Interpretation of Cartilage Damage at Routine Clinical MRI: How to Match Arthroscopic Findings

Abstract

This review is intended to aid in the interpretation of damage to the articular cartilage at routine clinical MRI to improve clinical management. Relevant facets of the histologic and biochemical characteristics and clinical management of cartilage are discussed, as is MRI physics. Characterization of damage to the articular cartilage with MRI demands a detailed understanding of the normal and damaged appearance of the osteochondral unit in the context of different sequence parameters. Understanding the location of the subchondral bone plate is key to determining the depth of the cartilage lesion. Defining the bone plate at MRI is challenging because of the anisotropic fibrous organization of articular cartilage, which is susceptible to the “magic angle” phenomenon and chemical shift artifacts at the interface with the fat-containing medullary cavity. These artifacts may cause overestimation of the thickness of the subchondral bone plate and, therefore, overestimation of the depth of a cartilage lesion. In areas of normal cartilage morphology, isolated hyperintense and hypointense lesions often represent degeneration of cartilage at arthroscopy. Changes in the subchondral bone marrow at MRI also increase the likelihood that cartilage damage will be visualized at arthroscopy, even when a morphologic lesion cannot be resolved, and larger subchondral lesions are associated with higher grades at arthroscopy. The clinical significance of other secondary features of cartilage damage are also reviewed, including osteophytes, intra-articular bodies, and synovitis.

Online supplemental material is available for this article.

Work of the U.S. Government published under an exclusive license with the RSNA.

SA-CME LEARNING OBJECTIVES

After completing this journal-based SA-CME activity, participants will be able to:

• ■ Discuss which properties of the osteochondral unit have the greatest effect on its MRI appearance.

• ■ Describe the nonmorphologic features of cartilage damage.

• ■ Identify the sources of pain from cartilage damage.

Introduction

Accurate assessment of the depth and extent of damage to the articular cartilage in a joint is important for clinical decision making, patient counseling, and prognostication. An ever-growing range of surgical and nonsurgical management options are available, and the choice among them is determined, in part, by the depth, extent, and location of damage to the cartilage.

MRI is the optimal noninvasive method for assessment of articular cartilage and other internal derangements. However, the appearance of cartilage is especially susceptible to changes in imaging sequence parameters and orientation in the magnetic field at MRI, which may make interpretation more challenging. A working knowledge of how cartilage structure affects its appearance is important to accurate interpretation of normal and damaged cartilage.

This review provides a framework for interpreting damage to the articular cartilage in synovial joints at routine clinical MRI with the use of histologic, biochemical, and clinical management insights. The clinical significance of secondary features of damage to the cartilage are reviewed, including subchondral bone marrow changes, osteophytes, intra-articular bodies, and synovitis. Advanced imaging sequences, quantitative techniques, and machine learning algorithms are not discussed.

Anatomy of the Osteochondral Unit

The mature osteochondral unit is composed of the hyaline articular cartilage, the subchondral bone plate, and the subchondral spongiosa (Figs 1, 2, E1).

Figure 1.

Illustration of osteochondral unit structure and associated MRI appearance. The extracellular matrix of the superficial zone is composed of a mat of interwoven collagen fibrils that resists shear forces. The middle and deep zones contain leaves of collagen fibrils that are aligned to the plane of the compressive load. Collagen fibrils are rooted in the thin calcified layer and do not cross into the subchondral bone plate or the subarticular spongiosa. Gradation of signal intensity at MRI is primarily related to the effects of anisotropy (A). Low signal intensity in the region of the cartilage-bone interface (B) at MRI is related to deep-layer anisotropy, the calcified layer, the bone plate, and chemical shift artifacts. A bare area is present at the margins of joints in the space between the cartilage and joint capsule and is covered by synovium.

Figure 2.

Appearance of the osteochondral unit at MRI, micro-CT, and histologic analysis using tibial plateau specimens from donors without known disease. (A) 3-T MR image with 30-µm in-plane resolution and the sample in an anatomic orientation shows the layered appearance of cartilage related to the structural organization of the extracellular matrix, with lower signal intensity in the highly organized deep and superficial zones. (B) Quantitative polarization microscopic image with color representing the orientation of the collagen fibers (color wheel at upper left) and the intensity representing optical retardance, shows that the deep and superficial zone collagen orientations are nearly perpendicular to one another, with a relatively isotropic middle zone. (C) Micro-CT image shows the high-attenuation thin calcified layer and bone plate and the subarticular trabecula. (D) Digitized image of a histologic slide shows the junction between the deep zone of uncalcified cartilage and the calcified layer, which is referred to as the tidemark. The junction between the calcified layer and the irregular bone plate, referred to as the cement line, can also be identified, and blood vessels are noted to perforate through the bone plate (arrow). (Safranin O and van Gieson stains; digitized using a ×40 objective lens.)

