Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 3.
Published in final edited form as: J Biomech. 2025 Jul 12;190:112864. doi: 10.1016/j.jbiomech.2025.112864

Quantitative MRI-measured composition changes despite small mechanical measures in tibiofemoral cartilage of healthy adults under applied load

Marit E Scott a,b, Bruce D Beynnon a,c, Andrew S Borah c,d, John C Ramsdell a,e, Mickey I Krug c, Mack G Gardner-Morse c, Michael J DeSarno f, Jiming Zhang g,h, Matthew Geeslin h, Mathew J Failla c,i, Timothy W Tourville c,i, Niccolo M Fiorentino a,c,j,*
PMCID: PMC13229481  NIHMSID: NIHMS2098844  PMID: 40683074

Abstract

A crucial step in understanding the onset and progression of cartilaginous disease, such as osteoarthritis, is to study how cartilage mechanics and composition relate in response to controlled loading in disease-free joints. Both knees of 10 healthy participants were imaged with a 3 T magnetic resonance imaging (MRI) scanner at two timepoints (7 ± 3 days apart). Quantitative MR images for T1ρ and T2* were acquired with the knee in two states: i) a traditional setup without load applied, and ii) while a loading device applied a 40% bodyweight load to the plantar aspect of the foot. Associations between mechanical metrics (cartilage deformation, cartilage strain, change in bone-bone distance, and change in cartilage contact area) and compositional metrics (T1ρ and T2* relaxation times) were identified. Significant decreases in bone-bone distance were seen in all compartments in response to load. Articular cartilage thickness consistently decreased, but differences were significant for only half of the medial and lateral compartments in the tibia and femur. Strains ranged from 4.9% in compression to 0.3% in tension. No significant changes were found in cartilage contact area. T1ρ and T2* relaxation times changed significantly with the application of load, with the femoral and tibial cartilage exhibiting opposite responses. No significant associations were observed between mechanical and compositional metrics for T1ρ scans, but T2* scans had three significant relationships. Results from this work demonstrate that loading can induce tibiofemoral articular cartilage composition changes, as assessed with T1ρ and T2*, even with small magnitude measurements of mechanics.

Keywords: Cartilage, MR-compatible, qMRI, Load, Knee

1. Introduction

Healthy articular cartilage is essential for proper synovial joint function and overall musculoskeletal performance. Articular cartilage absorbs load and facilitates joint translations and rotations with minimal friction. Because cartilage is avascular and has a slow rate of turnover, maintaining adequate nutrient supply and metabolism depends on an optimal biomechanical and biochemical environment (Andriacchi et al., 2009; Fujii et al., 2022). In turn, a disruption in cartilage homeostasis resulting from an injury, extended period of immobilization, or change in gravitational environment during spaceflight can trigger a degenerative pathway (Hardy, 2024; Liphardt et al., 2020). A disruption in cartilage homeostasis can lead to the onset of osteoarthritis (OA), which is defined by cartilage degeneration, loss of cartilage thickness, bone remodeling, joint pain, and disability. Many previous studies have investigated the mechanosensitive nature and the complex biochemical composition of cartilage separately without directly relating them. Therefore, while the relationship between tibiofemoral cartilage mechanics and composition remains largely unknown, it is an important part of understanding how cartilage responds to load.

Detecting biomechanical and biochemical changes in cartilage in vivo presents a significant challenge. Quantitative magnetic resonance imaging (qMRI) provides a contemporary approach to quantifying morphology in cartilage and the compositional changes that may precede them (Matzat et al., 2013). Specialized qMRI parameters T1ρ and T2* provide valuable insight into the matrix composition of articular cartilage which consists of water, proteoglycans, and a dense collagen network (Cohen et al., 1998; Kuettner, 1992). During MR scanning, free water protons relax faster than the water protons that are bound to proteoglycans and the collagen matrix – increased T1ρ relaxation times are associated with loss of proteoglycans (PGs), and increased T2* relaxation times are associated with a more disorganized collagen matrix, both of which are signs of early cartilage degeneration (Argentieri et al., 2018; Matzat et al., 2013).

Traditionally, MR imaging of the knee is performed while the participant is lying in a relaxed supine position without externally applied loads. However, a lack of applied load limits the ability to investigate the properties that cartilage exhibits while under physiologic loading conditions (e.g. standing). MRI-compatible loading devices apply a participant-specific compressive load to the tibiofemoral joint for the duration of the scanning session (Argentieri et al., 2024; Jerban et al., 2020). Comparing qMR images taken in both the unloaded and loaded state permits the calculation of mechanical and compositional change metrics that are highly useful in studying the mechanisms of cartilage degeneration. Because most imaging centers do not possess an MRI-compatible loading device, we have a limited understanding of how cartilage mechanics and matrix composition respond to mechanical loading.

