Dr. Chow is currently a fourth-year Radiology Resident at Brigham and Women's Hospital/Harvard Medical School in Boston, MA. He graduated summa cum laude from McGill University, Montreal, Canada, with a BSc in Biochemistry in 1994 and received his MD and Master of Surgery from McGill University in 1998. Dr. Chow will remain at Brigham and Women's Hospital to begin a fellowship in Musculoskeletal Radiology in July 2003.

Magnetic resonance (MR) imaging of the knee is a useful modality for assessing internal derangements of the knee. Recently, imaging of articular cartilage has been studied intensely owing to advances in surgical therapy for focal articular cartilage disease. MR imaging developments have shown clear advantages over other modalities in noninvasively detecting and grading articular cartilage defects. This report will focus on MR imaging techniques, pitfalls and limitations in detecting cartilage lesions, current therapy, and the future of articular cartilage imaging.

In recent years, intense interest in articular cartilage imaging has been generated by several advances in the surgical treatment of focal articular cartilage injury. In the United States, there is a high prevalence of articular cartilage disease; >75% of the population >65 years of age have evidence of osteo-arthritis and 1.6% of the entire population have symptomatic osteoarthritis of the knee. ¹ In a retrospective review of all arthroscopies performed at a single institution during a 4-year period, >60% of 31,516 knee arthroscopies showed articular cartilage damage, and 37% of patients <40 years of age showed isolated articular cartilage injury without ligamentous or meniscal pathology. ² It is generally believed that, over time, these articular cartilage defects eventually lead to osteoarthritis, the incidence of which is extremely common in the United States and increases with age. ^3-5 Osteoarthritis costs the U.S. economy nearly $65 billion per year in direct expenses, lost wages, and lost productivity. ⁶ After cardiovascular disease, osteoarthritis is the second leading cause of chronic disability in America. ⁷ This has created a necessity for accurate noninvasive detection and evaluation of articular cartilage disease and monitoring of therapy. A complete evaluation is needed to select the criteria that will best determine which cartilage defects are amenable to treatment and which patients will benefit from surgical treatment. ⁸

The clinical symptoms associated with articular cartilage defects are variable and range from almost no pain to chronic pain, joint effusions, and immobility. For cartilage evaluation, physical examination and history are inaccurate and radiography is an indirect and insensitive modality. While arthroscopy is the current standard method for assessing articular cartilage integrity, certain disadvantages to this method are evident. Surveillance by arthroscopy is a minimally invasive procedure but requires trocar insertion and anesthesia. Arthroscopy is also limited to the evaluation of surface integrity and cannot accurately assess subchondral pathology. ^9,10 In addition, arthroscopy gives the orthopedic surgeon insufficient preoperative information about what to expect from the surgery. This hinders their ability to counsel patients regarding treatment options and rehabilitation.

Imaging plays an important role in noninvasively assessing patients who are suitable for surgical treatment. Magnetic resonance (MR) imaging plays an integral and complementary role in noninvasively assessing the integrity of articular cartilage beneath the surface articulations and within subchondral bone. This discussion of MR imaging of articular cartilage will focus mainly on the knee because research has been concentrated in this area.

Modalities for imaging articular cartilage

Radiography is an indirect method of assessing cartilage by measuring joint narrowing. ¹¹ It is inaccurate and insensitive for pure articular cartilage disease, but may be revealing in osteochondral injuries, such as osteochondritis dissecans. ¹² Ultrasound has been used in assessing ligaments and other soft-tissue structures in the knee; however, it is unable to delineate cartilage abnormalities clearly. ¹³ Computed tomographic arthrography has also been used in assessing internal derangements of the knee, but is more invasive than conventional MR imaging. Optical coherence tomography--a high-resolution micron scale imaging method that measures the intensity of back-reflected infrared light--is being researched intensely but has not been clinically proven. ¹⁴ Surgical arthroscopy is regarded as the standard method, but it is invasive and costly. To this day, conventional MR imaging has been the mainstay of noninvasive articular cartilage imaging.