Articular Cartilage

Healthy articular cartilage contains a relatively small number of chondrocytes (1%–5% by volume), and the bulk of the tissue consists of an extracellular matrix composed of water (65%–85%), collagen (10%–20% by wet weight), and proteoglycans (5%–10% by wet weight). Articular cartilage is generally organized into superficial (tangential), middle (transitional), deep (radial), and calcified zones (Figs 1, 2). The superficial zone is composed of a thin mat of densely packed interwoven collagen fibrils aligned roughly parallel to the articular surface that resists shear forces. The middle and deep zones have an extracellular matrix composed of densely packed collagen fibrils in a columnar organization that manage the compressive load. These fibrils radiate like blades of grass from the bone in the deep zone and curve through 90° arcs in the middle zone (1). The fibrous organization of the extracellular matrix is vertically oriented centrally and becomes more oblique at the joint margin (1). The fibrous nature of articular cartilage becomes apparent at MRI with higher spatial resolution (Figs 2, 3). The relative proportions of these zones vary by location in the joint, and every joint surface is different. Compressive loads are additionally managed by internal hydrostatic pressure driven by the hydrophilic properties of the proteoglycan molecules, with expansion constrained by the surrounding collagen meshwork.

Figure 3.

Resolution limitations of clinical MRI. 3-T MR images (repetition time, 3000 msec; echo time, 60 msec) show a tibial plateau explant at 0.31 × 0.56 × 2 mm (A), which is at the upper limits of clinical spatial resolution, and at 0.06 × 0.06 × 0.8 mm (B), which is higher than clinical spatial resolution. At higher spatial resolution, the cartilage architecture is revealed, with delineation of the radially oriented collagen fibrils. Photograph of the gross specimen (C) shows scuffing and dulling of articular cartilage (dz), which can also be seen at arthroscopy. The adjacent surface is normal (nl). These types of superficial degeneration are not detectable at clinical MRI because of resolution constraints; however, at a resolution beyond routine clinical resolution (B), perturbation of the superficial zone can be seen (arrow in B).

The anisotropy of the collagen architecture is a major factor in the appearance of articular cartilage at MRI (2). The gradual change in shading of signal intensity seen at MRI of normal articular cartilage is related to changes in the angle between the anisotropic collagen fibrils and the main magnetic induction field (B₀) (Fig 4). Like tendon tissue, the water molecules between the dense collagen fibrils are restricted. The charges on the collagen protein cause the water molecules to assume a preferential alignment (unlike bulk water that is randomly oriented) and the residual dipole coupling results in phase shifts, generally promoting T2 decay (3). However, there is strong orientational variation, with the shortest T2 relaxation (lowest signal intensity) when fibers are parallel to B₀. As the fibers are tilted away from the direction of B₀, dipole-dipole interactions decrease, with maximal T2 prolongation observed at 54.7° (and 125.3°), which is the "magic angle.”

Figure 4.

Effect of magic angle phenomenon on signal intensity in the deep zone. MR images show a patella explant angled (arrow) toward (A, C) and at 55° from (B, D) the main magnetic induction field (B₀) using proton-density–weighted (repetition time, 2000 msec; echo time, 13 msec) (A, B) and intermediate-weighted (repetition time, 2000 msec; echo time, 65 msec) (C, D) sequences. At parallel orientations to B₀, the anisotropic collagen fibrils promote rapid T2 decay (low signal intensity) because of dipole to dipole interactions. As the fibrils are tilted away from the direction of B₀, interactions decrease, with maximal T2 prolongation (high signal intensity) observed at 54.7° (the magic angle). Longer echo times promote lower signal intensity in the deep zone.

This effect explains the changes in cartilage signal intensity at different depths and at the curved margins of joints. For example, on images of the knee from sagittal fluid-sensitive MRI sequences, signal intensity is commonly elevated at the femoral trochlea anteriorly and at the femoral condyles posteriorly and is decreased in both the trochlea and condyles inferiorly. Similarly, on images of the shoulder from coronal fluid-sensitive MRI sequences, cartilage signal intensity is commonly elevated at the medial aspect of the humeral head and decreased at the superior aspect of the humeral head. In fact, because of the magic angle effect, the signal intensity in the same region of healthy cartilage can vary by greater than 200%, depending on its orientation to B₀ (Fig 4) (2). These anisotropic effects are more apparent on images from long–echo-time sequences (intermediate-weighted or T2-weighted MRI) than on those from short–echo-time sequences (T1-weighted or proton density–weighted MRI).

The appearance of the deep zone is most affected by anisotropy, because the fibrils are straight and compact in the deep zone. In the middle zone, the fibrils arch in an intermixing manner, resulting in loss of anisotropy. Therefore, the middle zone normally maintains a relatively elevated signal intensity or isointensity at all orientations to B₀. The thin superficial zone is variably seen at clinical MRI. Although the fibrils of the superficial zone are oriented roughly parallel to the articular surface, each articular surface has a specific directional pattern of fibril orientation (4,5). Thus, orientation-dependent relaxation times depend both on the angle of the surface to B₀ and the rotation angle around the axis of the surface (4).