The purpose of this study was to quantify tibiofemoral articular cartilage mechanical change metrics (cartilage deformation, cartilage strain, change in bone-bone distance, and change in contact area) and determine their relationships with the cartilage matrix composition measured through qMRI relaxation times of T1ρ and T2* in response to applied 40% bodyweight load. We hypothesize that the 40% bodyweight applied load will induce quantifiable, nonzero mechanical metrics, and that there will be greater magnitude changes in mechanics for longer T1ρ and T2* relaxation times.

2. Methods

2.1. Participants

Ten young healthy participants (6 females, 4 males, age: 23 ± 2.4 years, body mass index: 25.3 ± 3.5 kg/m2) with normal, asymptomatic knees, no history of seeking clinical care for a knee injury or disease, and no history of lower limb injuries or disease were included in the study after receiving their written informed consent. The study was approved by and performed under the rules and regulations of the IRB at our university.

2.2. MR image acquisition and joint loading

MR image acquisition parameters and joint loading timing have been described in detail previously (Ramsdell et al., 2024). Briefly, participants’ left and right knees were imaged at two separate visits (7 ± 3 days apart) using a 3 T Phillips MR scanner (Achieva, Philips Healthcare, Best, The Netherlands) with a 16-channel radiofrequency coil. Both visits consisted of four sets of MR image acquisitions per knee, including two MR sequences (T1ρ and T2*) and two loading conditions (unloaded and loaded) (Fig. 1). In the unloaded condition, the participant was oriented in a traditional supine position. In the loaded condition, the participant was positioned within a custom-developed pneumatic-driven MRI-compatible loading device that applied a 40% bodyweight load to the plantar aspect of the subjects’ foot and subsequently to their knee (Fig. 2). The left knee was imaged first, after 15 min of off-weighting to provide time for the cartilage volume to be restored (Van Ginckel and Witvrouw, 2013). The right knee was imaged next, having been off-weighted for the entire duration of the four left knee acquisitions. T1ρ images were reconstructed to a voxel size of 0.49 × 0.49 × 1.5 mm3, and T2* images were reconstructed to a voxel size of 0.44 × 0.44 × 2 mm3.

Fig. 1.

Fig. 1.

Open in a new tab

Flowchart describing the MRI scanning sessions for each participant.

Fig. 2.

Fig. 2.

Open in a new tab

Schematic representation of custom MRI-compatible loading device used to apply compressive loads during scanning sessions. The function of the pulley was to hold the participant’s body in a steady position to prevent the loading device from pushing them up the back support.

2.3. Image segmentation

Subchondral bones and articular cartilage of the tibiofemoral joint were manually segmented in the sagittal plane using OsiriX (Rosset et al., 2004) and split into medial and lateral compartments. Segmentations consisted of points describing the outer boundary of each anatomical structure. Segmentations were performed on images from the rise of the tibial spines to the outer edge of the compartment. Segmentations were reviewed by multiple study team members, and any potential disagreements were discussed among the individuals until a consensus was reached. A similar segmentation approach was utilized in a previous study of cartilage thickness in anterior cruciate ligament injury injured patients using T1-weighted fast-field echo scans, where the intra-class coefficient of tibial cartilage thickness ranged from 0.85 to 0.97 in different regions (Argentieri et al., 2014) and was 0.94 for femoral cartilage thickness (Pius et al., 2022). Segmentations were imported into MATLAB (The MathWorks Inc., Natick, MA) for further processing. The following steps were performed separately for T1ρ and T2* datasets.

2.4. Post-processing

2.4.1. Registration

Loaded and unloaded segmentations were registered spatially so that loaded and unloaded cartilage could be compared using the unloaded bone as a reference (Fig. 3). Femur segmentations were registered to a standard reference cylinder point cloud (25 mm radius, 70 mm height). All other segmentations associated with that limb (femoral cartilage, tibia, and tibial cartilage) followed the same transformation. Loaded bone segmentations, split by compartment, were registered to their corresponding unloaded bone segmentations using pcregistericp in MATLAB, for a total of four registrations per limb (Medial Femur, Lateral Femur, Medial Tibia, Lateral Tibia). Any remaining offset between the bone locations on the unloaded and loaded scans was calculated at each point on the unloaded bone. Cartilage segmentations followed the same transformation of the corresponding bone registration.

Fig. 3.

Fig. 3.

Open in a new tab

Steps of post-processing for both T1 and T2* images. Point-based segmentations are performed on the MR images. Loaded femur and tibia segmentations are registered to unloaded femur and tibia segmentations. The loaded-unloaded transformations are used on the loaded cartilage segmentations. All segmentations are upsampled to create point clouds with increased point density and evenly spaced points. Upsampled point clouds are used in ROI selection.

2.4.2. Upsampling

Because the original segmentations consisted of sparse, unevenly spaced points, they were modified to create finely sampled, consistently spaced point clouds. A surface mesh with 0.5 mm grid resolution was fit to segmentations using linear interpolation. The grid coordinates of the surface mesh were interpolated in the x, y, and z directions by a ratio of 4 so that all points were 0.125 mm apart in the x and y directions. This ratio was determined through a sensitivity analysis to balance precision and computational efficiency.