MR imaging techniques

The advantages of MR imaging are cross-sectional multiplanar capability, high spatial resolution, and good contrast resolution. However, small articulations and early cartilage changes are still challenging. Many different clinical imaging sequences are available today. These imaging sequences rely on the differences in contrast resolution between articular cartilage and adjacent structures such as joint fluid, subchondral bone, and menisci, as shown in Table 1. ¹⁵ The mainstay of MR pulse sequences optimized for articular cartilage visualization are fast-spin echo (FSE) (or turbo-spin echo [TSE]), ^16,17 and three-dimensional (3D) spoiled gradient-recalled echo (3D-SPGR) (or fast low-angle shot [FLASH]). ^18,19 These two pulse sequences exploit the dark cartilage/ bright fluid and bright cartilage/dark fluid differences, respectively. ¹⁵ For evaluating focal cartilage defects, MR arthrography, which uses the dark cartilage/bright fluid technique, has generated much interest; however, it has the added advantage of joint distension. ²⁰ Spin-echo (SE) imaging, routinely used in imaging of the knee, reveals a bilaminar appearance of cartilage with a dark basal layer and an intermediate signal upper layer. However, owing to the similar signal intensity of subchondral bone with the basal cartilage layer, the thickness of the cartilage is underestimated. ¹⁵ Although the cartilage-fluid interface is well visualized on T2-weighted SE images, poor definition of the cartilage-subchondral bone interface limits its use in the evaluation of articular cartilage. ¹⁵ This report will concentrate on FSE and 3D-SPGR techniques.

New MR imaging techniques are currently under development in order to optimize the sensitivity and specificity of detecting and grading articular cartilage defects. These new developments include magnetization transfer (MT), delayed gadolinium-enhanced MR imaging of cartilage (dGEMRIC), sodium nuclear magnetic resonance (NMR) imaging, T2 mapping, driven equilibrium Fourier transform (DEFT), and fluctuating equilibrium magnetic resonance imaging (FEMR). Although these imaging protocols have yet to be utilized in routine clinical practice, the potential to understand the pathophysiology of focal articular cartilage disease and the pathogenesis of osteoarthritis holds great promise. These imaging techniques will be discussed in detail later in this report.

FSE (TSE) with fat suppression

Fast-spin-echo technique generates a higher signal-to-noise ratio and improved image contrast when compared with SE technique. ^16,17,21 The superficial layers of cartilage appear intermediate in signal intensity while joint fluid appears bright. With fat suppression, the basal layer of cartilage appears slightly more hyperintense compared with subchondral bone and bone marrow, but more hypointense when compared with joint fluid. Marrow edema and contusions can be detected easily by adding a chemical shift selective fat-suppression technique. This technique can also be used to adequately evaluate other structures within the knee, including menisci, ligaments, and bone marrow. Thus, FSE with fat suppression is a useful technique in routine knee imaging. Metallic artifact is also much decreased with FSE as compared with gradient-echo techniques. All of this is achieved with a shorter acquisition time compared with SE, with certain investigators advocating an effective echo time (TE) of approximately 34 msec for the greatest sensitivity. ²¹ With collagen loss and increasing water content in degenerating cartilage, T2 prolongation occurs and proton density increases, resulting in higher signal intensity. ¹⁶ The disadvantages of FSE include decreased spatial resolution compared with 3D-SPGR due to the two-dimensional (2D) acquisition and blurring artifacts of short T2 structures, such as cartilage and menisci. ¹⁵ There are technical developments being investigated for 3D-FSE, and with shorter echo spacing available, it is possible to reduce blurring artifacts for short TE imaging.

3D-SPGR (3D-FLASH) with fat suppression

Numerous advantages are achieved with 3D-SPGR imaging of articular cartilage. A consistent trilaminar appearance of the articular cartilage--bright against dark joint fluid, subchondral bone, and bone marrow--is evident. ²² Greater spatial resolution and contrast-to-noise (CNR) ratio are attained because of the 3D acquisition. ²³ The higher spatial resolution enables multiplanar reformatting, which allows for additional manipulation and volumetric quantification of cartilage volume using commercially available software. ^24-27 Curved surfaces of articular cartilage, such as the femoral trochlea and patellar cartilage, can also be evaluated more accurately with standard and user-specified planes. One disadvantage of this technique is greater susceptibility to metallic artifacts, limiting its use in postoperative knees. In addition, evaluation of other important knee structures, such as ligaments and menisci, are not assessed adequately; therefore, this sequence is used in addition to other routine imaging sequences of the knee. Furthermore, 3D-SPGR also has a longer acquisition time than FSE techniques.

