishth, 2011; Lambers et al., 2013; Malekipour et al., 2020). The effect of accumulated bone microdamage on the mechanical properties of SCB remains poorly understood.

In order to understand the effects of microdamage in SCB, it is important to be able to quantify it. Current studies on the effect of pre‐existing bone microdamage on the mechanical properties of SCB have qualitatively scored bone microdamage based on microscale changes that are visible on plain micro‐computed tomographic images (μCT) (Malekipour et al., 2018, 2020). This method is limited to detecting large microcracks, where separation of the crack surfaces is greater than the spatial resolution of the μCT, or that contain highly calcified material that provides contrast with surrounding bone matrix (Bouxsein et al., 2010; Boyde, 2003). Bone microdamage can be quantified histologically using bulk‐staining or scanning electron microscopy (SEM). These methods offer high resolution but are destructive, limited to two dimensions and require long processing times (Burr & Hooser, 1995; Norrdin & Stover, 2006). Contrast‐enhanced μCT techniques, using lead uranyl acetate (LUA) or barium sulphate as radiocontrast microdamage labels, have been used in trabecular and cortical bone for three‐dimensional quantification of bone microdamage (Brock et al., 2013; Hernandez et al., 2014; Lambers et al., 2013; Schaffler et al., 1994; Tang & Vashishth, 2007). We and others have utilised LUA to label bone microdamage in SCB from the equine fetlock joint; however, the suitability of this method for quantifying SCB has not been determined (Luedke et al., 2022; Pearce et al., 2022). In SCB, the hyaline articular cartilage adheres at the articular surface and is adjacent to the site from where microcracks propagate. Likewise, the SCB densification that occurs during adaptation to exercise is accompanied by reorganisation of the bone matrix leading to a rather more complex SCB structure compared to trabecular and diaphyseal cortical bone. Therefore, an assessment of the effectiveness of using LUA for microdamage detection and contrast‐enhanced μCT technique for bone microdamage quantification in SCB is necessary. The two aims of this study were to determine the penetration and localisation of LUA in dense equine SCB using two different labelling protocols and to determine the accuracy of contrast‐enhanced imaging for quantifying damaged bone volume in SCB of various bone densities by comparing μCT and SEM images.

2. MATERIALS AND METHODS

2.1. Material

Cylindrical SCB specimens (6 mm diameter and 10 mm depth; n = 22) were obtained from the palmar aspect of the distal condyles of third metacarpal bones of 15 Thoroughbred racehorses (8 right and 7 left limbs). Of the 15 horses, 6 were female, 2 entire males and 7 geldings. The mean age was 2.9 ± 2.1 years (range 1–8 years) with a mean race start of 7.9 ± 15.2 (range 0–48). Subchondral bone core drilling was performed as previously described (Martig et al., 2013). Specimens were wrapped in gauze soaked in 0.9% saline and stored frozen at −20°C. Specimens were divided into different groups (Figure 1), to determine the effect of different staining methods or presence of hyaline cartilage. The hyaline articular cartilage (HAC) was excised with a scalpel in 12 specimens but kept intact in 10 specimens to investigate if HAC affected the penetration of radiocontrast agent and labelling of SCB microdamage. Six out of the 12 specimens from which the HAC had been excised were from a previous study that applied cyclic loading to specimens until failure (Shaktivesh et al., 2020). Specimens were subjected to ex vivo loading on the articular surface to create experimental bone damage. Briefly, specimens were tested at room temperature in unconfined compression (applied in load‐control mode) under a sequence of varying load amplitudes and frequencies simulating that of a training racehorse. The sequence of dynamic loads composed of a series of compression‐compression sinusoidal loading cycles with load amplitudes and frequencies ranging from 32 to 91 MPa and 1 to 2 Hz. This sequence was applied to the specimens repeatedly until failure, that is propagation of macroscopic fracture. The remaining six specimens with excised HAC were not subjected to ex vivo loading. Ten specimens with intact HAC were also not subjected to ex vivo loading therefore preserving the pre‐existing in vivo bone microdamage in those specimens.

