Understanding Magnetic Resonance Imaging of Knee Cartilage Repair: A Focus on Clinical Relevance

Morphological MRI Techniques

MRI systems with high static magnetic fields can increase signal to noise ratio (SNR) and produce greater spatial resolution images than those with lower magnet strength, an important feature for cartilage imaging. Therefore, 1.5-Tesla (T) and 3-T magnet systems with extremity coils are recommended. In particular, the higher SNR allows isotropic 3D images to be acquired in a clinically relevant acquisition time. Advances in coil technology and parallel imaging techniques have also contributed to reduced scan times. The same acquisition techniques are used for repair tissue and native cartilage. High spatial resolution is essential to obtain images of diagnostic quality. In-plane resolution of 0.3 mm has been recommended to resolve fibrillation of the native cartilage surface using fat suppressed 3D gradient-recalled echo imaging. ⁵ Similarly, fissures and incomplete cartilage repair tissue integration require high-resolution imaging. In this section, we describe the pulse sequences most commonly used for morphological assessment of cartilage and cartilage repair tissue.

Two-Dimensional (2D) Fast Spin Echo (FSE)

Two-dimensional FSE or turbo spin echo imaging sequences provide excellent SNR, contrast between tissues, and faster acquisition times. ⁶ 2D-FSE is the core imaging technique for the musculoskeletal system, in general, and the acquisition type most commonly used for clinical assessment of cartilage lesions; it is part of the cartilage imaging protocol recommended by International Cartilage Repair Society. ⁷ The most common MRI acquisitions for morphological assessment of cartilage include proton density-weighted (PDW), intermediate-weighted (IW), and T2-weighted (T2W) sequences with or without fat suppression. ⁸ On the other hand, T1W images are not considered to be cartilage-sensitive because they only allow poor differentiation between joint effusion and cartilage unless used in conjunction with MR arthrography. By comparison, fluid-sensitive images provide good evaluation of cartilage, ligaments, menisci, and edema-like marrow signal. T2W images provide greater contrast between cartilage surface and effusion and can detect subtle changes such as fibrillation, although the signal intensity (SI) in the cartilage near the osteochondral junction becomes low due to decreased signal intensity in the cartilage with longer TE times, which makes differentiation of the deeper layers of cartilage within the radial zone and at the tidemark challenging. PDW images (TE < 30 ms) provide good cartilage surface contrast and have shown high sensitivity for internal cartilage signal changes. IW imaging with longer echo times (TEs) (30-60 ms) combines the advantages of T2W and PDW imaging and is less prone to magic angle artifacts. ⁹ Care should be taken to adjust the TE on each magnet system to optimize visualization of the cartilage-fluid and cartilage-bone interfaces.

Three-Dimensional (3D) FSE

Two-dimensional FSE imaging often suffers from the need for thicker slices with resultant partial volume averaging effects that may be greater than in 3D acquisitions. ⁸ To optimally view the cartilage and joint space, 2D-FSE also requires imaging acquisition in multiple planes, which is time consuming. Isotropic, or near isotropic, 3D sequences can produce higher spatial resolution and high-quality reformatted images in any plane. All image planes are potentially provided from a single sequence obviating the multiple acquisitions required to image all planes with standard 2D sequences. 3D-FSE acquisitions use an additional phase-encoding gradient for spatial encoding along the slice-select direction and generate a set of contiguous slices. ¹⁰ The clinically available 3D-FSE sequences rely on flip-angle modulation to reduce blurring and parallel imaging to reduce scan time. ⁸

Three-dimensional FSE sequences have a diagnostic accuracy similar or slightly inferior to 2D-FSE for evaluation of focal cartilage defects and other joint structures. It has been shown that 3D-FSE can be applied for multitissue assessment of the whole joint.^{11 -13} Additionally, IW 3D-FSE sequences with fat saturation allow for evaluation of subchondral bone, although image contrast is better with water sensitive 2D-FSE images.

