Radiography of knee OA

The Kellgren and Lawrence (KL) scale is a semi-quantitative scoring system often used to assess radiographic OA severity, providing a global composite OA score on a 0–4 scale . It is widely used and well known, and has recently been validated in knees by using trained nonclinicians compared to experienced radiologists . The semi-quantitative Osteoarthritis Research Society International (OARSI) classification is most often applied when quantifying individual radiographic features of OA , but is more time consuming than scoring KL.

There are certain concerns on how the KL grades have inconsistently been labeled and applied in knee OA studies, especially the important cut-off value of KL grade 2 that represents definite OA. Longitudinal studies examining new-onset disease or disease progression according to the traditional KL scoring system may end up with joints characterized in a non-uniform manner across studies, and a modification has been proposed by Felson and colleagues to increase sensitivity to change . They suggest incident OA according to KL grade 2 as having both joint-space loss and osteophytes, and an alternative grade 2 for those having only incident osteophytes (“grade 2/ost”), as applied in the Osteoarthritis Initiative (OAI) and the Multicenter Osteoarthritis (MOST) study.

The limitations of CR in assessing OA structural progression in advanced disease have been highlighted in a recent study by Guermazi and colleagues . They reviewed MRI changes over 30 months in knee OA joints with radiographic KL grade 4 at the baseline. MRI frequently detected further cartilage loss and fluctuation of bone marrow lesions (BMLs), effusion, synovitis, and Hoffa-synovitis at the follow-up. Thus, KL grade 4 knees can still progress and the term “end stage” seems to be no longer appropriate.

Wirth and colleagues recently published data from the OAI, comparing different radiographic methods of measuring joint-space width (JSW) . They found a greater responsiveness when applying location-specific measures of JSW instead of standardized minimum JSW (mJSW) measures. Eckstein and colleagues presented similar data from the same cohort, showing that a change in location-specific JSW was a stronger predictor for knee replacement than mJSW . Both studies demonstrate a need to revise mJSW as a structural endpoint for DMOAD intervention trials by regulatory agencies, and instead use location-specific measurement of JSW in future studies. However, MRI-detected cartilage thickness measurements were shown to demonstrate superior sensitivity to change compared to the two radiographic methods , further supporting the need to include MRI in longitudinal OA studies.

Trabecular bone structure reflects the structural progression of OA. Trabecular bone is constantly remodeled in response to stress, and can be measured by fractal analysis of radiographs. Longitudinal studies have shown that alterations in the trabecular bone can predict incident radiographic JSN and MRI-measured cartilage thinning in knee OA, as well as predict the necessity for joint replacement . Wolski et al. have published a similar scoring method for hand OA, which could be potentially useful for early detection and prediction of hand OA . Being potentially modifiable , the integrity of subchondral bone may serve both as an outcome measure and a target for novel treatments, but it needs further validation against other outcome measures in longitudinal studies.

Radiography of hand OA

Traditionally, it has been understood that the association between CR features and symptoms is weak, and it is difficult to find an association between the total amount of pathology at patient-level (i.e. when summed from semi-quantitative CR scores) and self-reported pain at rest or during activities. At the joint level, however, a recent study has suggested a positive cross-sectional association between radiographic features of hand OA and hand pain , and radiographic progression, especially incident erosions, were associated with incident joint tenderness .

CR is often used in studies to confirm the diagnosis and severity of hand OA, and to separate erosive from non-erosive disease, which is often evaluated according to the Anatomic Phase Scoring System by Verbruggen et al., based on the assumption that hand OA undergoes predictable phases and can be scored subsequently , or the Ghent University Scoring System (GUSS) that shows better ability to detect progression over a shorter period of time . However, CR has a limitation in its two-dimensionality and thus cannot detect small erosions. A study compared radiography with MRI and found four times as many erosive joints with MRI . Another recent study found that erosions are frequently found on MRI and ultrasound in radiographic non-erosive joints, and inflammatory features are common in both erosive and non-erosive hand OA . The similar pattern of radiographic features further supports the theory that the difference between erosive and non-erosive OA is quantitative rather than qualitative, consistent with erosive OA being a severe form of hand OA rather than a separate entity .