FIGURE 1.

A chart showing the subchondral specimens used in this study and their distribution into different groups. SCB, subchondral bone.

2.1.1. Microdamage labelling

Specimens were labelled with LUA using two incubation methods to determine how preparation of reagents affected the penetration of LUA into SCB (Table 1). The specimens with excised HAC were divided into two groups such that two specimens with similar pre‐label bone volume fraction (BV/TV) were assigned to different groups (each group had either three specimens with excised HAC subjected to loading or three specimens with excised HAC and not subjected to loading). One group was labelled using a LUA incubation method described previously (Method 1) (Schaffler et al., 1994). In this method, the LUA reagent was prepared by mixing 8% uranyl acetate in 70% acetone with 20% lead acetate in 70% acetone. The specimens were incubated with shaking for 14 days and under vacuum for 12 h on the first day. The second group of specimens was incubated in 8% uranyl acetate in 70% acetone for 7 days, washed briefly and then incubated in 20% lead acetate in 70% acetone for an additional 7 days with shaking and under vacuum for 12 h on the first day of each incubation (Method 2). The latter method produced better penetration of LUA into the SCB (discussed in Section 3) and was repeated in specimens with intact HAC that were not subjected to ex vivo loading. After the final staining procedures, all specimens were washed in 70% acetone for 14 days with sonication for 30 min each day, followed by incubation in 1% ammonium sulphide in acetone for 1 day and a final wash in 70% acetone.

TABLE 1.

The groups of SCB specimens, the methods used for lead uranyl acetate labelling and the μCT scan settings used in this study.

Samples	Labelling method	μCT scan settings
3 SCB with excised HAC (non‐loaded) 3 SCB with excised HAC (loaded)	Method 1: Incubated with a mixture of uranyl acetate (8% w/v) and lead acetate (20%) in 70% acetone for 14 days.	Pre‐label: 70kVp, 200 μA, integration time‐600 ms, 0.5 mm aluminium filter 10 μm3 voxel. Post‐label: 70kVp, 200 μA, integration time‐600 ms, 0.5 mm copper filter and 5 μm3 voxel.
3 SCB with excised HAC (non‐loaded) 3 SCB with excised HAC (loaded)	Method 2: Incubated with uranyl acetate (8%) in 70% acetone for 7 days, then lead acetate (20%) in 70% acetone for 7 days.
10 SCB with intact HAC (non‐loaded)		Pre‐ and post‐label: 70kVp, 200 μA, integration time‐600 ms, 0.5 mm aluminium filter and 4 μm3 voxel.

Abbreviations: HAC, hyaline articular cartilage; μCT, microcomputed tomography; SCB, subchondral bone.

2.1.2. Micro‐computed tomography

Images of all specimens were obtained by μCT (μCT50, SCANCO Medical, Brüttisellen, Switzerland) in transverse sections using scan settings shown in Table 1. Pre‐label scan (obtained prior to LUA staining and embedding) for specimens with excised HAC was performed at isotropic voxel of 10 μm³ and X‐ray settings: 70 kVp, 200 μA, 0.5 mm aluminium filter and 600 ms integration time. A post‐label scan (obtained after LUA staining but prior to embedding) for these specimens was performed at 5 μm³ voxel and X‐ray settings: 70 kVp, 200 μA, 0.5 mm copper filter and 600 ms integration time. The pre‐ and post‐label scans for specimens with intact HAC were performed at 4 μm³ voxel and X‐ray settings: 70 kVp, 200 μA, 0.5 mm aluminium filter and 600 ms integration time.

2.1.3. Back‐scattered scanning electron microscopy

Specimens that had been stained with LUA were embedded in polymethyl methacrylate and sectioned along their long axis using a diamond blade on a low‐speed precision saw (Isomet 1000, Buehler, Lake Bluff, IL, USA). Block surfaces were polished with diamond suspension and carbon coated under high vacuum (Safematic CCU‐010, Switzerland). SEM images were then obtained using a back‐scattered detector and electron beam voltage of 20 kV and current of 20 nA at 10 mm working distance and 250 times magnification (FEI Teneo Volumescope, Thermo Fisher Scientific, Hillsboro, OR, USA). Overlapping SEM image tiles were stitched using MAPS (ThermoFisher Scientific) and differences in the penetration and localisation of LUA labels in bone matrix were assessed visually between incubation methods and with or without HAC.