2.5. Cartilage thickness, deformation, and strain

Deformation and strain of cartilage were measured within a region of the cartilage defined by a point of contact. Regions of interest (ROIs) for medial and lateral compartments were defined by first taking a weighted centroid of the points on the loaded femoral cartilage that were closest to the points on the loaded tibial cartilage (Ramsdell et al., 2024). From there, all points on the loaded femoral cartilage within a 6 mm radius in the x and y directions of the weighted centroid were included in the ROI (Ramsdell et al., 2024). The same x and y coordinates of this initial ROI were used to define the ROIs of all other point clouds within that limb and compartment so that all ROIs were aligned in the z-direction (Fig. 4A). Unloaded cartilage thickness was measured by calculating the distance between the points in the unloaded subchondral bone ROI to their corresponding closest points on the unloaded cartilage ROI using findNearestNeighbors in MATLAB (Maier et al., 2017). Loaded cartilage thickness followed a similar approach but either added or subtracted the loaded-unloaded registration offset at each point, depending on the z-coordinate of each point on the unloaded bone compared to the z-coordinate of its corresponding closest point on the loaded bone. Deformation of the ROI was defined as the difference between loaded and unloaded thickness at each point, averaged over the ROI. Strain was defined as the deformation divided by the unloaded thickness, averaged over the ROI. (See Fig. 5 for example ROI strain result in the four medial/lateral femoral/tibial compartments.).

Fig. 4.

Fig. 4.

Open in a new tab

Regions of interest (ROIs). A) Cartilage thickness/deformation/strain ROIs. B) Bone-bone distance ROIs. C) Contact area ROIs. D) Composition ROIs.

Fig. 5.

Fig. 5.

Open in a new tab

Example of deformations at each point within the ROIs for one limb of one subject. Negative values (blue) indicate “compression” while positive values (red) indicate “expansion.”.

2.6. Bone-bone distance

Bone-bone distance ROIs were comprised of points within a 6 mm radius of the weighted centroid of the closest points between the femur and tibia (Fig. 4B). The average distance between points in the femur and tibia ROIs was calculated for both compartments and loading conditions. Change in bone-bone distance was calculated by subtracting the unloaded values from the loaded values.

2.7. Cartilage contact area

Contact area ROIs were defined by a surface mesh generated using pc2surfacemesh in MATLAB (Fig. 4C). The surface mesh was formed by points of the femoral cartilage and tibial cartilage point clouds that were within 1 mm (roughly 2 pixels distance) from each other. The contact area of the ROI was calculated for both compartments and loading conditions. Change in contact area was calculated by subtracting the unloaded values from the loaded values.

2.8. T1ρ and T2* relaxation times

Cartilage composition ROIs and calculation of T1ρ and T2* were described in more detail by a prior publication from our group (Ramsdell et al., 2024). The cartilage composition ROIs were defined by creating a ‘contact cylinder’ with a 6 mm radius whose centroid was the tibiofemoral cartilage-on-cartilage contact (Fig. 4D). All cartilage voxels existing within the cylinder were included in the ROI for that compartment. Spin lock/echo time images were registered successively to the spin lock/echo time images that the segmentation was performed on, and a mono-exponential decay function was fit to the signals for each pixel. The functions were: S(TSL) = S0*e-TSL/T1ρ for T1ρ, with S = signal, S0 = signal at TSL = 0, and fit variables S0 and T1ρ; and S(TE) = S0*e-TE/T2* for T2*, with S = signal, S0 = signal at TSL = 0, and fit variables S0 and T1ρ. Fits were performed in MATLAB using the lsqcurvefit function. T1ρ and T2* relaxation times were averaged over all pixels within the 6 mm ROI for the unloaded and loaded MR acquisitions. Three scans were excluded from the analysis due to movement artifacts affecting image volume registration. Change in composition was calculated by subtracting the unloaded relaxation time values from the loaded values.

2.9. Statistical analysis

Descriptive statistics for thickness, bone-bone distance, contact area, and T1ρ/T2* relaxation times are presented in terms of mean +/− standard deviation. For each of these measures, comparisons of unloaded vs loaded means, across both visits and both limbs, were made using linear mixed effects models with random effect of subject and fixed effects (and repeated factors) of visit, limb, and load status (loaded vs unloaded). These analyses were done separately for each compartment. Linear mixed effects regression analyses were performed to test for significant associations between mechanical metrics and compositional metrics from T1ρ and T2* scans across both visits and both limbs. These were models with random effect of subject and fixed effects (and repeated factors) of visit and limb. Therefore, these models were testing for significant associations between these measures while controlling for effects of visit and limb. These analyses were also performed separately for each compartment. Statistical significance level alpha was set a priori to 0.05 for all analyses. All statistical analyses were done using SAS version 9.4 statistical analysis software (SAS Institute, Inc., Cary, NC, USA).