The scoring of semi-quantitative JSN remains reader-dependent and limited to categorical grades of 0–3. Further, the joint spaces of finger joints are particularly small, making it difficult to assess the changes over a shorter duration. A semi-automated JSW measurement has been proposed, and a recent report by Damman et al. demonstrated that this method can detect small changes in joint space in early hand OA . However, when evaluating construct validity, the progression of semi-automated JSW showed weaker association with baseline inflammatory features than traditional JSN progression , and the application in hand OA clinical trials needs further evaluation.

Radiography of hip OA

Assessment of OA severity is often made semi-quantitatively according to the classification of Croft (grades 0–5) or KL (grades 0–4) , whereas quantitative mJSW in the superior region of the joint has been recommended as the structural endpoint in DMOAD trials . The JSW in hips is more strongly correlated with cartilage thickness than in knees, and is often used to assist in the diagnosis of OA with a cut-off value of ≤2.5 or ≤2.0 mm. However, data on mJSW being the most reliable or responsive measurement (as opposed to JSW in a fixed or another area of the joint) are limited.

Only 9–16% of hip joints with frequent pain in the Framingham cohort and OAI study showed evidence of radiographic hip OA, and 21–24% of joints with evident hip OA were frequently painful . Another study comparing MRI and radiography found severe cartilage damage in 26% of hip joints with KL score 0–1, and labral tear in 57% of the same joints . These findings indicate that in many cases, the diagnosis of hip OA may be missed or wrongly diagnosed if relied on radiography alone.

Another radiographic approach to hip OA is femoroacetabular impingement (FAI), in which a disturbed range of motion results from an abnormal contact between the head of femur and rim of the acetabulum. It is common in young adults and predisposes to later hip OA development . A recent study by Steppacher et al. demonstrated promising 10-year results for FAI treatment with surgical hip dislocation, osteoplasty, and labral reattachment, where 80% had not progressed to total hip arthroplasty or developed radiographic worsening of OA after 10 years . Radiographic predictors for failure were related to over- and under-treatment of acetabular rim trimming and should be used in follow-up after such procedures .

MRI and quantification of OA pathology

There are several available semi-quantitative MRI scoring systems for OA in the knee, hip, hand, and shoulder that have recently been reviewed in detail by Guermazi and colleagues . The knee differs from other joints by having more scoring systems available (reflecting the predominant focus on knee OA in the literature), each with strengths and limitations . Many of these scoring systems employ 0–3 or 0–4 semi-quantitative scores for compartment pathologies, and occasionally have within-grade scoring (e.g. 0.5 increments) in an attempt to improve responsiveness to change over time . Furthermore, teaching atlases and training sets of MRI scans improve reader reliability of scoring, and an electronic template that overlies MRI images has been demonstrated to improve reader reliability .

MRI as a valid method to assess inflammatory and structural features in hand OA also provides additional information over radiographs . Kortekaas and colleagues examined the validity of the Oslo Hand OA MRI score , and found good reliability and significant associations between pain and synovitis . They also compared MRI with ultrasound, favoring contrast-enhanced MRI in detecting synovitis. However, some limitations of the Oslo Hand OA MRI scoring system resulted in a revision and subsequent development of the OMERACT Hand OA MRI scoring system (HOAMRIS) , which included synovitis, erosive damage, cysts, osteophytes, cartilage-space loss, malalignment, and bone marrow lesions on 0–3 scales. Further validation may lead to exclusion of less important features from the proposed scoring system. Furthermore, an OMERACT MRI scoring system for thumb base OA is currently being developed and tested for reliability.