2.1.4. Quantification of damaged bone

Bone microdamage was quantified as uptake of LUA in the SCB by μCT and SEM for specimens labelled using Method 2.

Damaged bone volume was quantified on μCT images for two regions using CT Analyser (Skyscan CTAn version 1.3.11.0, Bruker). The cylindrical volume of interest was defined by bone diameter, 1 mm from the cut‐edge and to a depth of 2 mm from the bone/cartilage interface for the superficial region or to 5 mm for the whole SCB (Figure 2a). These ROIs are representative regions of bone microdamage presence (superficial region) and bone areas commonly selected for assessment of the mechanical properties (whole SCB) (Malekipour et al., 2020; Rubio‐Martinez et al., 2008). Gaussian blur, smoothing and despeckle were applied and segmentation was performed using global threshold settings. Minimum threshold values, determined to be representative by visual inspection, were set individually for specimens with excised HAC (minimum greyscale values: 20–30 for bone, and 45–150 for labelled bone). Fixed minimum threshold values were used for specimens with HAC (minimum greyscale values: 60 for bone, and 150 for labelled bone). For the superficial region and whole SCB, bone volume (BV), total volume (TV), labelled bone volume (LV) and bone surface area (BS; in pre‐label images) were determined. A fraction of bone tissue that is labelled (LV/BV) was calculated and to adjust for non‐specific labels present on bone surfaces, labelled bone volume fraction corrected for bone surface area (LV/BV BS⁻¹) was also determined.

FIGURE 2.

An illustration of the regions of interest selected for measurement of labelled bone volume on micro‐computed tomographic images (a) and for the measurement of labelled bone area on SEM images (b). SCB, subchondral bone; SEM, scanning electron microscopy.

Labelled bone area was determined on SEM images for two areas of interest defined by a width extending 1 mm from the specimen edge and 2 mm deep for superficial regions and 5 mm for whole SCB, starting from the articular cartilage‐bone interface (Figure 2b). Smoothing, median filter of 3.0 and despeckle was applied and images were segmented using thresholds set by visual inspection of individual images. Labelled bone area (L.Ar; total area of bone matrix labelled by LUA) and bone area (B.Ar; area of labelled and unlabelled bone) were determined using ImageJ (ImageJ, NIH, USA). Furthermore, bone surface labels that were not associated with microdamage or crack lines were manually erased using a paintbrush tool and the area of labelled bone associated with bone microdamage (D.Ar) was determined. The percentage of total labelled area associated with bone microdamage (D.Ar /L.Ar ×100) was determined and compared between specimens, based on their bone area fractions (B.Ar/T.Ar).

2.1.5. Statistical analyses

Measures of bone microdamage (LV/BV; LV/BV BS⁻¹) and BV/TV were presented as mean ± SD and compared between groups. Due to the small number of samples in each group, statistical comparison was performed using a two‐tailed Mann–Whitney U test. To determine if there were differences due to damage propagated by ex vivo experimental loading, a comparison was made between the group of specimens with excised HAC that were loaded and the specimens with excised HAC that were not loaded. To determine if there were differences due to the presence of HAC, a comparison was made between the group of specimens with excised HAC that were not loaded and specimens with intact HAC that were not loaded.

3. RESULTS

3.1. Penetration of label

Lead uranyl acetate was detectable as intense bright labels in the bone matrix on μCT and SEM images.

On SEM images, specimens with excised HAC that were not subjected to loading, when labelled using Method 1, had incomplete distribution of LUA on bone surfaces (Figure 3a), compared to similar specimens labelled using Method 2 which had complete distribution of LUA on bone surfaces across the bone (Figure 3b).

FIGURE 3.