MR arthrography

The concept of MR arthrography parallels conventional arthrography in that distension of the joint with a contrast agent helps to visualize and delineate surface abnormalities more precisely. With MR arthrography, enhancement of joint fluid can be achieved with either direct injection of a dilute solution of gadolinium into the joint (intra-articular [IA] or direct MR arthrography), or by intravenous ([IV] or indirect MR arthrography) injection of contrast material with subsequent diffusion of the contrast into the joint. ²⁸ This contrast diffusion out of the blood pool and into the extracellular space is achieved with active exercising of the joint for approximately 10 to 20 minutes before imaging for the greatest enhancement of joint fluid. This technique has been used to greatest advantage in evaluating the shoulder joint for rotator cuff and glenoid labrum tears. ²⁹ Subsequently, it has also been shown to be an excellent method for evaluating articular cartilage. ^30,31 By revealing enhanced joint fluid beneath an osteochondral lesion, it is currently the most useful technique in determining stability of an osteochondral fragment. ³² It has also been shown to have high specificity and accuracy in the detection and grading of patellar cartilage defects when compared with conventional SE and 3D-SPGR imaging. ²⁰

Some of the disadvantages of direct (IA) MR arthrography are that it is more invasive and more time-consuming than indirect (IV) MR arthrography. Also, direct MR arthrography requires a radiologist, assisted by a techologist or a nurse, to inject the diluted gadolinium solution into the joint under fluoroscopy, using a sterile technique. Total time required for the technique depends on familiarity with conventional arthrography techniques, but is usually approximately 20 to 30 minutes. Imaging time is the same for both direct and indirect MR arthrography. With indirect MR arthrography, there is no direct control of the degree of joint distension.

Grading and detection of cartilage defects

There are many grading schemes for focal articular cartilage defects. The modified Outerbridge classification, adopted from its original use in grading chondromalacia patella, is based on morphological appearance at arthro-scopy and uses a 5-point system ³³ : grade 0 represents normal cartilage; grade 1 lesions display softening of the cartilage; grade 2 lesions show shallow ulceration or blister-like swelling; grade 3 lesions show partial-thickness fibrillation or deep ulceration not extending to bone (Figure 1); and grade 4 lesions are ulcerations with exposure of the subchondral bone (Figure 2). Another grading scheme, proposed by Bauer and Jackson, ³⁴ categorizes articular cartilage injuries according to arthroscopic patterns of cartilage fracture. MR imaging grading schemes correspondingly follow this system: grade 1 lesions show signal abnormality within the cartilage; grade 2 lesions have surface irregularity; grade 3 lesions show partial thickness loss (<50%); and grade 4 lesions show full-thickness cartilage loss to subchondral bone (Figure 3).

Detection of articular cartilage lesions on MR imaging relies on signal intensity changes and contour abnormalities. Fat-suppressed 3D-SPGR images are useful for assessing surface undulations or irregularities, whereas FSE images are useful for detecting abnormal signal intensity and contour changes if a joint effusion is present. MR arthrography can show both contour abnormalities and signal changes within the substance of the articular cartilage. However, MR imaging is relatively insensitive to grade 1 lesions. ^17,19,20 With the use of FSE, 3D-SPGR, or MR arthrography, the sensitivity rate for detecting lesions improved as the grade of cartilage lesions increased: the sensitivity rates ranged from 32% for grade 1, 72% for grade 2, 94% for grade 3, and 100% for grade 4. ¹⁷ Published data have shown sensitivity and specificity for the detection of cartilage lesions to range from 81% to 100% for correlation within 1 arthroscopic grade. ^16-21,35