Cartilage volume decreases with age (6). The age-related nonpathologic reduction in the volume of cartilage is related to changes in hydration rather than to degenerative cartilage loss, which is a pathologic process, regardless of age. Specifically, changes in cartilage volume may be due to alterations in aggrecan, the major proteoglycan in the articular cartilage. These changes include changes in size, structure, sulfation, and accumulation patterns (7–10). Although type II collagen, the most abundant matrix protein in cartilage, has a half-life of more than 100 years (11), the half-life of aggrecan is 25 years in cartilage (12). Because aggrecan’s hydrophilic properties are responsible for maintaining the high water content in articular cartilage, these changes are thought to lead to decreased cartilage volume. Because the signal intensity of articular cartilage is most strongly affected by the matrix structure, age-related qualitative alteration of signal intensity is not expected and has not been reported.

Finally, it is important to be aware that every joint surface has bare areas where there is no covering cartilage (Fig 1). In most joints, bare areas are at the margins of the articular surface; however, several joints have central bare areas that could be confused with cartilage defects. For example, in the shoulder, there may be a bare spot in the central glenoid fossa (associated with the tubercle of Assaki); in the elbow, there may be a transverse bare band in the trochlear notch of the ulna; and in the hip, there may be a bare spot in the superior acetabulum centrally (supra-acetabular fossa) or more medially (stellate crease). Also, some articular surfaces have eye-catching marginal bare areas to be aware of. For example, in the ankle, there is a bare area at the anterior medial aspect of the tibial plafond (ie, the notch of Harty), and in the elbow, there is a large bare area at the posterior aspect of the capitellum (ie, pseudodefect).

Subchondral Bone

The subchondral bone consists of the subchondral bone plate or articular bone plate and the subarticular spongiosa. The subchondral bone mechanically and metabolically supports the articular cartilage, maintaining the joint shape and absorbing shock.

The thickness of the bone plate varies both within and between joints and is best assessed using CT (Fig 2) (13). Overlying the bone plate is a discrete band of mineralized cartilage referred to as the calcified layer, which is firmly anchored to the bone plate by an irregular interdigitating interface referred to as the cement line (Figs 2, E1). Collagen fibrils pass only into this calcified layer of the cartilage and do not pass into the bone plate (Fig 1). The interface between the calcified cartilage and the overlying noncalcified cartilage is referred to as the tidemark and is vulnerable to damage from shear forces, which may result in cartilage delamination (14,15). The bone plate is not an impermeable structure and is invaded by hollow spaces that provide a direct connection between the uncalcified cartilage and the marrow cavity of the spongiosa (Fig 2). Through these channels, blood vessels and nerves can reach the deep portions of the cartilage (16). The calcified layer is removed in abrasion chondroplasty, allowing blood to flow through these channels into a well-shouldered cartilage defect, with the expectation that introduction of biologic factors will result in fibrocartilage fill (17,18).

The appearance of the bone plate is influenced by the receiver bandwidth of the image. The receiver bandwidth refers to the range of frequencies used to encode (or read out) an image. This is a parameter that can be changed at the will of the operator. The representation of bandwidth differs slightly depending on the vendor of the system used, but GE Healthcare, Siemens Healthineers, and Canon Medical all display bandwidth (in hertz per pixel) on the operator console. At 3 T, the chemical shift difference between fat and water is approximately 440 Hz, and the fat-water shift is calculated by dividing 440 Hz by the bandwidth (in hertz per pixel), yielding the number of shifted pixels. The number of shifted pixels is directly displayed on the console of Philips systems. By multiplying the number of pixels with the resolution in the frequency-encoding direction (typically 0.3–0.5 mm), the exact shift in millimeters can be calculated. Knowledge of this phenomenon is important because chemical shift artifact may artifactually thicken or thin the appearance of the bone plate in the frequency-encoding direction (Fig 5). Chemical shift artifact is exacerbated when the bandwidth is decreased. This is typically done to improve signal-to-noise ratios, such as on images from T2-weighted MRI sequences. Fat-suppression techniques reduce or eliminate chemical shift artifact, depending on how strongly the fat signal is suppressed. Dixon and water excitation methods of fat suppression eliminate chemical shift artifact. The strength of fat saturation using chemical shift–selective (CHESS) methods can typically be adjusted. Often the spoiler gradient for fat is purposefully weak to maintain contrast between fat and tendons, ligaments, and bone, and in this scenario, chemical shift artifact is reduced but not eliminated.

Figure 5.

Effect of chemical shift artifacts on the appearance of the subchondral bone plate. (A, B) Coronal T1-weighted MR images of a knee joint were acquired with a standard clinical protocol at 3 T, including routine bandwidth (± 41.7 kHz, 236.8 Hz per pixel) images with default caudal-cranial (A) and reversed cranial-caudal (B) frequency-encoding directions (dashed arrows). Fat shifts by nearly two pixels (fat-water frequency shift of 440 Hz divided by 236.8 Hz per pixel), which is equal to 0.9 mm at the imaged spatial resolution. This is at least 2–3 times thicker than the true thickness of the subchondral bone plate (13). Spatial misregistration at the interface with the fat-containing medullary cavity causes the bone plate to appear thicker (black arrow) in the tibial plateau and is obliterated in the femoral condyle (white solid arrow) with the default frequency-encoding direction (dashed arrow in A) and with the reversed encoding direction (dashed arrow in B). (C) Coronal T2-weighted MR image shows that these artifacts are mitigated with the application of routine fat suppression and the default frequency-encoding direction (dashed arrow).