Back‐scattered SEM images of specimens labelled with lead uranyl acetate. (a, b) Specimens with excised articular cartilage that were not subjected to ex vivo uniaxial compression loading. (c, d) Specimens with excised articular cartilage that were subjected to ex‐vivo uniaxial compression loading. (e) Specimen with complete HAC layers, and (f) Specimen with disrupted superficial HAC layer. (a) and (c) were prepared using a known labelling method (Method 1). (b), (d), (e) and (f) were prepared using a modified labelling method (Method 2). HAC, hyaline articular cartilage; SEM, scanning electron microscopy.

The specimens with excised HAC that were subjected to loading had LUA predominantly in bone matrix surrounding crack lines. When labelling was performed using Method 1, these samples had unevenly distributed LUA with more intense label around the outer part of the specimens and faint towards the centre (Figure 3c). Specimens with excised HAC that were subjected to loading, when labelled using Method 2, had uniformly intense LUA distributed around crack lines across the specimens (Figure 3d).

The distribution of label was variable in specimens with intact HAC. Specimens with complete HAC layers (n = 4) had variable distribution of LUA in the SCB with two specimens having less uniform LUA distribution in SCB (Figure 3e), whereas specimens with partial HAC loss (n = 6) had more uniformly distributed LUA in SCB (Figure 3f). Specimens with larger pores had some LUA uptake in the external pores (Figure 3e,f). On closer observation, this LUA uptake was due to filling of the outer large pores by bone particles that were lodged in the pores during bone core drilling.

3.2. Localisation of label

Lead uranyl acetate label was present in the bone matrix lining discontinuities (microcracks) and areas of bone matrix with no visible linear gaps (diffuse microdamage) in the superficial region of specimens that were not subjected to loading (Figure 4a). Similar pattern but more extensive damage was also seen in the superficial and deep regions of specimens that were loaded to failure (Figure 4b). Linear microcracks extending obliquely in the calcified cartilage and the superficial SCB were not identified in specimens with excised HAC that were not subjected to loading where the calcified cartilage had been removed.

FIGURE 4.

Corresponding region of specimens with excised HAC imaged with microCT and back‐scattered SEM imaging showing the localisation of labels with bone microdamage. Regions indicated by yellow boxes in low magnification SEM images are shown at higher magnification. LUA labels were present in areas with diffuse microdamage (green box on right), linear microcracks (yellow arrows), and on bone surfaces (green arrows) in both non‐loaded (a) and loaded samples (b). Partial calcified cartilage was present in (a) but absent in (b). Scale bar = 1 mm. HAC, hyaline articular cartilage; LUA, lead uranyl acetate; μCT, microcomputed tomography; SEM, scanning electron microscopy.

Localisation of LUA to microdamage in specimens with intact HAC is shown in Table 2 and representative images are presented in Figure 5. Linear microcracks extending obliquely in the calcified cartilage and the superficial SCB were identified in five specimens with the bone matrix surrounding these microcracks labelled in two specimens with partial loss of the articular cartilage but not in three specimens with intact HAC (Figure 5a). In three specimens, calcified microcracks that have bright radiocontrast material (high‐density mineral infills) within the space between linear bone discontinuities were present, but there was no halo of LUA label detected in the bone matrix surrounding these microcracks (Figure 5b). Diffuse microdamage present in the calcified cartilage and SCB was labelled in five specimens. Labels present around diffuse microdamage appear faint in two specimens with complete HAC layers whereas two specimens with partial HAC loss had more intense label (Figure 5c).

TABLE 2.

Specimens with intact hyaline articular cartilage and the assessment of lead uranyl acetate label around bone microdamage in the subchondral bone (SCB), diffuse microdamage present in SCB and linear microcracks at joint surface.