3. Results

3.1. T1ρ results

When combined across both visits and both limbs, mean cartilage thickness change ranged from −0.09 mm to −0.02 mm (negative indicates compression) (Table 1). Differences in cartilage thickness were statistically significant in the medial femur and lateral tibia compartments (p = 0.04 and p = 0.02, respectively). Cartilage was thickest in the lateral tibia for both unloaded and loaded conditions (2.91 mm ± 0.65 mm and 2.82 ± 0.68 mm, respectively). The greatest deformation occurred in the lateral tibia (−0.09 mm ± 0.17 mm). The greatest strain occurred in the medial (−3.1% ± 8.0%) and lateral tibia (−3.1% ± 6.7%). Bone-bone distance decreased significantly in both the medial and lateral sides with the application of load (p = 0.03 and p < 0.0001, respectively). There were no significant changes in contact area in any compartment. Mean T1ρ relaxation times were significantly different between loading conditions for all compartments (p ranged from 0.004 to 0.01). However, an opposite effect was observed in the femur and tibia. During the loaded condition, the medial femur and lateral femur had longer T1ρ relaxation times while the medial tibia and lateral tibia had shorter T1ρ relaxation times.

Table 1.

Mechanical and compositional metrics derived from T1ρ scans.

Medial Femur Medial Tibia Lateral Femur Lateral Tibia
Thickness (mm) Unloaded 1.87 ± 0.44 1.91 ± 0.34 1.81 ± 0.37 2.91 ± 0.65
Loaded 1.81 ±0.46 1.90 ±0.29 1.75 ±0.34 2.82 ± 0.68
p-Value 0.04 0.47 0.11 0.02
Deformation (mm) Deformation −0.06 ± 0.16 −0.02 ± 0.12 −0.06 ± 0.16 −0.09 ± 0.17
Strain (%) Strain −3.1% ± 8.0% 0.3% ± 7.6% −2.3% ± 9.0% −3.1% ± 6.7%
Bone-Bone Distance (mm) Unloaded 3.97 ± 0.62 3.97 ± 0.62 4.51 ± 0.81 4.51 ± 0.81
Loaded 3.85 ± 0.57 3.85 ± 0.57 4.28 ± 0.78 4.28 ± 0.78
p-Value 0.03 0.03 < 0.0001 < 0.0001
Difference −0.12 ± 0.20 −0.12 ± 0.20 −0.23 ± 0.20 −0.23 ± 0.20
Contact Area (mm2) Unloaded 223.73 ± 55.88 223.73 ± 55.88 167.35 ± 57.41 167.35 ± 57.41
Loaded 234.04 ± 60.89 234.04 ± 60.89 171.33 ± 62.46 171.33 ± 62.46
p-Value 0.24 0.24 0.26 0.26
Difference 10.31 ± 39.73 10.31 ± 39.73 3.99 ± 22.93 3.99 ± 22.93
T1ρ (ms) Unloaded 32.29 ± 7.05 43.88 ± 5.61 31.34 ± 6.34 37.13 ± 4.18
Loaded 37.88 ± 6.88 40.48 ± 6.52 36.60 ± 5.79 34.69 ± 3.95
p-Value 0.01 0.002 0.003 0.004
Difference 5.59 ± 7.20 −3.41 ± 4.60 5.26 ± 7.14 −2.44 ± 2.98
Open in a new tab

Values are mean ± standard deviation. P-values reflect loaded-unloaded differences, bolded p-values are statistically significant (p < 0.05).

3.2. T2* results

When combined across both visits and both limbs, mean cartilage thickness change ranged from −0.12 mm to −0.01 mm (negative indicates compression) (Table 2). Cartilage thickness changes were only significant in the medial and lateral tibia (p = 0.01, p = 0.03, respectively). Cartilage was thickest in the lateral tibia for both unloaded and loaded conditions (2.70 mm ± 0.66 mm and 2.58 mm ± 0.73 mm, respectively. The greatest deformation occurred in the lateral tibia (−0.12 mm ± 0.26 mm). The greatest strain also occurred in the lateral tibia (−4.9% ± 11.1%). Bone-bone distance decreased significantly in both the medial and lateral sides with the application of load (p = 0.002 and p < 0.0001, respectively). There were no significant changes in contact area in any compartment. Mean T2* relaxation times were significantly longer in the loaded condition for both the medial and lateral femur (p < 0.0001 and p = 0.001, respectively). Mean T2* relaxation times were significantly shorter in the medial tibia (p = 0.01) in the loaded condition, but the magnitude of change was relatively small in the medial tibia (−1.12 ms ± 2.51 ms) compared to the medial (8.20 ms ± 5.95 ms) and lateral (9.63 ms ± 8.53 ms) femur.

Table 2.

Mechanical and compositional metrics derived from T2* scans.