Synovium

The synovial membrane is a delicate, highly vascular inner membrane of the joint capsule and is vital to the maintenance of joint homeostasis. Synovial tissue contains fibroblast-like synovial cells that produce synovial fluid, including the lubricant hyaluronic acid, and macrophage-like synovial cells that remove debris and control the proinflammatory and anti-inflammatory cytokine balance in the synovial fluid (19,20). Chondrocytes are primarily nourished by diffusion from the synovial fluid and primarily depend on anaerobic metabolism.

MRI Interpretation of Cartilage Damage

Modern surgical and MRI grading systems define the depth of cartilage damage using a similar schema, with four major categories. The lowest category denotes cartilage softening, the second category denotes cartilage damage involving less than 50% of the cartilage thickness, the third category denotes cartilage damage involving greater than 50% of the cartilage thickness, and finally, the fourth category denotes full-thickness cartilage damage. Using these four categories in clinical practice is both efficient and free from possible miscommunication related to a numeric scoring system (Table 1). The extent of cartilage damage is also important to define for clinical decision making and patient counseling about the prognosis. Most surgical and MRI scoring systems quantify the amount of damage by providing an estimate of the surface area involved. In clinical practice, the extent of disease can be effectively classified by description, such as focal, medium-sized, large-sized, and diffuse, with measurements provided for focal full-thickness defects and delamination.

Table 1:

Surgical and MRI Grading Systems for Depth of Cartilage Damage

Regarding focal cartilage lesions, softening or partial-thickness lesions are handled similarly regardless of depth and may be ignored or treated with chondroplasty, which refers to smoothing of degenerative cartilage or trimming of unstable margins. Full-thickness lesions may benefit from marrow stimulation (eg, abrasion chondroplasty or microfracture), chondrocyte implantation (autologous or allogenic), or osteochondral grafts (autologous or allogenic). A precise measurement of cross-sectional size is perhaps most important when defining an isolated full-thickness defect, because lesion size may be the determining factor for choosing among treatment options. For example, osteochondral autografts and marrow stimulation procedures such as microfracture and abrasion chondroplasty to punctate bleeding bone are typically reserved for small defects, measuring up to approximately 2 cm² (17,18). Larger lesions are most often treated with an osteochondral allograft transplant.

Figure 6 shows the elements of the normal MRI appearance of the knee and ankle joint cartilage, with CT arthrograms for reference.

Figure 6.

Normal articular cartilage in a 15-year-old adolescent boy with knee pain being assessed for meniscal tear (A–C) and in a 28-year-old man with ankle pain (D–F). (A, D) Coronal CT arthrograms of the knee (A) and ankle (D) show the true thickness of the subchondral bone plate (bp) beneath the articular cartilage (c). (B, C, F) Intermediate-weighted (B) and intermediate-weighted fat-suppressed (C) MR images of the knee and T2-weighted fat-suppressed MR image (F) of the ankle show how the low signal intensity of the deep cartilage may exaggerate the thickness of the bone plate, depending on the echo time and the orientation of the articular surface to B₀ because of anisotropy (a) in the deep zone (dz), with increased signal intensity at the magic angle (ma). cs = chemical shift artifact, mz = middle zone, sz = superficial zone. (E) T1-weighted MR image of the ankle shows how chemical shift artifact thickens the appearance of the talar dome bone plate (cs). The higher signal intensity of the middle zone (mz) is seen between the thin low signal intensity of the superficial zone (sz) and the variable signal intensity of the deep zone (dz). Arthroscopy revealed normal cartilage in the knee and ankle (not shown).

Changes in Signal Intensity of Articular Cartilage

Signal intensity changes may be reflective of composition change or morphologic lesions that are beyond the spatial resolution of clinical MRI sequences. Changes in signal intensity are most easily identified with the use of quantitative T2-weighted mapping techniques but can also be detected with fluid-sensitive (intermediate-weighted or T2-weighted fast spin-echo) MRI sequences, which are routinely used for clinical assessment of cartilage (5,21). Investigators (21) have reported that both increases and decreases in signal intensity on T2-weighted MR images may be seen with cartilage degeneration at arthroscopy and may progress to morphologic defects at subsequent MRI (Figs 7, 8, E2) (22). Hyperintense lesions in morphologically normal cartilage have a 94% positive predictive value for detection of degeneration at knee arthroscopy (23). These areas are typically ill defined, located in abnormally expanded cartilage, and correlate with areas of softening at arthroscopy (Fig 7). An increase in T2 relaxation time is thought to be due to multiple factors including increased water content, decreased macromolecular content, and disruption of the collagen matrix ultrastructure (24–27). Hypointense lesions in morphologically normal cartilage have a 64% positive predictive value for detection of degeneration at knee arthroscopy (21). These areas involve the middle zone, may be linear or globular, and are most commonly areas of fibrillation or fissuring at arthroscopy (Fig 8). A decrease in relaxation time may be due to factors such as disruption of tissue anisotropy, magnetization transfer effects (ie, increased exposure of “bound” protons on disrupted collagen to bulk water), or mature fibrocartilage (28). Care must be taken to distinguish these areas from the normal pattern of joint-specific anisotropy, and radiologists should familiarize themselves with areas of cartilage that are usually relatively hyperintense or hypointense. Familiarity with the appearance of truncation artifacts related to the use of Fourier transforms to reconstruct MRI signal into images is also helpful (Fig 7).