	Hyaline articular cartilage	Diffuse microdamage	Linear microcracks
			Non‐calcified	Calcified
SCB 1	Partial	ND	Labelled	ND
SCB 4	Partial	Labelled	ND	ND
SCB 6	Partial	Labelled	Labelled	Not labelled
SCB 7	Partial	ND	ND	ND
SCB 8	Complete	Labelled	ND	ND
SCB 9	Complete	Faintly labelled	Not labelled	ND
SCB 10	Partial	ND	ND	ND
SCB 12	Partial	ND	ND	ND
SCB 14	Complete	ND	Not labelled	Not labelled
SCB 15	Complete	Faintly labelled	Not labelled	Not labelled

Abbreviations: Faintly labelled, microdamage/microcrack detected with less intense LUA; Labelled, microdamage/microcracks detected with bright intense LUA; ND, microdamage/microcrack not detected; Not labelled, microdamage/microcrack was detected with no LUA label.

FIGURE 5.

MicroCT and back‐scattered SEM images of specimens with hyaline articular cartilage (HAC) labelled with lead uranyl acetate (LUA) by Method 2. Regions indicated by yellow boxes in low magnification SEM images are shown at higher magnification. Hyaline articular cartilage was intact and oblique linear microcracks were not labelled (arrows in higher magnification image) (a). The superficial HAC layer was disrupted, calcified microcracks (arrow heads) were not labelled by LUA and a non‐calcified microcrack (white arrow) was labelled by LUA (b). The superficial hyaline cartilage was disrupted and an area of diffuse microdamage with LUA uptake is present in the subchondral bone adjacent to a low density bone area with no bone damage and no LUA uptake (c). Scale bar = 1 mm. μCT, microcomputed tomography; SEM, scanning electron microscopy.

In all specimens, LUA label was present on bone surfaces to a variable extent. A thicker layer of label was present on bone surfaces with lesser greyscale values, at the interface of the pre‐existing and newly formed bone matrix.

3.3. Quantification of damaged bone

Labelled bone volume measurements by μCT for each specimen are shown in Data 1. For both the whole SCB and the superficial region, LV/BV and LV/BV BS⁻¹ were higher in specimens with excised HAC (loaded and non‐loaded) than in the non‐loaded specimens with intact HAC (Table 3).

TABLE 3.

Labelled bone volumetric parameters derived for three SCB specimen groups labelled using sequential labelling method (Method 2).

	LV (mm3)	LV/BV	LV/BV BS−1 (mm−2)	BV/TV	BS (mm2)	BS/BV (mm−1)
Superficial region
SCB with excised articular cartilage and loaded (n = 3)	6.477 ± 2.40	0.190 ± 0.06a	10.824 ± 4.4a	0.939 ± 0.06	178.96 ± 18.0	5.318 ± 0.9a
SCB with excised articular cartilage and non‐loaded (n = 3)	1.956 ± 2.64	0.057 ± 0.08a,b	4.065 ± 4.18a.b	0.98 ± 0.03	103.88 ± 52.8	2.956 ± 1.6a,b
SCB with articular cartilage and non‐loaded (n = 10)	0.162 ± 0.19	0.005 ± 0.005b	0.386 ± 0.6b	0.964 ± 0.05	195.12 ± 303.9	5.744 ± 3.8b
Whole SCB
SCB with excised articular cartilage and loaded (n = 3)	15.485 ± 8.10	0.189 ± 0.11c	5.822 ± 3.91c	0.951 ± 0.01	338.01 ± 37.7	4.053 ± 0.4c *
SCB with excised articular cartilage and non‐loaded (n = 3)	4.565 ± 0.99	0.055 ± 0.017c,d	2.471 ± 1.23c,d	0.942 ± 0.09	287.61 ± 228.3	3.638 ± 3.3c,d
SCB with articular cartilage and non‐loaded (n = 10)	0.334 ± 0.33	0.005 ± 0.005d	0.135 ± 0.14 d	0.931 ± 0.08	508.59 ± 350.7	7.483 ± 5.7d

Note: Mean ± SD of values presented. Groups with the same letters (a, b, c, or d) are significantly different.

Abbreviations: BS/BV, bone surface per defined volume of bone; LV, labelled volume; LV/BV, labelled volume fraction; LV/BV BS⁻¹, labelled volume fraction, corrected for bone surface; SCB, subchondral bone.

^*Estimate was affected by loading.

