T2* mapping in an equine articular groove model: Visualizing changes in collagen orientation

Abstract T2* mapping is promising for the evaluation of articular cartilage collagen. In this work, a groove model in a large animal is used as a model for posttraumatic arthritis. We hypothesized that T2* mapping could be employed to differentiate between healthy and (subtly) damaged cartilage. Eight carpal joints were obtained from four adult Shetland ponies that had been included in the groove study. In this model, grooves were surgically created on the proximal articular surface of the intermediate carpal bone (radiocarpal joint) and the radial facet of the third carpal bone (middle carpal joint) by either coarse disruption or sharp incision. After 9 months, T2* mapping of the entire carpal joint was carried out on a 7.0‐T whole‐body magnetic resonance imaging (MRI) scanner by means of a gradient echo multi‐echo sequence. Afterwards, assessment of collagen orientation was carried out based on Picrosirius Red‐stained histological sections, visualized by polarized light microscopy (PLM). The average T2* relaxation time in grooved samples was lower than in contralateral control sites. Opposite to the grooved areas, the “kissing sites” had a higher average T2* relaxation time than the grooved sites. PLM showed mild changes in orientation of the collagen fibers, particularly around blunt grooves. This work shows that T2* relaxation times are different in healthy cartilage vs (early) damaged cartilage, as induced by the equine groove model. Additionally, the average T2* relaxation times are different in kissing lesions vs the grooved sites.

adult Shetland ponies that had been included in the groove study. In this model, grooves were surgically created on the proximal articular surface of the intermediate carpal bone (radiocarpal joint) and the radial facet of the third carpal bone (middle carpal joint) by either coarse disruption or sharp incision. After 9 months, T2* mapping of the entire carpal joint was carried out on a 7.0-T whole-body magnetic resonance imaging (MRI) scanner by means of a gradient echo multi-echo sequence. Afterwards, assessment of collagen orientation was carried out based on Picrosirius Red-stained histological sections, visualized by polarized light microscopy (PLM). The average T2* relaxation time in grooved samples was lower than in contralateral control sites. Opposite to the grooved areas, the "kissing sites" had a higher average T2* relaxation time than the grooved sites. PLM showed mild changes in orientation of the collagen fibers, particularly around blunt grooves. This work shows that T2* relaxation times are different in healthy cartilage vs (early) damaged cartilage, as induced by the equine groove model. Additionally, the average T2* relaxation times are different in kissing lesions vs the grooved sites. and alterations to the collagen fiber network. 2 Although, up until today, there is no cure for osteoarthritis, early detection of cartilage damage is essential for (future) therapy planning in order to prevent or decelerate the progressive irreversible damage to the joint. T2* mapping is sensitive to water content and to the degree of orientation of the collagen fiber network within the articular cartilage. 3 In addition, zonal variations in the depth of the articular cartilage could be visualized by employing T2* mapping. 4 Since articular an ultra-high field scanner to achieve a high spatial resolution which is needed to correctly map the thin layer of cartilage. It has been shown that T2* mapping is feasible on a 7T magnetic resonance imaging (MRI) with diagnostic imaging quality. 5 However, T2* relaxation time on 7T is significantly shorter than on 3T, which has to be taken into account for the imaging protocols used. 6 T2* mapping has been successfully applied within a number of joints in the human body, such as the hip, 5,7 knee, 8,9 and ankle. 10,11 To test whether T2* mapping could serve as a biomarker for (early) cartilage degeneration in a controlled fashion, one could employ an animal model such as a groove model, serving as a model for posttraumatic arthritis. The cartilage groove model has been used for OA research in rats, 12 dogs, 13 sheep, 14 and horses, 15 and has shown its potential to create posttraumatic arthritic changes in the joint. In this model, the cartilage layer is grooved through the superficial, middle, and deep zone (leaving the calcified layer intact) and the joint is subjected to intensified loading.
The horse in particular is an interesting model because of the similarities between equine and human cartilage. 16 Recently, the groove model has been applied in the equine carpal joint in two variants: grooves were surgically created by either blunt disruptions or sharp incisions (blunt grooves and sharp grooves, respectively). The contralateral joint was sham-operated and used as a control. 17 It is believed that blunt grooves and sharp grooves create different types of damage and, consequently, different progression of the disease.
In this study, we investigated the difference in cartilage integrity between grooved and control sites and between the two groove types. For that purpose, ultra-high field MRI was employed by means of a T2* mapping sequence to gain insight into cartilage quality. T2* is sensitive to magnetic susceptibility differences, and shortens in proximity to transitions between tissues like water to bone. We hypothesized that T2* relaxation times (a) are shorter in grooved cartilage than in control cartilage, (b) are different in the "kissing sites" (ie, the contact surface of the grooved cartilage) compared with grooved cartilage or control cartilage, and (c) can distinguish between blunt and sharp grooves.

| Subjects
Four adult female Shetland ponies with a mean age of 7.3 years (SD, 3.9 years; range, 4-13 years) and with a mean body weight of 203 kg (SD, 21.8 kg; range, 171-220 kg) were included in this study ( Table 1).