Spoiled Gradient Recoiled Echo (SPGR)

Fat suppression or water excitation 3D-SPGR, or similar fast low-angle shot (FLASH), imaging produces a contrast in which the SI of cartilage is greater than the surrounding tissues and joint fluid. ⁸ However, unreliable image contrast between the cartilage and fluid or joint tissues may result in poor detection of subtle cartilage defects. ¹⁴ An additional limitation is susceptibility artifacts, which can be prominent at some cartilage repair sites. Another problem with gradient recalled echo based (GRE) sequences, in general, is insensitivity to bone marrow edema and other joint pathology.^15,16 While GRE sequences have been used for focal cartilage defect assessment,^17,18 several recent studies have shown that they are less sensitive for focal defect detection than other techniques.^16,19,20 These sequences are, however, excellent for cartilage segmentation and quantification of cartilage volume and thickness due to the good image contrast between cartilage and subchondral bone.^21,22

Dual-Echo Steady-State (DESS)

DESS is another GRE imaging technique for cartilage that was introduced in the 1990s, ²³ but did not become widely used until 3D-DESS could be performed with a short imaging time. Flip angles may be adjusted to optimize cartilage-fluid contrast to improve cartilage lesion detection. ²⁴ Fat saturation or water excitation is typically used with DESS. The diagnostic performance of 3D-DESS for cartilage lesions is commensurate with 3D-GRE sequences.

Balanced Steady-State Free Precession (bSSFP)

Three-dimensional bSSFP sequences provide excellent contrast between cartilage and adjacent synovial fluid with bright fluid SI compared to cartilage. ²⁵ Magnetic field inhomogeneities may result in off-resonance artifacts worsened by long TRs. Short TRs can improve the images but result in longer acquisition times. The diagnostic accuracy of bSSFP for morphologic assessment of cartilage is similar to that of standard 2D- and 3D-GRE sequences.^25,26 Fat suppression, water excitation, or a Dixon technique such as IDEAL are commonly used with bSSPF. ²⁷ The use of combined IDEAL and bSSFP may result in higher SNR and cartilage-fluid contrast to noise ratio. ²⁸

MRI Observation of Cartilage Repair Tissue (MOCART)

MOCART is a reproducible semiquantitative scoring system for assessment of morphological cartilage repair. MOCART originally defined 9 structural variables but was modified to assess 11 variables, taking advantage of higher resolution 2D and isotropic 3D MR images, as summarized in Table 1 . ²⁹ This “3D-MOCART” improves evaluation by localizing the features within the repair site and the relation of the repair site to the weight-bearing regions of the joint, evaluating the repair tissue-cartilage interfaces or “border zones,” in every plane, and providing a detailed subchondral bone assessment. ²⁹ Since all of the original MOCART features are assessed by 3D-MOCART, for the sake of simplicity we will refer to both types as “MOCART.” ²⁹ The usefulness of MOCART in a randomized controlled clinical trial of autologous cartilage tissue implants has been demonstrated in a study by Anderson and colleagues through its 5-year follow-up period. ³⁰

OPEN IN VIEWER

Table 1.

Summary of MOCART Scoring System for Cartilage Repair Tissue.

Features Evaluated	Scores
Defect fill (degree of defect repair and filling of the defect in relation to the adjacent cartilage)	0%; 0% to 25%; 25% to 50%; 50% to 75%; 75% to 100%; 100%; 100% to 125%; 125% to 150%; 150% to 200%; >200%
Cartilage interface (integration with adjacent cartilage to border zone in two planes)	Scored using sagittal (femur, patella, trochlea, tibia), coronal (femur, tibia), and axial (patella, trochlea) planes. Complete; demarcating border visible (split-like); defect visible <50%; defect visible >50%
Bone interface (integration of the transplant to the subchondral bone; integration of a possible periosteal flap)	Complete; partial delamination; complete delamination; delamination
Surface (constitution of the surface of the repair tissue)	Surface intact; surface damaged <50% of depth; surface damaged >50% of depth; adhesions
Structure (constitution of the repair tissue)	Homogeneous; inhomogeneous or cleft formation
Signal intensity (intensity of MR signal in the repair tissue in comparison to the adjacent cartilage: normal = identical to adjacent cartilage; nearly normal = slight areas of signal alterations; abnormal = large areas of signal alteration)	Normal; nearly normal; abnormal
Subchondral lamina (constitution of the subchondral lamina)	Intact; not intact
Chondral osteophytes (osteophytes within the cartilage repair area)	Absent; osteophytes <50% of repair tissue; osteophytes >50% of repair tissue
Bone marrow edema (maximum size and localization in relation to the cartilage repair tissue and other alterations assessed in the 3D MOCART score)	Absent; small (<1 cm); medium (<2 cm); large (<4 cm); diffuse
Subchondral bone (constitution of the subchondral bone)	Intact; granulation tissue; cyst
Effusion (approximately size of joint effusion visualized in all planes)	Absent; small; medium; large