Figure 7.

Signal hyperintensity of the cartilage lesion in a 27-year-old woman with left knee pain. Axial T2-weighted fat-suppressed MR image (A) and arthroscopic image (B) show an area of signal hyperintensity and mild swelling in the medial facet of the patella (arrow) that correlates to cartilage softening with probing during arthroscopy (B). Truncation artifact is noted in the lateral facet of the patella (ta).

Figure 8.

Cartilage lesion in the central femoral trochlea of a 22-year-old woman with right knee pain. (A, B) Axial T2-weighted fat-suppressed (A) and coronal intermediate-weighted fat-suppressed (B) MR images show a typical hypointense lesion (arrow). (C) Arthroscopic image shows that the lesion correlates with a partial-thickness cartilage fissure (arrow).

When present, calcium pyrophosphate deposition (ie, chondrocalcinosis) may cause characteristic scattered, punctate, hypointense signal foci, which are most easily detected with gradient-echo MRI sequences (Fig E2). Recent investigations using data from the Osteoarthritis Initiative of the National Institutes of Health (29) have shown the presence of chondrocalcinosis to be associated with a higher prevalence of cartilage and meniscal damage and greater risk of cartilage and meniscal degeneration over 4 years (30).

Morphologic Changes of Articular Cartilage

Common terminology of cartilage damage is detailed in Table 2, and lesions are illustrated in Figures 9–12. More acute injuries tend to have more linear features or more squared-off margins, and more chronic disease tends to have rounded or gradated margins.

Table 2:

Cartilage Damage Terminology

Figure 9.

Superficial partial-thickness cartilage fissuring in the patella of a 27-year-old woman with left knee pain. Axial CT arthrogram (A), axial T2-weighted fat-suppressed MR image (B), and arthroscopic image (C) show the lesion (black arrow in A, white solid arrow in B and C). Also note the alteration in signal intensity in the underlying deep zone of the patellar cartilage (open arrow in B), which further indicates regional chondral degeneration. Superficial morphologic defects can be detected more easily with higher spatial resolution, which in this case is greatest on the CT arthrogram.

Figure 12.

Full-thickness cartilage damage with subchondral bone marrow lesion in the medial talar dome of a 55-year-old woman with left ankle pain. Intermediate-weighted MR images with caudal-cranial (A) and left-right (B) frequency-encoding directions (dashed arrow in A and B) and (C) coronal T2-weighted fat-suppressed MR image with a caudal-cranial frequency-encoding direction (dashed arrow in C) show an area of essentially full-thickness damage (solid arrow). Chemical shift (cs) artifactually thickens the appearance of the talar dome bone plate (bp) when the frequency-encoding direction is caudal-cranial (A, C) but not when the frequency-encoding direction is left-right (B) or when fat suppression is applied (C). Minimal friable tissue was seen at the base of the defect at arthroscopy (not shown). The presence of subchondral bone marrow edema–like changes (e) and cystlike changes (c) increases the likelihood of finding advanced cartilage damage at arthroscopy. Intact cartilage at the medial aspect of the joint has low signal intensity in the deep zone related to anisotropy (a).

Figure 11.

Full-thickness cartilage defect in the left humeral head in a 48-year-old man with shoulder pain. (A) Coronal CT image shows intact joint space and a normal thin bone plate (bp) for reference. (B, C) Sagittal T2-weighted fat-suppressed MR image (B) and arthroscopic image (C) show a full-thickness defect (large arrow) with delamination along the margins (small arrows in B) and superficial cartilage damage along the glenoid fossa (arrowheads). Intact cartilage at the superior aspect of the head has low signal intensity related to anisotropy (a), where the collagen fibrils of the deep zone are parallel to B₀.