Pitfalls and limitations in MR articular cartilage imaging

Pitfalls in detecting chondral lesions on MR imaging include a focal bright signal in articular cartilage at 55º relative to the main magnetic field. This is due to the highly organized structure of collagen fibers in cartilage causing the T2 relaxation anisotropy--the so-called "magic-angle effect." ³⁶ An attempt to reduce this artifact can be made by increasing the TE (increasing T2 weighing). Chemical shift artifacts can also arise at fat-cartilage interfaces and can lead to spatial misregistration in the frequency direction; however, this can be eliminated by fat suppression. ²³ Truncation artifacts develop from the undersampling of signal arising from small objects with high contrast. This occurs with 3D-SPGR, producing a false laminar appearance of cartilage. ^37,38 Finally, metallic susceptibility artifact arises in the postoperative knee; gradient-echo imaging is much more sensitive to this artifact than is the FSE technique, which limits the ability of 3D-SPGR in this context.

The spatial resolution and image contrast of current imaging sequences somewhat limit the determination of cartilage defects. When thin fissures and flap tears are present, adequate image contrast and spatial resolution are necessary to identify a thin cleft at the articular surface (Figure 4). Assessment of curved articular surfaces also presents difficulty owing to partial volume averaging effects and may lead to overestimation or underestimation of defect size and severity. Differentiation of full-thickness from partial-thickness defects requires absolute image contrast between the deep zones of cartilage and subchondral bone. Despite these limitations, MR imaging is reasonably successful in the accuracy of grading cartilage lesions. Various studies have shown >85% sensitivity and >90% specificity for moderate-to-severe cartilage lesions (grades 2 to 4). ^16-21 Given the current limitation of spatial resolution and partial volume averaging, the spatial resolution of lesion dimensions has been reported to be accurate to 0.5 mm in vivo. ²⁵ These spatial and contrast resolution requirements emphasize the need for high-field MR imaging systems with dedicated coils to improve the signal-to-noise ratio.

Clinical challenges and surgical therapy

Currently, there is little consensus as to which cartilage defects can be managed conservatively versus lesions that are amenable to surgical therapy. Classification of articular defects into degenerative, inflammatory, and posttraumatic etiologies helps define the patient population and type of chondral defects or injury that will benefit most from surgical treatment. ³⁹ Degenerative chondrosis or advanced chondromalacia changes are typically multifocal defects that have shallow angles and indistinct margins, usually occurring along the posterior condyles. Meanwhile, inflammatory arthritis usually shows diffuse and uniform articular cartilage thinning without focal defects. This is presumably due to the effect of inflammatory pannus and catabolic enzymes within the joint fluid. ⁴⁰

Acute chondral/osteochondral injury characteristically shows an acute angle of the lesion wall with sharp borders occurring at weight-bearing surfaces. Pure chondral fractures occur more frequently in the skeletally mature patient because of shearing forces propagating along the tidemark, a natural cleavage plane between the nonmineralized cartilage layers from the deepest calcified layer of cartilage and the subchondral bone plate. ⁴¹ Osteochondral fractures generally occur in the skeletally immature patient because tidemark has not been defined, thus allowing injuring forces to penetrate the calcified cartilage and subchondral bone plate. ⁴² It has been suggested that approximately 78% of acute chondral/osteochondral injuries are associated with focal subchondral edema at the site of injury, and that these lesions are associated with treatable cartilage defects. ⁴³

Some investigators have proposed an algorithm for symptomatic full-thickness articular cartilage defects with surgical repair as first-line treatment (Figure 5). ⁴⁴ Medical or conservative therapies have been ineffective at preserving joint function and stopping progressive degeneration that may lead to osteoarthritis in this clinical situation. Articular cartilage, or hyaline cartilage, is histomorphologically composed of a well-organized network of type II collagen fibers and interspersed chondrocytes in a hydrated gel matrix that contains proteoglycans and hyaluronate. ³ Articular cartilage has little regenerative or reparative capacity when damaged; when repair tissue is formed, it is fibrocartilage and its biomechanical properties are significantly different from hyaline cartilage. ³ Because fibrocartilage is unable to withstand joint loading and degenerates over time, clinical symptoms return. ³³ This often requires reconstructive knee surgery and eventually a prosthesis. ³