When the repair tissue is as thick as the adjacent cartilage, defect fill is scored as 100%. Hypertrophy (>100%) and incomplete repair (<100%) are recorded in 25% increments. Kreuz et al. found that graft hypertrophy over 150% after autologous chondrocyte implantation (ACI) of full-thickness chondral defects in the knee usually required surgery while less severe hypertrophy could be watched. ³¹

The cartilage-interface variable scores the integration of the repair to adjacent cartilage at the border zone. Absence of a gap between the repair tissue and native cartilage is considered “complete.” Incomplete integration is either a fissure-like border zone or a measurable defect and is scored as less than or more than 50% of the length of the defect in coronal, axial, and sagittal directions. Of note, the conspicuity of fissures at the area of integration with the host cartilage is a function of both the in-plane and through-plane resolutions.

The bone interface grades integration of the repair tissue to the subchondral bone. It is described as complete integration, partial delamination, or complete delamination. For periosteal ACI, partial or complete delamination of the periosteal flap is noted.

The repair tissue surface is considered “intact” when it is even and smooth, without evidence of damage or irregularity. When damage such as fibrillation or ulceration is present, it is scored as involving less than or more than 50% of the tissue thickness. Although rare, synovial adhesions attached to the repair tissue are recorded under the “surface” variable.

The repair tissue structure is classified as “homogenous” when a normal, layered cartilage appearance is seen over the entire repair site. If the structure is lost, it is classified as inhomogeneous. Additionally, if a cleft is present in the structure, it is mentioned specifically.

Based on the visual inspection of MR images, the SI of the repair tissue is considered normal if it is identical to the adjacent native cartilage. If the SI is hyper- or hypointense, it should be classified as “nearly normal” or “abnormal,” depending on the extent and/or degree of altered SI. This SI variable is based on any cartilage-dedicated MR sequence; to simplify scoring, no specific acquisition type is specified.

The subchondral lamina, or bone plate, between the repair tissue and the bone marrow is classified as either intact or “not intact” if irregular.

Chondral osteophytes, that is, ossifications within the repair site above the level of the adjacent subchondral bone plate, are scored as absent or less than or more than 50% of the repair tissue thickness. Of note, chondral osteophytes are not uncommonly seen following the microfracture.

Bone marrow edema subjacent to the repair site is classified according to diameter as small (<1 cm), medium (<2 cm), large (<4 cm), or diffuse. The size of bone marrow edema is calculated in cm². If the edema is related to other pathology by MOCART scoring, this is noted.

MOCART also scores the subchondral bone marrow adjacent (beneath) to the repair site as intact if there is no abnormality, or note is made of granulation tissue, subchondral sclerosis, and cysts. Joint effusions are estimated by visual inspection and graded as absent, small, moderate, or large.

Cartilage Repair Osteoarthritis Knee Score (CROAKS)

MOCART is comprehensive for assessment of the repair site itself; however, assessment of the other structures of the joint is paramount to assess longitudinal outcomes and development of osteoarthritis. The CROAKS combines features of MOCART and the MRI osteoarthritis knee score, ³² which is an established semiquantitative scoring system for whole organ assessment of the knee, to provide a comprehensive, reproducible tool for longitudinal postoperative assessment after surgical cartilage. ³³

Table 2 shows the specific features outside of the repair site, assessed for multiple joint subregions by CROAKS. These include bone marrow lesions beyond the repair site, subchondral cysts, nonoperated cartilage status, osteophytes, synovitis, effusion, menisci, and anterior and posterior cruciate ligaments. Other periarticular features evaluated include the pes anserine bursa, illiotibial band, prepatellar bursa, infrapatellar bursa, and presence of popliteal cysts, ganglion cysts, and/or loose bodies.

OPEN IN VIEWER

Table 2.

Summary of MRI Compositional Techniques.