Sequence parameters affect grading of cartilage damage because they affect delineation of the articular surface and the underlying bone plate and the spatial resolution and signal intensity of articular cartilage. T1-weighted MRI sequences provide poor contrast between joint fluid and the cartilage surface, making superficial lesions difficult to define unless arthrograms are acquired. T2-weighted MRI sequences provide excellent distinction between the joint fluid and cartilage surface but may provide poor internal cartilage signal intensity, making it difficult to define the bone plate to determine lesion depth. Proton-density–weighted MRI sequences yield the highest signal-to-noise ratio, typically allowing the highest spatial resolution, and provide excellent contrast between cartilage and both the joint fluid and the bone plate. Intermediate-weighted sequences are similar to proton-density–weighted sequences, but with longer echo times, giving them more T2 weighting (Table 3). Intermediate-weighted MRI sequences provide higher overall signal intensity and spatial resolution while still being sensitive to T2-weighted signal intensity alterations indicative of early cartilage degeneration. Proton-density–weighted and intermediate-weighted MRI sequences are routinely used to assess the articular cartilage during routine clinical imaging. Cartilage lesions are generally better defined with thinner sections; and for this reason, isotropic imaging sequences such as three-dimensional fast spin-echo or spoiled gradient-echo T2*-weighted MRI sequences may be particularly helpful in joints with thin cartilage such as those of the hip.

Table 3:

Definition of Clinical MRI Parameters

Because of the effect of sequence parameters, care must be taken not to confuse the low-signal-intensity deep fibers and the bone plate when differentiating deep partial-thickness and full-thickness cartilage lesions (Figs 10, E3). For most sequences, cartilage damage needs to extend nearly to the subchondral bone marrow to allow confident diagnosis of a full-thickness lesion.

Figure 10.

Deep partial-thickness cartilage flap in a 53-year-old man. (A) Sagittal T2-weighted fat-suppressed MR image shows a fissure (large arrow) and chondral delamination (small arrow) in the weight-bearing portion of the medial-femoral condyle. (B) Sagittal proton-density–weighted MR image shows the depth of this lesion, with some cartilage tissue (t) remaining deep to the surface defect (arrow) and the flap. (C) Coronal intermediate-weighted fat-suppressed MR image also shows the partial-thickness damage (large and small arrows). t = cartilage tissue. Note how both the femoral condyle and tibial plateau cartilage are difficult to discern from the underlying bone on the T2-weighted fat-suppressed image, which may be a source of overestimation of cartilage loss and cartilage lesion depth. See Figure E3 for comparison of normal, deep-partial, and full-thickness cartilage lesions at this location.

The reported sensitivity of both two-dimensional and three-dimensional MRI sequences for detection of cartilage softening and superficial partial-thickness damage in the knee is less than 50% when compared with that of arthroscopy (31–33). This relatively low sensitivity is primarily attributed to suboptimal spatial resolution and is joint dependent. Moreover, an in-plane spatial resolution of 0.15 mm is required to reliably distinguish superficial morphologic changes of cartilage, which is beyond the spatial resolution of most MRI sequences used in clinical practice (ie, typically 0.3–0.5-mm in-plane spatial resolution and 3–4-mm section thickness) (Fig 3) (34). The agreement between MRI and arthroscopic results is also poor for differentiation of superficial and deep partial-thickness lesions (35); however, the accuracy of MRI is greater for deeper lesions, particularly in those that involve more than 50% loss of chondral substance, where reported sensitivity is higher than 75% (31–33). This interpretive problem may, in part, be related to diffuse cartilage loss (36). At arthroscopy, only the cartilage surface is seen, with no gauge of overall thickness without a defect as a reference point. Therefore, diffuse superficial loss at MRI may be interpreted as "dull" or only containing minor scuffs at arthroscopy. Moreover, interpretive discrepancies between superficial and deep cartilage defects may occur when there is a background of superficial cartilage loss, with the radiologist considering the absolute depth and the surgeon considering the relative depth of the defect. Fortunately, the distinction between partial-thickness and full-thickness cartilage lesions is more clinically important than the distinction between grades of partial-thickness lesions.

Care should be taken when using the term osteoarthritis when interpreting MRI examinations. The term osteoarthritis describes not only morphologic severity, but also clinical severity. Furthermore, the term osteoarthritis implies generalized disease, typically correlates well with radiographic findings, and has a different management path than that for localized disease. The clinical management of a severe focal cartilage defect is distinct from the clinical management of severe osteoarthritis, which is limited to physiotherapy and arthroplasty. Therefore, it may be misleading to equate a focal high-grade chondral abnormality with severe osteoarthritis in a joint, especially when radiographs are normal or near normal.

Finally, the appearance of cartilage may also be altered by reparative fibrocartilage, which may fill areas of previously full-thickness cartilage damage, particularly after surgical bone marrow stimulation treatment (Fig 13). Whether it happens because of microscopic cracks or is surgically induced by a microfracture instrument, penetration of the bone plate allows a cartilage defect to become populated with platelets, growth factors, and bone marrow–derived mesenchymal stem cells, which mediate a fibrocartilaginous repair process. The fibrocartilage that forms in the defect is not as organized and has inferior biomechanical properties to those of hyaline cartilage but is superior to exposed bone. Shortly after treatment, the reparative fibrocartilage appears hyperintense to native cartilage on fluid-sensitive MR images and may be difficult to differentiate from fluid or may appear very thin (18). As the repair tissue matures, its signal intensity decreases, and it may eventually appear hypointense in comparison with native cartilage. Defects that demonstrate poor fill, incomplete peripheral integration, or persistent underlying edema-like marrow signal intensity 2 years after treatment are associated with worse outcomes (18).