The study was authorized by the Utrecht University Animal Experiments
Committee and the Central Committee for Animal Experiments (AVD108002015307).

| Surgical procedure
The grooves were induced in a randomly chosen front limb through an arthrotomy at two locations: the radial facet of the third carpal bone (in the middle carpal joint), and the dorsoproximal surface of the intermediate carpal bone (in the radiocarpal joint). Blunt and sharp grooves were randomly assigned to either one of the joints (Table 1).
Blunt grooves were induced with a hooked arthroscopic probe with a sharpened tip; sharp grooves were made with a surgical blade such that incisions were of equal depth and could not exceed 400 μm. The contralateral joints were sham-operated and used as controls. The ponies were subjected to an exercise program for 8 weeks, starting 3-weeks post-surgery. 17 The ponies were euthanized in week 39, after which the carpal joints were harvested and stored at −20°C.

| Polarized light microscopy
Osteochondral samples of the whole-joint surface, including the grooved sites, the kissing sites, and the contralateral controls, were

| RESULTS
From the 32 sites available (four ponies, two legs/pony, four VOI/leg), a total of 30 sites were included for VOI analysis. Excellent highresolution T2* imaging could be performed showing T2* contrast as shown in Figure 2A, with T2* maps of sharp-grooved and bluntgrooved cartilage ( Figure 2C,D). The left middle carpal joint of pony 2 contained a large air bubble ( Figure 2B) caused by the surgery that was performed upon euthanasia (out of the scope of this paper).
Therefore, these two sites in this joint were excluded from the analysis.
The median T2* relaxation time of the sample group of grooved sites was lower (6.06 ms) compared to their contralateral (7.43 ms) counterparts in the blunt-grooved sample group and similarly in the sharp-grooved group (7.41 and 9.58 ms, respectively), as shown in Figure 3. Additionally, the median T2* relaxation time of the grooved sites was lower (6.11 ms) as compared to their opposite kissing sites (7.17 ms). Similar patterns were observed in the sharpgrooved sample group (7.77 and 8.34 ms, respectively) ( Figure 4).
No substantial differences have been observed in T2* relaxation time between groove types. Median T2* relaxation times have been summarized in Table 2.
When Picrosirius Red-stained sections were analyzed under polarized light, typical collagen fiber structures could be recognized in control sites, that is, parallel-oriented fibers in the superficial layer when compared with healthy controls. 18 These results confirm that T2* mapping can be used as a measurement for damaged cartilage.
A limitation of this work is that due to the randomization, grooved sides and groove types were not evenly distributed (ie, three right and one left front limb(s) grooved; three sharp-grooved radiocarpal joints, one blunt-grooved radiocarpal joint, and vice versa for the middle carpal joint). This relates to the differences in T2* relaxation time between blunt grooves and sharp grooves which are present, but the differences between the joint types (radiocarpal vs middle carpal) have to be taken into account. The underlying limitation of this is the small sample size within this study-we included four ponies, with eight legs, divided into four groups (being contralateral control blunt, blunt-grooved, contralateral control sharp, and sharp-grooved).
One of the limitations of T2* mapping, in general, is that T2* mapping is prone to magic angle artifacts. By placing the samples consequently in the same way in the scanner (dorsal side upwards which caused the cartilage surfaces to be orthogonally oriented towards the magnetic field) we tried to minimize the influence of magic-angle effects. Cartilage T2 and, therefore, also, T2* was shown to approximately follow the angular dependence of the nuclear dipole-dipole interaction, which has its maximum at 55°. 20 This effect is even more pronounced in highly ordered collagenous structures such as the meniscus. 21 Within this work we specifically chose to implement T2* mapping instead of T2 mapping because it can be implemented with a higher spatial resolution and can be acquired faster, providing more specific information on local tissue disruption Medians based on grooved and kissing surfaces ( Figure 3). b Medians based on grooved and control sides (Figure 4). that causes steep transitions between tissues of different magnetic susceptibilities. 22 Nevertheless, other MR methods, less sensitive to magic-angle artifacts, could have been implemented. T1rho imaging has also been shown to be prone to magic angle effects, depending on which sequence is used. 23,24 Diffusion tensor imaging (DTI) would also have been an excellent alternative to gain insight in the directionality of the collagen fibril network. The potential of DTI imaging in articular cartilage has been shown ex vivo, where a high spatial resolution of 45 µm was shown in the rat knee. 25 In vivo work on 7T MRI has shown that DTI can possibly show differences between layers in articular cartilage. 26 Another limitation of the T2* mapping is the number of echo times, which could be increased for a clinical T2* mapping protocol The acquisition time of the T2* mapping protocol used was long due to the high spatial resolution, which has to be taken into account for clinical implementation. The scan time was long because the head coil was used with a corresponding SAR model, which is too conservative for scanning of extremities. With our knee coil, we could increase the SAR (IEC limit for extremities 40 W/kg local SAR instead of 10 W/kg for the head) and, therefore, decrease the acquisition time. Additionally, the SENSE factor can be increased by using receive arrays close to the volume of interest, even decreasing the acquisition time further.
In conclusion, this work shows that T2* relaxation times are different in healthy cartilage vs (early) damaged cartilage as induced by an equine groove model. Additionally, the median T2* relaxation times are different in kissing lesions vs the grooved sites.