Medial Femur Medial Tibia Lateral Femur Lateral Tibia
Thickness (mm) Unloaded 1.66 ± 0.49 2.00 ± 0.52 1.49 ± 0.30 2.70 ± 0.66
Loaded 1.63 ± 0.52 1.91 ± 0.52 1.48 ± 0.31 2.58 ± 0.73
p-Value 0.43 0.01 0.69 0.03
Deformation (mm) Deformation −0.03 ± 0.17 −0.10 ± 0.19 −0.01 ± 0.14 −0.12 ± 0.26
Strain (%) Strain −1.3% ± 11.6% −4.1% ± 10.5% 0.1% ± 10.8% −4.9% ± 11.1%
Bone-Bone Distance (mm) Unloaded 3.57 ± 0.73 3.57 ± 0.73 3.97 ± 0.73 3.97 ± 0.73
Loaded 3.48 ± 0.70 3.48 ± 0.70 3.69 ± 0.73 3.69 ± 0.73
p-Value 0.002 0.002 < 0.0001 < 0.0001
Difference −0.09 ± 0.17 −0.09 ± 0.17 −0.27 ± 0.22 −0.27 ± 0.22
Contact Area (mm2) Unloaded 230.53 ± 75.06 230.53 ± 75.06 130.31 ± 50.21 130.31 ± 50.21
Loaded 241.56 ± 71.79 241.56 ± 71.79 131.54 ± 51.67 131.54 ± 51.67
p-Value 0.20 0.20 0.78 0.78
Difference 11.03 ± 47.67 11.03 ± 47.67 1.23 ± 21.56 1.23 ± 21.56
T2* (ms) Unloaded 28.83 ± 6.88 29.59 ± 4.42 25.44 ± 3.49 23.75 ± 3.17
Loaded 37.03 ± 8.82 28.46 ± 4.08 35.07 ± 8.36 24.49 ± 3.59
p-Value < 0.0001 0.01 0.001 0.07
Difference 8.20 ± 5.95 −1.12 ± 2.51 9.63 ± 8.53 0.74 ± 2.68
Open in a new tab

Values are mean ± standard deviation. P-values reflect loaded-unloaded differences, bolded p-values are statistically significant (p < 0.05).

3.3. Associations between mechanical metrics and compositional metrics

There were no significant associations between mechanical metrics and compositional metrics for T1ρ scans, and three significant associations from T2* scans (Table 3). Two of the significant associations were with change in bone-to-bone distance, which had significant associations with change in composition in the medial tibia (p = 0.04) and unloaded composition in the lateral femur (p = 0.04). Change in contact area was also associated with change in composition in the lateral tibia (p = 0.01).

Table 3.

Associations between mechanical change metrics and compositional metrics.

Mechanical Measure p-Values
MF MT LF LT
T1ρ Change in Composition Deformation 0.40 0.85 0.90 0.26
Strain 0.37 0.96 0.99 0.18
Change in Bone-Bone Distance 0.25 0.34 0.43 0.38
Change in Contact Area 0.91 0.84 0.83 0.57
Unloaded Composition Deformation 0.55 0.99 0.70 0.09
Strain 0.69 0.97 0.54 0.09
Change in Bone-Bone Distance 0.64 0.86 0.77 0.72
Change in Contact Area 0.97 0.62 0.47 0.66
T2* Change in Composition Deformation 0.44 0.11 0.37 0.86
Strain 0.37 0.14 0.51 0.81
Change in Bone-Bone Distance 0.60 0.04 0.13 0.17
Change in Contact Area 0.49 0.63 0.63 0.01
Unloaded Composition Deformation 0.98 0.26 0.60 0.72
Strain 0.88 0.29 0.97 0.93
Change in Bone-Bone Distance 0.74 0.64 0.04 0.60
Change in Contact Area 0.83 0.27 0.94 0.16
Open in a new tab

Change metrics are Loaded – Unloaded. Bolded p-values are statistically significant (p < 0.05). MF: medial femur. MT: medial tibia. LF: lateral femur. MF: medial femur.

4. Discussion

The purpose of this study was to quantify changes in tibiofemoral cartilage mechanics and determine how these changes are related to cartilage matrix composition via qMRI relaxation times in response to an applied compressive load. Overall, our hypotheses were only supported in part, with the applied load inducing quantifiable, nonzero significant changes in most mechanical metrics (cartilage deformation and bone-to-bone distance but not contact area), but mechanical metrics were only related to T2* relaxation times in three instances with no significant relationships with T1ρ relaxation times.