Figure 13.

Fibrocartilage filling in a full-thickness cartilage defect in a 39-year-old woman with knee pain that began after a fall. Sagittal T2-weighted fat-suppressed MR image acquired initially (A) shows a full-thickness cartilage defect in the lateral femoral condyle (arrow); image acquired 9 months later (B) shows partial spontaneous filling with fibrocartilage (arrow); and image acquired 8 months after subsequent chondroplasty and microfracture treatment (C) shows nearly complete fill with fibrocartilage (curved arrow).

Subchondral Bone Changes

Changes in the subchondral bone marrow at MRI increase the likelihood of identifying overlying cartilage damage at arthroscopy, even when a morphologic lesion cannot be resolved on MR images (Fig 12) (21,37). These lesions include bone marrow edema–like signal intensity and cystlike lesions (38). These terms are favored over edema and cyst because fibrovascular tissue may be a major constituent of these processes, and they show enhancement at MRI after intravenous administration of contrast material. Fat suppression on images from fluid-sensitive sequences helps to accentuate these lesions. Larger subchondral lesions are associated with higher-grade defects at arthroscopy (37), and enlarging subchondral lesions are associated with progression of overlying cartilage loss (39). Cystlike features favor a diagnosis of chronic disease. The subchondral bone is well innervated, and the presence of subchondral lesions correlates with patient pain (40).

There are two proposed causes for subchondral changes from cartilage damage (16). The synovial intrusion theory describes microstress fractures in the bone plate, allowing a physical connection between the joint cavity and the subchondral bone, through which synovial fluid travels. The bone contusion theory describes microstress fractures developing in the subchondral trabecula, leading to bone necrosis and formation of cystlike lesions.

Discussion of subchondral abnormalities is incomplete without briefly addressing nomenclature. The term osteochondral lesion is nonspecific and has been used to refer to any lesion that involves the cartilage and subchondral bone (38,41). The use of more specific terms is recommended when possible (eg, a chronic full-thickness cartilage defect with subchondral cyst). An osteochondral defect is a more specific type of osteochondral lesion and is characterized by a focal defect in the cartilage and the underlying bone. Osteochondritis dissecans is a specific term referring to a repetitive microtrauma injury of the developing subchondral bone, resulting in disruption of endochondral ossification and conferring risk of instability to the osteochondral unit (38,41). Moreover, osteochondritis dissecans may progress from subchondral edema-like signal intensity to an osteochondral defect, if it does not heal.

Osteophytes

Osteophytes in synovial joints are defined as central, marginal, or synovial. Central osteophytes develop from the bone plate in areas where cartilage has abnormal structure and composition (42). Adjacent and overlying damage to cartilage is frequently present (42–45). However, the central osteophyte itself is often covered by tissue and is rarely visible at arthroscopy (Fig 14) (43). Moreover, the typical mechanism of central osteophyte formation likely involves microscopic fracture of the articular plate, without an overlying cartilage defect (45,46). Bone formation occurs at the fracture site as part of the healing process, resulting in a central osteophyte rising beneath the articular cartilage. Occasionally, a central osteophyte forms after an initial cartilage defect, followed by a reparative response, filling out the defect with fibrocartilage and a central osteophyte (42,45). An identical process, described as subchondral bone overgrowth, occurs frequently after surgical microfracture (Fig 14) (47).

Figure 14.

Central osteophyte formation (subchondral bone overgrowth) in a 39-year-old woman with knee pain. (A) Initial sagittal T2-weighted fat-suppressed MR image shows a full-thickness cartilage defect in the lateral femoral condyle (arrow). (B, C) Arthroscopic images show the full-thickness defect (*) identified and treated and the microfracture (mf). (D) Sagittal proton-density–weighted MR image obtained 9 years later shows a central osteophyte to have filled the previous chondral defect (curved arrow). (E) Subsequent arthroscopic image shows the area to be covered by fibrocartilage, with no central osteophyte visible (arrow). A central osteophyte arising after microfracture treatment is a subchondral bone overgrowth that is typically small. Fraying (f) of the lateral meniscus free margin is also noted.

Marginal osteophytes are common features of osteoarthritis. Multiple clinical studies (48–52) have shown that marginal osteophytes detected on radiographs are the most sensitive finding for prediction of cartilage lesions. The cytomorphology and gene expression patterns in formation of marginal osteophytes resemble those of a healing fracture callus, forming from stimulation of the periosteum in the bare areas at the margins of joints (53,54). Synovial osteophytes have similar biologic characteristics to those of marginal osteophytes but occur in areas away from the articular cartilage margin, such as the buttressing osteophytes seen along the femoral neck in hips with osteoarthritis.