Many different surgical approaches for decreasing symptoms, restoring joint function, and preventing osteo-arthritic degeneration have been attempted with variable success. These include debridement and lavage, abrasion and microfracture techniques, periosteum and perichondrium transplantation, osteochondral autograft and allograft plug transfer (mosaicplasty), and autologous chondrocyte implantation. ^45,46 With the new advances in surgical treatment of focal cartilage defects, MR imaging has become an important noninvasive means of monitoring the effectiveness of treatment (Figure 6). Currently, the accepted modality for monitoring treatment is arthroscopy with biopsy, an invasive and expensive means of follow-up.

Future of articular cartilage imaging

The appropriate deployment and selection of newer treatment interventions for osteoarthritis depends on the development of better methods for the longitudinal assessment of the disease process. New MR imaging sequences are contributing to our understanding of the natural history of focal cartilage defects and the progression to osteoarthritis. Although these sequences have yet to be used in routine clinical imaging, they show great promise in understanding the biochemical, physiologic, and functional characteristics of articular cartilage.

Magnetization transfer (MT) imaging uses an off-resonance radiofrequency pulse to interrogate the integrity of the collagen matrix in cartilage, showing decreased signal in an intact collagen matrix. ^47,48 Several studies have reported that MT imaging can be used to differentiate reliably among articular cartilage, adjacent joint fluid, and inflamed synovium. ^10,47,48 Delayed gadolinium-enhanced MR imaging of cartilage and sodium imaging methods can indirectly measure the glycosamino-glycan (GAG) concentration in articular cartilage, a constituent of the proteoglycan macromolecule that largely accounts for the biomechanical compressive strength of cartilage. ⁴⁹ The functional significance of GAG-depleted cartilage is the loss of the compressive strength of articular cartilage, suggestive of the early stages of cartilage degeneration. ⁴⁹ T2 mapping of articular cartilage has also been investigated for detecting early degeneration (Figure 7). Spatial distribution measurements of T2 relaxation times can be mapped and may reveal areas of increased water content, correlating with damage to the collagen matrix. ⁵⁰ Other promising pulse sequences include driven equilibrium Fourier transform imaging and steady-state free-precession methods--such as fluctuating equilibrium MR imaging (Figures 8 and 9). Both of these techniques use bright synovial fluid signal while maintaining cartilage signal based on the ratio of T1/T2 in tissue with high spatial resolution due to the 3D-acquisition technique. ⁵⁰

In addition to these novel pulse sequences, applications such as quantitative image processing and analysis techniques of cartilage thickness and volume with 3D reconstruction allow monitoring of medical and surgical therapies. ^10,27,51 Biomechanical analysis can also be obtained from a computer-generated model of a subject with fiducial markers climbing stairs, which may give insight into the "wear and tear" aspects of degenerative chondrosis. ^52,53 All of these pulse sequences and novel applications may play a significant role in evaluating progress in new chondro-regenerative drug therapy, such as intra-articular hyaluronate or oral chondroitin sulfate, ³³ or new surgical treatments. ^45,46

Conclusion

MR imaging, with its superior soft-tissue contrast, is the best technique available for assessing normal articular cartilage and cartilage lesions. With this modality, noninvasive imaging can be performed before treatment to determine the best possible therapy and after treatment to assess the response to therapy. With the promise of higher field MR imaging systems and new imaging sequences to evaluate articular cartilage with greater signal-to-noise ratio, CNR, and spatial resolution, radiologists will be better prepared to assess the clinical significance of early diagnoses and treatment options.

Acknowledgment

The author wishes to thank Dr. Carl S. Winalski (Brigham and Women's Hospital, Harvard Medical School) for his support in reviewing the manuscript and generously providing the images and technical assistance. Special thanks go to Drs. Philipp Lang, John A. Carrino, and Frank J. Rybicki (Brigham and Women's Hospital, Harvard Medical School) for reviewing the manuscript and providing thoughtful input and assistance. Dr. Garry Gold (Stanford University) and Dr. Timothy J. Mosher (Pennsylvania State University) also provided additional images.