Labelled and damaged bone area measurements (L.Ar, D.Ar/B.Ar) by SEM for specimens with HAC are shown in Table 4 and raw data are presented in Data 2. The percentage of labelled bone area associated with bone microdamage (D.Ar/L.Ar ×100) in specimens with whole SCB B.Ar/T.Ar greater than 0.85 (n = 7) was 61.13 ± 24.84 (range 12%–92%) for whole SCB, and 62.92 ± 22.52 (range 22%–92%) for the superficial region. One of these specimens (SCB 8) had a low percentage of labelled bone (16% for whole SCB and 22% for superficial region), and microdamage was not visible on the SEM image. The D.Ar/L.Ar percentage in specimens with B.Ar/T.Ar of the whole SCB less than 0.85 was 13.83 ± 2.03 (range 12%–16%) for whole SCB and 22.22 ± 6.3 (range 16%–28%) for the superficial region (Figure 6).

TABLE 4.

Labelled bone area parameters derived for the non‐loaded SCB samples with intact hyaline articular cartilage labelled using sequential labelling method (Method 2).

	Labelled bone area (L.Ar, mm−2)	Damaged bone area (D.Ar, mm−2)	D.Ar/L.Ar (%)	Bone area (B.Ar, mm2)	Total area (T.Ar, mm2)	L.Ar/B.Ar	B.Ar/T.Ar
Superficial region	0.210 ± 0.15	0.117 ± 0.13	50.71 ± 27.08	8.801 ± 1.13	9.56 ± 1.31	0.024 ± 0.018	0.922 ± 0.031
Whole SCB	0.738 ± 0.69	0.464 ± 0.64	46.94 ± 30.56	19.77 ± 2.18	22.60 ± 1.85	0.0367 ± 0.033	0.875 ± 0.073

Note: Mean ± SD of values presented.

Abbreviations: B.Ar, bone area; B.Ar/T.Ar, bone area fraction; D.Ar, damaged bone area; D.Ar/L.Ar %, proportion of total labelled area associated with bone microdamage; L.Ar, labelled bone area; L.Ar/B.Ar, labelled bone area fraction; SCB, subchondral bone; T.Ar, total tissue area.

FIGURE 6.

A scatterplot of the percentage of total bone area labelled that was localised to bone microdamage distributed by the bone area fraction (B.Ar/T.Ar) of the whole subchondral bone or the superficial region. SCB, subchondral bone.

4. DISCUSSION

The effectiveness of using LUA for detecting and quantifying bone microdamage in SCB was investigated in this study. Greater penetration of LUA into dense SCB was achieved using serial incubation of labelling reagents (modified protocol, i.e. Method 2). Both diffuse and linear microdamage and microcracks present in the SCB were labelled by LUA, except the linear microcracks extending obliquely from the calcified cartilage to the SCB in specimens with intact HAC or when the microcracks had been hypermineralised. In all specimens, LUA labelled the bone surfaces even when microdamage was not identified on μCT and BSEM images. Non‐specific bone surface labelling accounted for less than 50% of total label in the majority of specimens with greater bone density.

We achieved better penetration of LUA into the equine SCB specimens using sequential incubation with the labelling reagents. The reagent mixture that had been used in previous studies (repeated as Method 1 in this study) formed a precipitate that was viscous limiting the penetration of the contrast agent into the SCB. Equine SCB undergoes a substantial structural transformation from a trabecular to a cortical bone subtype due to adaptive bone modelling at the onset of race training (Boyde, 2003). Unlike cortical bone which has a high bone volume fraction with an organised Haversian system (layers of bone arranged concentrically around a central vascular pore and oriented parallel to each other with interconnections of the pores), the pores in the superficial SCB are disordered and the majority are closed (Martig et al., 2018). This morphology may limit SCB penetration of radiocontrast reagents.

Compared to specimens that had complete HAC layers, specimens with superficial HAC loss produced more consistent labelling of the microcracks in the superficial SCB suggesting that the HAC restricted the penetration of labels into this region. Where articular cartilage was intact, the superficial and middle layers were saturated with the LUA label which was also present at the hyaline‐calcified articular cartilage interface; however, there was no diffusion of materials across to the HAC‐bone interface. The articular cartilage with its high proportion of negatively charged proteoglycans may restrict the passage of positively charged label to the calcified cartilage‐bone interface due to high affinity for the label (Mow et al., 1992; Roughley & Lee, 1994).