Formation of marginal osteophytes is driven by cytokine mediators that are released from synovial macrophages into the synovial fluid in response to breakdown products from damaged cartilage (19,20,55–61). Transforming growth factor β (TGF-β) and bone morphogenetic protein 2 (BMP-2) are key cytokine mediators in the formation of osteophytes (55,59,60,62,63). Cytokine mediation of marginal osteophyte growth explains why compartments in the knee with marginal osteophytes and intact cartilage at MRI and arthroscopy are commonly seen in knees that have other compartments with both osteophytes and cartilage lesions (64). Moreover, damaged cartilage in one compartment of the knee may cause marginal osteophytes at another compartment with intact cartilage because of the shared joint cavity. Similarly, the greatest predictor of marginal osteophyte size in a single compartment is the summation of the highest-grade cartilage lesion from each compartment, rather than the highest grade of cartilage lesion in the same compartment (64). An important point from this research is that if the cartilage in an articular surface appears to be normal at MRI, then it is more likely to be normal at arthroscopy, regardless of the presence or the size of a marginal osteophyte (Fig 15).

Figure 15.

Marginal osteophytes at compartments with intact cartilage in a 33-year-old woman with knee pain. (A) Axial T2-weighted fat-suppressed image shows a full-thickness cartilage defect at the patella (arrow) with subchondral bone marrow edema–like signal intensity (e). Cartilage was normal in the medial and lateral compartments at MRI and arthroscopy (not shown). (B) Coronal intermediate-weighted MR image shows marginal osteophytes at compartments with intact cartilage (arrowheads), which were stimulated to grow from the cartilage damage in the patella.

Synovitis

Osteoarthritis is not exclusively a disorder of the osteochondral unit and can be considered an organ failure of the whole joint, including the ligaments and synovium. The term synovitis describes hyperemia and hypertrophy of the synovium. Mild synovitis and joint effusion are frequently present in primary osteoarthritis as a response to cartilage damage. More severe synovitis raises the possibility of inflammatory arthropathy, which is a response to immune activity in the synovium, with notably different inflammatory cells and cytokines than those that are present in osteoarthritis (20,65). The synovium is well innervated, and the presence of synovitis correlates with patient pain (19,40).

Contrast-enhanced MRI is the most accurate method to assess for the presence and degree of synovitis. Normal synovium has a defined, smooth, thin morphology at contrast-enhanced imaging, typically with a thickness of less than 1 mm (Fig E4). The degree of synovial hypertrophy and hyperemia determine the degree and acuity of synovitis. Intravenous contrast material is slowly excreted from the synovium into the joint, which may artifactually thicken the appearance of the synovium after 10 minutes (68). At fluid-sensitive imaging, synovial tissue is often indistinguishable from joint fluid; however, there are two indirect indicators of synovitis that may be used at routine noncontrast MRI (Fig 16) (66,67). Edema along the subsynovial margins of the intra-articular fat pads, referred to as Hoffa-synovitis, is a sensitive but nonspecific indicator of joint synovitis (Fig 16) (66,67). Also, the size of a joint effusion may function as a surrogate for synovitis, referred to as effusion-synovitis. The presence of Hoffa-synovitis and effusion-synovitis were shown to predict the incidence and progression of osteoarthritis in multiple studies (69–73). Comparison with T1-weighted images may, at times, be helpful for distinguishing synovial thickening from fluid, but contrast-enhanced imaging remains the standard of care for assessment of synovitis.

Figure 16.

Hoffa synovitis and effusion synovitis in a 43-year-old woman. (A) T2-weighted fat-suppressed MR image shows edema along the subsynovial margins of the intra-articular fat pads, referred to as Hoffa-synovitis (hs), and what appears to be a medium-sized joint effusion, referred to as effusion-synovitis (es). Cartilage damage is noted in the central femoral trochlea (arrowheads). (B) Contrast-enhanced T1-weighted fat-suppressed MR image shows irregular synovial thickening, indicating synovitis (s) and a small joint effusion (e). Note how the inflamed synovial tissue may be indistinguishable from joint fluid, and how the size of the joint effusion is smaller than expected on the contrast-enhanced image (B). The normal morphology of the Hoffa fat pad is also well defined in this case, with deep superior and inferior recesses creating the characteristic whale tail shape. See Figure E4 for comparison with normal joint synovium.

Intra-articular Bodies

Because articular cartilage is primarily nourished by the synovial fluid, pieces of displaced damaged cartilage commonly continue to grow, forming rounded bodies. In this way, long-standing intra-articular cartilaginous bodies may become larger than their donor sites, and it is not uncommon to see large bodies in remote joint recesses, such as in a popliteal cyst of the knee or a bicipital recess of the glenohumeral joint. The central areas in these cartilaginous bodies may undergo ossification and are then referred to as osteocartilaginous bodies. Intra-articular bodies may be stable and adhered to the synovium or unstable and loose. A body can be confidently identified as unstable when it changes position at serial imaging, or it may be suggested when it is surrounded by fluid at cross-sectional imaging.

Conclusion

Understanding the anatomy of the osteochondral unit in the context of MRI sequence parameters is essential for accurate interpretation of articular cartilage damage. An appreciation of the differences that exist between MRI and arthroscopic interpretation of cartilage abnormalities can aid in surgical planning, patient education, and cross-disciplinary discussion of cartilage abnormalities.