The compressive strains observed in this study are mostly on par with those reported in previous studies under similar conditions. Both Cofatana et al. and Subburaj et al. also reported compressive strains (also referred to as negative percent changes in cartilage thickness or negative percent deformations) of up to 7% when a 50% bodyweight load was applied to healthy participants’ knee joints via MRI-compatible loading device (Cotofana et al., 2011; Subburaj et al., 2012). Wang et al reported slightly higher compressive strains of up to 15.5% in healthy participants with a similar study design but only included four participants (Wang et al., 2015). Previous studies have also reported cartilage strain that was calculated by comparing cartilage thickness directly before and after performing an activity rather than using a loading device, such as single-leg hopping (maximum strains of 8% in the tibial cartilage and 6% in the femoral cartilage) (Sutter et al., 2019) and walking (maximum strains of 8% in the tibial cartilage and 7% in the femoral cartilage) (Lad et al., 2016). Overall, the strains measured from the present study were similar but slightly lower than those reported in the literature. This could be because we used a lower load magnitude of 40% body weight. However, it is also important to consider the differences in how the MR images were acquired. All the aforementioned studies used either 3D spoiled gradient-echo (3D-SPGR) or double-echo steady state (DESS) MR sequences with pixel sizes ranging from 0.3 mm to 0.41 mm. In the present study, we used T1ρ and T2* sequences which have the advantage of being sensitive to cartilage matrix biochemistry but have a larger pixel size (0.49 mm for T1ρ and 0.44 m for T2*). Pixel size can affect image quality and thus impact findings due to the difficulty in differentiating cartilage layers, which may be important as cartilage deformation involves sub-millimeter changes in cartilage thickness on average that are smaller than the pixel size.

The bone-bone distance changes in each compartment were of similar magnitude in both T1ρ and T2* scans. This suggests that the bone-bone distance was the most repeatable and quantifiable mechanical metric. While the difference in contact area was positive when comparing loaded scans to unloaded scans, the lack of significant differences in contact area changes found in the present study contrasts with what has been reported previously. Both Suburraj et al. and Shin et al. found that contact area increased significantly in both medial and lateral compartments in healthy participants, and even greater increases were seen in OA participants (Shin et al., 2011; Subburaj et al., 2012). This discrepancy may be attributed to the method used to define contact regions. In the present study, sagittal plane images, including the tibial spine, were excluded due to worsening image quality with the rapidly changing bone and cartilage geometry as the tibial spine rose from the tibial plateau, even though this region will also be in contact under load. In contrast, Subburaj et al. defined contact regions directly from the coronal plane of 3D-SPGR images, which likely included tibial spine cartilage (Subburaj et al., 2012). A systematic reanalysis of the images in the present study using the approach of Subburaj et al. would provide a direct comparison between the two studies and highlight how methodological differences contribute to conflicting results (Subburaj et al., 2012).

We expected to see greater magnitude changes in mechanics for longer T1ρ and T2* relaxation times and bigger changes in T1ρ and T2* relaxation times with the application of load. However, only three out of the 64 mechanical-compositional associations were significant, and these were all in T2* scans. Subburaj et al. is the only other study reporting the associations between cartilage deformation/contact area and qMRI parameters (Subburaj et al., 2012). They stratified T1ρ and T2 values into high and low categories based on their group medians. The high-T1ρ group had greater changes in contact area and greater deformations than the low-T1ρ group, but only the contact area stratification reached significance. There were no differences in either contact area or deformation between the high-T2 and low-T2 groups. It is important to note that both normal (n = 20) and OA (n = 10) participants were included in this stratification, so the high-T1ρ and high-T2 likely included most of the OA participants who had higher T1ρ and T2 values because of disease.

The most notable finding regarding cartilage matrix composition was that the femoral and tibial cartilage demonstrated opposite effects on T1ρ and T2* relaxation times between unloaded and loaded conditions. This suggests that the composition within the ROI changes when the 40% bodyweight load is applied. We hypothesize that the load influences the net fluid flow between the femoral and tibial cartilage due to differences in permeabilities in the femoral and tibial cartilage (Mäkelä et al., 2015). Specifically, Makela et al. found higher permeability for tibial cartilage compared to femoral cartilage on the medial and lateral sides (Mäkelä et al., 2015). Higher permeability for tibial cartilage may explain the higher T1ρ values in tibial cartilage with loading due to water entering tibial cartilage and effectively lowering the relative concentration of proteoglycans, given that T1ρ is negatively correlated with concentration of proteoglycans (Duvvuri et al., 1997). In addition, we speculate that fluid flow throughout the same cartilage layer may explain regions of expansion and compression.

This study has potential limitations that must be considered when interpreting the results. The previously mentioned limitations in MR resolution created difficulty in image segmentation, which could have contributed to the discrepancy seen between the consistently significant decreases in bone-bone distance and the occasionally significant cartilage deformation. Specifically, the in-plane pixel size of the T1ρ (0.49 mm) and T2* (0.44 mm) may have contributed to the variability in cartilage deformation, especially when considering the pixel size was greater than the measured deformations (range −0.09 mm to −0.02 mm and −0.12 mm to −0.01 mm, respectively). Another potential limitation is that the orientation of each participant’s knee within the loading device is dependent on their anatomy and could influence the way that intersegmental loads were transmitted across the tibiofemoral joint. A full analysis of tibiofemoral translations and rotations would be required to further investigate the alignment of the lower limb within the loading device. Furthermore, the orientation of the tibia and femur were restricted to approximate standing based on the geometry of the radiofrequency coil, whereas a more flexed position may have better simulated different activities like walking. In addition, a larger load may have better approximated activities like walking or jogging. Finally, this study involved a small number of participants which may influence reaching significance for mechanical changes in response to applied loading.