MRI can be used to assess the severity of hip OA by addressing structures associated with inflammation (synovitis and effusion), BMLs, structural damage (osteophytes, subchondral cysts, cartilage loss, and labral tears), and predisposing geometric factors (femoroacetabular impingement and hip dysplasia). These are assessed either by the Hip OA MRI Scoring System (HOAMS), which measures nearly all structures , or by the Hip Inflammation MRI Scoring System (HIMRISS), which selectively scores active inflammatory lesions and BMLs across several smaller subregions of the joint .

MRI and OA synovitis

As stated, MRI can visualize all affected joint structures in OA. Synovitis is increasingly considered important in OA and is a possible target for structure-modifying OA drugs . There are several methods for detecting and quantifying synovitis in OA. Because of possible side effects and associated costs, it would be desirable not to use contrast-enhanced MRI in OA studies; however, its use does improve the certainty of detecting inflamed synovium , which has been related to pain and shown to predict progression of radiographic JSN . Following intra-articular corticosteroid injection, contrast-enhanced synovitis also correlated better with subjective pain improvement than conventional measurement of synovitis .

However, many studies have employed non–contrast-enhanced MRI (often STIR sequences) to assess inflammation using signal changes in Hoffa’s fat pad and intra-articular fluid assessment; the latter has been validated against synovial fluid volume measured by arthrocentesis . MRI-detected effusion increases the risk of cartilage loss and pain , and a change in synovitis and effusion has been related to concurrent change in pain .

MRI and OA bone

MRI is the only imaging modality that shows BMLs ( Fig. 1 ), which are seen as areas with high signal intensity on T2-weighted fat-suppressed, proton density fat-suppressed, or STIR sequences, and with irregular margins. BMLs have been associated with increased loading due to obesity, joint malalignment, and meniscal pathology, as well as pain and structural progression in hand, knee, and hip OA . Lesions with a similar appearance are seen in systemic inflammatory joint diseases, where they are referred to as ‘bone marrow edema’ and histologically represent inflammatory osteitis. In OA, trabecular remodeling, necrosis of fatty cells, fibrosis, and extracellular fluid accumulation have been found histologically .

MRI with T1 (left) and proton density fat-suppressed (right) sequences, showing rupture of the medial meniscus (arrow) with adjacent tibial bone marrow edema (star), and fissures and fringed cartilage (arrowhead) as signs of early OA development.

A recent study found that the size of BMLs was important; large BMLs were associated with knee pain, structural damage, and further disease progression after 48 months, whereas small baseline BMLs were of less clinical relevance . A decrease in BMLs over time, which was related to a decrease in pain, did not predict improvement of structural aspects of OA . In fact, regions with a decrease in BMLs showed a trend toward increased cartilage defects and increased JSN. Baseline BML size may, therefore, be more important than longitudinal BML change in predicting OA progression, and BML change is possibly not an optimal endpoint.

Modern imaging analysis using statistical shape modeling utilizes the 3-D information in MRI to provide accurate quantification, and this has recently resulted in insights into the importance of bone shape in OA. In a large longitudinal cohort from the OAI, bone shape changed annually (in an almost linear fashion) in people with higher KL grades than those selected for persistent KL of 0, and the bone shape change was a more responsive measure of OA progression over time than radiographic JSN or MRI cartilage thickness . The validity of bone shape as a biomarker has been demonstrated in another study using OAI data to demonstrate that it predicts subsequent total knee joint replacement . Accurate spatial awareness has allowed us to see for the first time that BMLs were closely aligned with adjacent areas of cartilage loss, supporting a biomechanical origin for these lesions .

Previous radiographic studies have revealed that knee OA exhibits symmetry to some extent, but MRI features have been less studied for symmetry. A recent brief report found a higher degree of symmetry of BMLs and meniscal damage than expected by chance, enhancing the importance of systemic person-based risk factors for knee OA development in addition to local risk factors . It was recommended by these authors to use the contralateral knee as a control in randomized trials on unilateral treatment, such as intra-articular injections, as both knees are subjected to the same systemic risk factors.

