Multiparametric MRI characterization of knee articular cartilage and subchondral bone shape in collegiate basketball players

Abstract Magnetic resonance imaging (MRI) is commonly used to evaluate the morphology of the knee in athletes with high‐knee impact; however, complex repeated loading of the joint can lead to biochemical and structural degeneration that occurs before visible morphological changes. In this study, we utilized multiparametric quantitative MRI to compare morphology and composition of articular cartilage and subchondral bone shape between young athletes with high‐knee impact (basketball players; n = 40) and non‐knee impact (swimmers; n = 25). We implemented voxel‐based relaxometry to register all cases to a single reference space and performed a localized compositional analysis of T 1ρ‐ and T 2‐relaxation times on a voxel‐by‐voxel basis. Additionally, statistical shape modeling was employed to extract differences in subchondral bone shape between the two groups. Evaluation of cartilage composition demonstrated a significant prolongation of relaxation times in the medial femoral and tibial compartments and in the posterolateral femur of basketball players in comparison to relaxation times in the same cartilage compartments of swimmers. The compositional analysis also showed depth‐dependent differences with prolongation of the superficial layer in basketball players. For subchondral bone shape, three total modes were found to be significantly different between groups and related to the relative sizes of the tibial plateaus, intercondylar eminences, and the curvature and concavity of the patellar lateral facet. In summary, this study identified several characteristics associated with a high‐knee impact which may expand our understanding of local degenerative patterns in this population.


| INTRODUCTION
The knee is vulnerable to articular cartilage degeneration and injury in jumping athletes who exert high compressive and shear forces during practice and competitive play. [1][2][3][4][5] Imparting large loads to the articular cartilage is a known risk factor for chronic musculoskeletal conditions such as early-onset osteoarthritis (OA) 1 and pain. 2 Accordingly, there is wide interest in studying associations between high-knee impact sports and long-term health of the knee joint.
Articular cartilage is of distinct concern due to its specialized function for distributing loads and its limited capacity for repair.
Previous studies have used magnetic resonance imaging (MRI) to find that degenerative changes are consistently prevalent in knee cartilage of basketball players across all levels of competition. 3-5 A 2005 study observed articular cartilage lesions on MRI in 47.5% of asymptomatic professional NBA players, with the majority of cartilage lesions found in the patellofemoral joint. 3

A recent study by
Pappas et al., 4 imaged 24 NCAA Division I collegiate basketball players and found increased abnormal findings (fat pad edema, patellar tendinopathy, articular cartilage, and meniscal injury) after one season of play in every knee imaged. 5 Though the high prevalence of abnormal imaging findings in high-knee impact athletes is well-established, biochemical changes of macromolecules associated with cartilage degeneration occur before visible morphological changes. 6,7 Biomechanical stiffness of articular cartilage is provided by the collagen and proteoglycan (PG) organization and content, respectively, of the extracellular matrix (ECM).
Damage to this macromolecular environment results in an increase of mobile water and a concomitant reduction in tissue stiffness.
Compositional MRI techniques, such as T 1ρand T 2 -relaxation time mapping, can quantify such changes in cartilage matrix biochemistry. 8,9 T 1ρ -relaxation times reflect interactions between movementrestricted water and surrounding large macromolecules and has been related to glycosaminoglycan (GAG) and PG content and early OA. Some studies demonstrated elevated T 1ρ -relaxation times with disruption of the ECM through decreased PG content via ex vivo enzymatic removal, 10,11 yet others have seen no relation between T 1ρ abnormalities and GAG. 12,13 While the mechanism is not yet fully understood, prolonged T 1ρ has associated with populations at risk of and living with OA. [14][15][16] Meanwhile, T 2 relaxation is associated with loss of collagen and disorganization of collagen fibrils. 16 T 2 is prolonged in the setting of the degeneration of articular cartilage. 11 Newer methods permit acquisitions of T 1ρ and T 2 in a single combined sequence and have been used to evaluate patients with anterior cruciate ligament injuries and those with OA, 16,17 but its use to investigate the status of knee cartilage health in young elite athletes is limited.
Quantitative analysis of T 1ρand T 2 -relaxation time maps are traditionally performed using region of interest (ROI)-based approaches, which presents several challenges: (1) cartilage ROIs are often segmented manually or semi-automatically and are prone to inter-and intrauser variation; (2) statistical analyses are performed based on the average T 1ρ or T 2 value of all voxels within the ROIs, limiting the spatial assessment of relaxation times within the defined regions. Methods for segmentation have recently advanced to be less reliant on manual input. Advanced segmentation methods transform images from individual knees to a single reference template, allowing comparison of local spatial distribution between subjects on a voxelby-voxel basis. This technique, voxel-based relaxometry (VBR), has been shown to agree with ROI-based analyses. 18 Notably, it can be performed in a fully automated fashion and can provide local information and patterns of imaging markers in articular cartilage evaluation. 18 Another component that plays a key role in the transmission of load across the knee joint is the geometric bone shape. Through skeletal homeostatic signal pathways, 19 high-intensity mechanical loading is associated with increased subchondral bone thickness and reduced bone resorption. [19][20][21] Stimulation of these pathways occurs in an anatomic site-specific manner depending on intensity and type of load. In turn, exercise-induced variations in bone architecture influence biomechanics of the knee joint 22,23 and incidence rates of injury 24 and OA. 25 Due to frequent heavy loads exerted onto the knees of athletes in high-knee impact sports, it is important to classify regional bone shapes in sports with low-and high-knee impact.
Statistical shape modeling (SSM) has recently gained traction as an analytical method for modeling variation in surface geometry from imaging. 26,27 Varying algorithms have demonstrated submillimeter level matching precision, allowing for analysis of complex three-dimensional (3D) shapes generated from medical imaging. [26][27][28][29] The purpose of this study was to use quantitative MRI techniques to characterize the articular cartilage and subchondral bone within the knee of two athletic groups: (1) a high-knee impact group consisting of collegiate basketball players, and (2) a non-knee impact group of collegiate swimmers. We hypothesized that the basketball players would demonstrate localized prolonged T 1ρand T 2 -relaxation times and bone shape differences as compared to the swimmers.   Table 1.
To assess biases in quantitative measurements across the sites of acquisition, an identical phantom was imaged on all scanners. The phantom was constructed with two instances of three varying amounts of agarose to encompass a range of relaxation times and scanned with the T 1ρ /T 2 MAPSS sequence. 17 The phantom acquisition was repeated two additional times at a single site to evaluate intrascanner variability. Coefficients of intrascanner variation ranged from 0.2% to 2.2%, while coefficients of interscanner variation ranged from 4.1% to 6.6%. 30