The advantages of the present study are that we looked at the relationship between mechanical metrics and qMRI parameters over the entire range of T1ρ and T2* values, as opposed to stratifying into high and low groups, and we are the first to include T2* sequences in these relationships. The main takeaway from this study is that significant changes in the articular cartilage composition of healthy knees can be seen even with small changes in mechanics. Studying how healthy articular cartilage responds to load is the first step in understanding the mechanical and biochemical changes that may be associated with the onset or progression of cartilage degeneration. This work could also be used to inform the development of a loading device for individuals experiencing joint unloading due to microgravity exposure, immobilization after surgery, rest following an injury, or limitations from disease.

Acknowledgements

We acknowledge Ashley Williams, Jiang Du, and Jose Raya for their advice during the execution of this project. We acknowledge Yansong Zhao (Clinical Scientist from Philips Healthcare) for proving the 3D MAPPS MRI sequence. Lastly, we acknowledge Ben Hamilton, Colby Fischer, Will Robinson, Victoria Walters, and Jeremy Frank for assisting with MRI segmentations.

Marit Scott’s contributions to this research were partially supported by NASA grant 80NSSC20M0122. Research reported in this publication was also supported by the National Institutes of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under award numbers R21AR077371 and R01AR076765. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

CRediT authorship contribution statement

Marit E. Scott: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Data curation, Conceptualization. Bruce D. Beynnon: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Conceptualization. Andrew S. Borah: Writing – review & editing, Validation, Methodology, Investigation, Data curation. John C. Ramsdell: Writing – review & editing, Software, Methodology, Formal analysis, Data curation. Mickey I. Krug: Writing – review & editing, Validation, Methodology, Data curation. Mack G. Gardner-Morse: Writing – review & editing, Validation, Software, Methodology. Michael J. DeSarno: Writing – review & editing, Formal analysis. Jiming Zhang: Writing – review & editing, Methodology, Investigation. Matthew Geeslin: Writing – review & editing, Formal analysis. Mathew J. Failla: Writing – review & editing, Resources, Methodology. Timothy W. Tourville: Writing – review & editing, Resources, Methodology. Niccolo M. Fiorentino: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