Compositional MRI Techniques	Cartilage Component Assessed			Advantages	Disadvantages
	Collagen Network	Water Content	GAG
T2 mapping	Yes	Yes	No	No need for intravenous (IV) gadolinium; well validated; easy to implement	Long scan time using multiecho spin-echo sequence; cannot assess calcified cartilage at osteochondral junction
T2* mapping	Yes	Yes	No	No need for IV gadolinium; shorter scan time than T2 mapping; can be used with UTE to assess calcified cartilage at the osteochondral junction	Not well validated; susceptible to magnetic field inhomogeneities and magic angle effects
T1 rho imaging	No	Yes	Yes	No need for IV gadolinium; sensitive to early degeneration; may complement T2/T2* mapping	Nonspecific in terms of cartilage components assessed; limited availability; long scan time
Sodium imaging	No	No	Yes	No need for IV gadolinium; correlates directly with GAG content	Needs specialized hardware and coils; long scan time; low spatial resolution
dGEMRIC	No	No	Yes	Direct assessment of GAG content by inverse relation of GAG accumulation/contrast uptake; well validated	Needs IV gadolinium; 90 minutes delay between exercise and scanning
gagCEST	No	No	Yes	No need for IV gadolinium	Difficult to implement on clinical systems due to technical complexity; needs ultra-high field MRI; not well validated
Diffusion-weighted imaging	Yes	No	Yes	No need for IV gadolinium; provides additional information regarding cartilage microarchitecture	Semiquantitative image processing is demanding; susceptible to movement artifacts
Ultrashort TE imaging (UTE)	Yes	Yes	No	Can be used to assess tissue with intrinsic short T2 such as cartilage near osteochondral junction; can be used in conjunction with T2, T2* and T1 rho imaging	Limited availability

References for Table II.^70-72

Compositional MRI

At the molecular level, cartilage consists of approximately 70% to 80% fluid and 20% to 30% solid extracellular matrix (ECM). ³⁴ The ECM is composed of a network of collagen fibrils and proteoglycan molecules; mostly aggrecans that consist of a protein core with covalently attached glycosaminoglycans (GAGs).^34,35 GAGs are negatively charged and make up the majority of the “fixed charge density” of the ECM. Electroneutrality is maintained by the partitioning of cations, mainly sodium (Na⁺), and anions, mainly chloride (Cl⁻), between the cartilage and joint fluid. The biomechanical properties of cartilage are largely due to water flow within the ECM and the cartilage-joint fluid distribution of water that is, in part, dictated by the fixed charge density.

Compositional MRI acquisitions provide a way to detect biochemical and microstructural changes in the cartilage ECM even before gross morphological changes occur. Although not in routine clinical use, these techniques have been used extensively in cartilage research. Compositional MRI can supplement morphologic imaging, by potentially defining the biomechanical quality of cartilage repair tissue. The basic concepts of the compositional MRI techniques currently available to evaluate cartilage repair tissue are summarized in Table 2 . Of note, the stratification of T2 relaxation times across the articular cartilage reflects changes in the orientation of collagen fibers in the extracellular matrix. ³⁶ The disorganized tissue found in cartilage repair sites generally has greater water mobility and therefore more prolonged T2 relaxation times. Laminar analysis of normal articular hyaline cartilage T2 maps, that is, separate measurement of regions of interest of the superficial and deep halves of the cartilage ( Fig. 1 ), has shown that T2 values are consistently higher at the articular level than at the bone interface. Restoration of a bilaminar structure with longer superficial T2 relaxation times in a cartilage repair site is indicative of maturation to a hyaline-like repair tissue.^37,38 While there is more literature on compositional MRI of nonoperated, degenerating cartilage, a substantial amount of research has applied these methods for evaluation of cartilage repair tissue. ³⁹

OPEN IN VIEWER

Figure 1.

Bilaminar T2 relaxometry. Example of manual segmentation of the cartilage of the medial tibia and weight-bearing femoral condyle on a sagittal multi-echo spin echo (MESE) sequence. The tibial cartilage is segmented from its anterior to posterior end, and the femoral cartilage throughout a weight-bearing region of interest. As cartilage T2 is known to display spatial variation with tissue depth, the segmented cartilage plates subdivided into the top (superficial) and bottom (deep) 50%, based on the local distance between the segmented cartilage surface and bone interface. Color coding displays intrachondral variations in regional T2.