In specimens that were not subjected to ex‐vivo loading (with or without HAC), LUA label was mostly localised in the superficial SCB, consistent with the location of pre‐existing diffuse and linear SCB microdamage propagated in vivo in equine SCB. Bulk‐staining with basic fuchsin and high‐resolution 2D imaging modalities have been used to demonstrate such in vivo SCB microdamage (Riggs et al., 1999; Whitton et al., 2018). However, oblique linear microcracks extending from the calcified cartilage to the SCB were observed in specimens with articular cartilage but not in specimens where HAC had been excised due to loss of the microcrack's profile with the excision of the calcified cartilage. Where calcified cartilage was partially retained or completely removed after excision, microcracks were not seen.

We observed that the linear microcracks that had been calcified had no label in their surrounding bone material even in specimens where the non‐calcified microcracks had been labelled. Calcified microcracks are older pre‐existing microcracks repaired with highly dense (mineralised) material infills (HDMI). These hypermineralised microcracks have been demonstrated in the metacarpal condyles, proximal sesamoid and carpal bones of racehorses. The presence of these HDMI likely seals these microcracks at the hyaline‐calcified cartilage interface preventing entry of labelling reagents (Boyde, 2021; Laverty et al., 2015; Whitton et al., 2018).

In addition to labelling bone microdamage, we observed LUA label uptake on bone surfaces in areas where microcracks are absent. Consistent with a previous finding in sheep rib cortical bone, a thicker layer of label was present along surfaces with recent mineralisation (Schaffler et al., 1994). The SCB from the distal metacarpus of a racehorse is an active site of modelling and remodelling with extensive osteonal and/or mineralising bone surfaces which can have greater affinity for LUA and result in the extensive bone surface labelling that we observed compared to human trabecular bone where surface labelling has not been reported (Holmes et al., 2014; Murray et al., 2001). This non‐specific surface label that is not localised to SCB microdamage is likely to affect the estimation of damaged bone volume measurement in SCB using contrast‐enhanced imaging techniques.

The specimens with excised HAC had greater LV/BV than those with intact HAC suggesting that more labels were retained in the SCB in the absence of HAC. This may be due to the limiting effect of HAC on the penetration of LUA into SCB which was consistent with our subjective observations. However, μCT scanning settings were different between the two groups with larger voxel size used in the specimens with excised HAC which may have contributed to an overestimation. In addition, the specimens with excised HAC had slightly greater BV/TV and were expected to have more pre‐existing bone microdamage than those with intact HAC as higher BV/TV is associated with more microdamage in equine SCB (Whitton et al., 2018).

We estimated that 50%–92% of the total LUA label was localised to bone microdamage in specimens with B.Ar/T.Ar greater than 0.85 but less than 20% or 30% for the whole SCB and superficial region respectively, when the specimen's B.Ar/T.Ar was less than 0.85. This implies that quantified damaged bone volume was more accurate in high‐density SCB where greater bone microdamage and less bone surface area are expected compared to less‐dense SCB, which is likely to have less bone microdamage and greater bone surface area. Whilst the bone fraction threshold of 0.85 was arbitrarily based on observations of the specimens from this study, the bone volume fraction of the distal metacarpal and proximal sesamoid bones of horses that are in race training are higher than 0.85 but less than 0.85 for untrained horses suggesting that quantitation of bone microdamage using LUA staining technique in bone specimens obtained from horses in race training will be more accurate than in specimens obtained from untrained horses (Ayodele et al., 2021; Martig et al., 2018). Similarly, the inclusion of the deeper region can also reduce the accuracy of estimating damaged SCB volume using this method due to its larger bone perimeter. Practically, the correlation of the bone microdamage in the superficial SCB and the mechanical properties of the superficial layer of the SCB or the whole SCB is likely to demonstrate the effect of the bone microdamage on the mechanical behaviour of the superficial SCB layer or the whole SCB.