  1. Andriacchi TP, Koo S, Scanlan SF, 2009. Gait mechanics influence healthy cartilage morphology and osteoarthritis of the knee (Suppl 1). J. Bone Joint Surg. Am 91, 95–101. 10.2106/JBJS.H.01408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Argentieri EC, Burge AJ, Potter HG, 2018. Magnetic resonance imaging of articular cartilage within the knee. J. Knee Surg 31, 155–165. 10.1055/s-0037-1620233. [DOI] [PubMed] [Google Scholar]
  3. Argentieri EC, Pekmezian A, Wach A, Zhu A, Bansal S, Breighner RE, Leatherman ER, Potter HG, Maher SA, Koff MF, 2024. Baseline-to-loaded changes in regional tibial cartilage thickness, T1ρ and T2: utilization of an MRI compatible loading device. J. Orthop. Res 42 (12), 2646–2658. 10.1002/jor.25956. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Argentieri EC, Sturnick DR, DeSarno MJ, Gardner-Morse MG, Slauterbeck JR, Johnson RJ, Beynnon BD, 2014. Changes to the articular cartilage thickness profile of the tibia following anterior cruciate ligament injury. Osteoarthr. Cartil 22 (10), 1453–1460. 10.1016/j.joca.2014.06.025. [DOI] [PubMed] [Google Scholar]
  5. Cohen NP, Foster RJ, Mow VC, 1998. Composition and dynamics of articular cartilage: structure, function, and maintaining healthy state. J. Orthop. Sports Phys. Ther 28 (4), 203–215. 10.2519/jospt.1998.28.4.203. [DOI] [PubMed] [Google Scholar]
  6. Cotofana S, Eckstein F, Wirth W, Souza RB, Li X, Wyman B, Hellio-Le Graverand M-P, Link T, Majumdar S, 2011. In vivo measures of cartilage deformation: patterns in healthy and osteoarthritic female knees using 3T MR imaging. Eur. Radiol 21 (6), 1127–1135. 10.1007/s00330-011-2057-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Duvvuri U, Reddy R, Patel SD, Kaufman JH, Kneeland JB, Leigh JS, 1997. T1rho-relaxation in articular cartilage: effects of enzymatic degradation. Magn. Reson. Med 38 (6), 863–867. 10.1002/mrm.1910380602. [DOI] [PubMed] [Google Scholar]
  8. Fujii Y, Liu L, Yagasaki L, Inotsume M, Chiba T, Asahara H, 2022. Cartilage homeostasis and osteoarthritis. Int. J. Mol. Sci 23 (11), 6316. 10.3390/ijms23116316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Hardy JG, 2024. Articular cartilage loss is an unmitigated risk of human spaceflight. npj Microgravity 10 (1), 104. 10.1038/s41526-024-00445-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Jerban S, Chang EY, Du J, 2020. Magnetic resonance imaging (MRI) studies of knee joint under mechanical loading: review. Magn. Reson. Imaging 65, 27–36. 10.1016/j.mri.2019.09.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Kuettner KE, 1992. Biochemistry of articular cartilage in health and disease. Clin. Biochem 25 (3), 155–163. 10.1016/0009-9120(92)90224-g. [DOI] [PubMed] [Google Scholar]
  12. Lad NK, Liu B, Ganapathy PK, Utturkar GM, Sutter EG, Moorman CT, Garrett WE, Spritzer CE, DeFrate LE, 2016. Effect of normal gait on in vivo tibiofemoral cartilage strains. J. Biomech 49 (13), 2870–2876. 10.1016/j.jbiomech.2016.06.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Liphardt A-M, Mündermann A, Heer M, Achtzehn S, Niehoff A, Mester J, 2020. Locomotion replacement exercise cannot counteract cartilage biomarker response to 5 days of immobilization in healthy adults. J. Orthop. Res 38 (11), 2373–2382. 10.1002/jor.24753. [DOI] [PubMed] [Google Scholar]
  14. Maier J, Black M, Bonaretti S, Bier B, Eskofier B, Choi J-H, Levenston M, Gold G, Fahrig R, Maier A, 2017. Comparison of different approaches for measuring tibial cartilage thickness. J. Integr. Bioinform 14 (2), 20170015. 10.1515/jib-2017-0015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Mäkelä JTA, Han S-K., Herzog W, Korhonen RK., 2015. Very early osteoarthritis ¨ changes sensitively fluid flow properties of articular cartilage. J. Biomech 48 (12), 3369–3376. 10.1016/j.jbiomech.2015.06.010. [DOI] [PubMed] [Google Scholar]
  16. Matzat SJ, van Tiel J, Gold GE, Oei EHG, 2013. Quantitative MRI techniques of cartilage composition. Quant. Imaging Med. Surg 3 (3), 162–174. 10.3978/j.issn.2223-4292.2013.06.04. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Pius AK, Beynnon BD, Fiorentino N, Gardner-Morse M, Vacek PM, DeSarno M, Failla M, Slauterbeck JR, Sturnick DR, Argentieri EC, Tourville TW, 2022. Articular cartilage thickness changes differ between males and females 4 years following anterior cruciate ligament reconstruction. J. Orthop. Res 40 (1), 65–73. 10.1002/jor.25142. [DOI] [PubMed] [Google Scholar]
  18. Ramsdell JC, Beynnon BD, Borah AS, Gardner-Morse MG, Zhang J, Krug MI, Tourville TW, Geeslin M, Failla MJ, DeSarno M, Fiorentino NM, 2024. Tibial and femoral articular cartilage exhibit opposite outcomes for T1ρ and T2* relaxation times in response to acute compressive loading in healthy knees. J. Biomech 169, 112133. 10.1016/j.jbiomech.2024.112133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Rosset A, Spadola L, Ratib O, 2004. OsiriX: an open-source software for navigating in multidimensional DICOM images. J. Digit. Imaging 17 (3), 205–216. 10.1007/s10278-004-1014-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Shin CS, Souza RB, Kumar D, Link TM, Wyman BT, Majumdar S, 2011. In vivo tibiofemoral cartilage-to-cartilage contact area of females with medial osteoarthritis under acute loading using MRI. J. Magn. Resonance Imaging: JMRI 34 (6), 1405–1413. 10.1002/jmri.22796. [DOI] [PubMed] [Google Scholar]
  21. Subburaj K, Souza RB, Stehling C, Wyman BT, Le Graverand-Gastineau M-P, Link TM, Li X, Majumdar S, 2012. Association of MR relaxation and cartilage deformation in knee osteoarthritis. J. Orthop. Res 30 (6), 919–926. 10.1002/jor.22031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Sutter EG, Liu B, Utturkar GM, Widmyer MR, Spritzer CE, Cutcliffe HC, Englander ZA, Goode AP, Garrett WE, DeFrate LE, 2019. Effects of anterior cruciate ligament deficiency on tibiofemoral cartilage thickness and strains in response to hopping. Am. J. Sports Med 47 (1), 96–103. 10.1177/0363546518802225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Van Ginckel A, Witvrouw E, 2013. Acute cartilage loading responses after an in vivo squatting exercise in people with doubtful to mild knee osteoarthritis: a case-control study. Phys. Ther 93 (8), 1049–1060. 10.2522/ptj.20120491. [DOI] [PubMed] [Google Scholar]
  24. Wang H, Koff MF, Potter HG, Warren RF, Rodeo SA, Maher SA, 2015. An MRI-compatible loading device to assess knee joint cartilage deformation: effect of preloading and inter-test repeatability. J. Biomech 48 (12), 2934–2940. 10.1016/j.jbiomech.2015.08.006. [DOI] [PubMed] [Google Scholar]

RESOURCES