| Morphological characterization
A board-certified musculoskeletal radiologist with 25 years of experience evaluated the MR images. Cartilage lesions were graded in a blinded fashion using the modified Noyes score, where Grade 0 classified cartilage with no lesions by PD-weighted MRI, and Grades 1 and above indicated increased signal intensity or cartilage defects.
For compositional analysis, all cases with cartilage lesions (modified Noyes ≥ 1) in any compartment, identified by morphological characterization, were not considered to focus on prestructural abnormalities and early signs of biochemical change. Sagittal MAPSS images in all echoes were rigidly registered to the first TSL/TE = 0 of each case using the VTK CISG registration toolkit. 31 Next, nonrigid registration to an atlas was then applied to all cases to morph the images to a common reference space. This was performed using Elastix, 32 a medical imaging registration toolbox based on maximizing mutual information between the fixed and moving images. The resulting nonrigid transformations between the atlas and each TSL/ TE = 0 cases were then applied to all other echoes/spin-lock images.
As all images were morphed to the same coordinate space, T 1ρ and T 2 maps were calculated on a voxel-by-voxel basis using Levenberg-Marquardt mono-exponential fitting.

| ROI-based relaxometry
Using a semiautomatic method based on edge-detection, 33

| Statistical shape modeling
Segmentation of femur, tibia, and PAT bones were performed automatically using V-Net, 34

| Statistical analysis
Morphological statistical analysis used a χ² test to assess the relationship of the prevalence of cartilage abnormalities between the two groups.

| Morphological evaluation
The prevalence of cartilage abnormalities was significantly higher in the basketball group (χ 2 = 6.658, p < .01), occurring in 24.6% of knees of basketball players and 6.3% of knees of swimmers (Table 2). By compartment, this increase was significant in the LFC ( χ 2 = 5.51, p < .05). Group analysis demonstrated similar results to those in Figure 2

| VBR analysis
Interpretation of the mean T 1ρ and T 2 maps from VBR-displayed prolongation near the trochlear groove and areas of shortening in the anterior and posterior regions of the tibiofemoral articulation.
Comparison of the two groups demonstrated significant differences by sport, with basketball players generally with longer T 1ρ and

| Bone shape analysis
The femur, tibia, and PAT were each described in domains defined by strain on the medial side of the tibial plateau with increased normalized walking speed, but not for the lateral side. 37 Additionally, medial compartment OA is the most common form of OA. 38,39 Our ROI-based analysis captured a pattern representative of this asymmetry; however, this method was not effective in finding local findings in other compartments. The traditional ROI-based analysis detected no differences between groups for patellar cartilage, despite previous research indicating the prevalence of imaging findings in this compartment. 2,5 Overall, the VBR analysis was more sensitive to local differences.
The prevalence of significant T 1ρ and T 2 prolongation in the medial External loading is known to influence subchondral bone shape and thickness via bone remodeling. 21,22 While our SSM results demonstrated no differences in femur shape, the significant modes of the tibia are especially relevant in controlling the biomechanics of the tibiofemoral joint. We found more symmetry between lateral and medial plateaus in basketball players as compared to swimmers.
Functionally, the lateral plateau is convex in shape and performs translational motion to the concave medial plateau. The anterolateral plateau, specifically, experiences tibial subluxation during knee flexion, indicating tibiofemoral internal rotation. 41 A high degree of rotation due to pivoting/cutting in basketball may contribute to the symmetry seen in the lateral plateau shape. Similarly, tibia mode 7 shows higher prominence of the medial spine in basketball players.
With its physical connection to the anterior cruciate ligament and its proximity to the medial meniscus, a vital tissue in shock absorption, the asymmetric heights could be explained by increased mechanical loading and subsequent bone remodeling. The size of the tibial plateau 42 and heights of the intercondylar eminence have been positively correlated with the prevalence of tibial osteophytes 43 and OA. 44 Therefore, tibia shape may be an important consideration in identifying the progression of knee kinematics, degeneration, and risk of injury in young athletes.
In regard to PAT shape, there was significant variation in the lateral facet between groups. The representative PAT of basketball players was more symmetric with a concave lateral facet, whereas that of swimmers was elongated and convex. Using Wiberg shape classification, 45  In summary, we identified several characteristics associated with high-knee impact athletes, including prolonged T 1ρ and T 2 relaxation in cartilage compartments and local depth-dependent differences, as well as bone shape variations in the tibia and PAT.