The limitations of the current study should be considered. The percentage of total LUA label localised to bone microdamage was evaluated using high‐resolution 2D images obtained by back‐scatted SEM and therefore our estimation of the actual proportion of bone label associated with bone microdamage was limited to one plane surface. The μCT settings used for post‐label scan of specimens without HAC were different from their pre‐label scans and compared to specimens with HAC as the μCT imaging was optimised to reduce artefacts due to presence of LUA. Therefore, the comparison of damaged bone volume between these groups should be interpreted with caution. Beam hardening artefacts were present in μCT images obtained at higher voxel size (low resolution) but this was reduced in images obtained at lower voxel size (high resolution). The sample size for specimens with excised HAC was low (n = 3); statistical comparison of the DV/BV between the non‐loaded and loaded specimens was not significant but may have proven to be significant with a higher sample number.

The current study demonstrates that pre‐existing microdamage accumulated in the equine SCB is detectable with LUA using modifications of an existing method used previously in cortical bone. Whilst using radiocontrast imaging provides a method of detecting and quantifying SCB microdamage that is better than the subjective assessment of bone microdamage based on the lytic bone changes seen on plain μCT in bone areas surrounding microdamage, it is important to consider the sources of error or inaccuracy such as the penetration of the radiocontrast agent into the SCB and non‐specific binding of label to bone surface in normal bone matrix. Sequential incubation with the reagents and disruption of the overlying HAC are necessary for adequate penetration into highly dense SCB. The removal of the HAC by excising it distorted the microcrack profiles extending from the calcified cartilage into the superficial SCB in this study. Chemical digestion of the articular cartilage using NaOH may be suitable for removing the HAC without disrupting the microcracks. Hypermineralised microcracks were not labelled by LUA; however, their high‐density mineral infills provide high contrast to the surrounding bone tissue enabling these microcracks to be detected by microCT or SEM, and quantified together with the LUA‐labelled microdamage in SCB. The non‐specific presence of LUA labels on bone surfaces reduced the accuracy of labelled bone volume (LV/BV) estimates, especially in specimens with lesser bone volume fraction. Labelled bone volume fraction, corrected for bone surface area (LV/BV BS⁻¹), is an alternative metric to LV/BV in the volumetric quantification of bone microdamage in SCB which takes the non‐microdamage related labelling of bone surfaces into consideration during microdamage quantification in SCB. This is particularly important when specimens from different bone sites from horses with variations in racing or exercise history, such as different loading magnitude and extent of bone adaptation, are compared, in which case the BV/TV of some of the specimens may be less than 0.85 (Pearce et al., 2022).

In conclusion, the detection and quantification of bone microdamage in equine distal limb SCB by contrast‐enhanced imaging required sufficient penetration of the LUA used as label. This was achieved by removal of the HAC and by using a sequential incubation method. Accounting for the presence of label on bone surfaces is important during the volumetric quantification of SCB microdamage in specimens with a range of microstructural properties. This contrast‐enhanced imaging method will be useful for quantifying pre‐existing SCB microdamage during future investigations of the biomechanical behaviour of SCB.

AUTHOR CONTRIBUTIONS

BAA: concept/design, acquisition of data, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript and approval of the article; FM: concept/design, data analysis/interpretation, critical revision of the manuscript and approval of the article; CNP: critical revision of the manuscript and approval of the article; EJM: concept/design, data analysis/interpretation, critical revision of the manuscript and approval of the article; RCW: concept/design, data analysis/interpretation, critical revision of the manuscript and approval of the article.

Supporting information

Data S1:

Ayodele, B.A. , Malekipour, F. , Pagel, C.N. , Mackie, E.J. & Whitton, R.C. (2024) Assessment of subchondral bone microdamage quantification using contrast‐enhanced imaging techniques. Journal of Anatomy, 245, 58–69. Available from: 10.1111/joa.14035

DATA AVAILABILITY STATEMENT

The data that supports the findings of this study are available in the supplementary material of this article.