MRI and early OA

MRI has been used to examine pre-radiographic changes in early OA. A recent case–control study by Roemer et al. reviewed repeated MRI images up to 4 years prior to the diagnosis of radiographic knee OA . They found that the presence of Hoffa synovitis, effusion synovitis, medial BMLs, and medial meniscal damage increased the risk of OA 2 years prior to incident radiographic OA, and that the number of present features increased the risk more than the presence of any single feature. Early MRI features such as effusion and BMLs that were found 3 or 4 years prior to onset of radiographic OA fluctuated more over time and were less strongly associated with incident OA.

Cartilage structure and composition through a novel range of MRI techniques (T2, T1rho, T2 relaxation times, and delayed contrast-enhanced MRI of cartilage (dGEMRIC)) have been intensively studied . Interestingly, in a study evaluating the years prior to incident radiographic OA, MRI cartilage damage only became significantly predictive of OA development one year prior to incident radiographic OA—but so did almost any other abnormal morphological MRI feature . In cases of existing radiographic OA, however, MRI-defined cartilage thinning (presence and worsening over time) is a robust predictor for radiographic knee OA progression .

The presence of meniscal damage has long been associated with the onset of OA development and progression. MRI has highlighted that meniscal damage is common in the elderly, and several studies support the important role of any meniscal damage, including tears, maceration, and substance loss inducing OA initiation , with risks increased (dose-dependently) by even minor valgus malalignment . Meniscal extrusion detected by MRI has also been reported as a separate risk factor for OA progression in the tibiofemoral joint, but not in the patellofemoral joint .

3D bone shape has also been demonstrated to predict incident radiographic OA as well as radiographic and pain progression in large analyses from OAI . This is a novel and promising imaging biomarker, and further studies will show whether 3D bone shape analysis can predict those with greater risk of rapid OA progression or serve as an intervention marker in clinical trials.

Finally, muscle strengthening has long been a highly effective therapy for OA knee pain, but muscle has been relatively poorly studied. There is emerging literature on muscle structure in OA, with a recent study examining characteristics of the vastus medialis muscle, which could identify patients at higher risk for OA progression . Baseline vastus medialis changes (having low area and high fat percentage in combination with obesity) were additional risk factors beyond age, sex, body mass index, and other factors, toward structural OA progression .

Ultrasound of cartilage

Ultrasound assessment of cartilage has some limitations. First, because of acoustic window only a limited area of the cartilage is available for scanning. In addition, as pathology in the cartilage develops, the physics of sound waves in this tissue changes. Moreover, the scanning angle is of major importance: linear measurements may be distorted if not performed perpendicular to the cartilage lining . Hence, the evaluation of cartilage thickness may be difficult in OA patients. Despite these limitations, several studies on cartilage in OA have been performed. They have shown ultrasound to be reliable and valid for the evaluation of cartilage pathology (i.e., altered echogenicity and thinning) in cross-sectional studies, primarily of large joints such as the knee ( Fig. 2 ) . The cartilage of knee joints has a normal thickness of about 2 mm, and superficial regions can easily be evaluated by ultrasound . However, in the small joints of the hands, there are conflicting results, with an important finding being that ultrasound may differentiate between normal and pathological finger-joint cartilage, whereas semi-quantitative scoring is not reliable .

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

Imaging in rheumatic and musculoskeletal conditions Imaging in the diagnosis and management of axial spondyloarthritis Imaging in the diagnosis and management of peripheral psoriatic arthritis—The clinical utility of magnetic resonance imaging and ultrasonography Musculoskeletal interventional procedures: With or without imaging guidance? Imaging of regional pain syndromes; from syndromes to conditions using imaging? Imaging in paediatric rheumatology: Is it time for imaging?

Stay updated, free articles. Join our Telegram channel

Join