 |
|
|
 |
Editorial
Magnetic Resonance Imaging for Spondyloarthritis — Avoiding the Minefield
WALTER P. MAKSYMOWYCH, FRCPC,
Department of Medicine, University of Alberta;
ROBERT G.W. LAMBERT, FRCPC,
Address reprint requests to Prof. W.P. Maksymowych, Department of Medicine, 562 Heritage Medical Research Building, University of Alberta, Edmonton, Alberta T6G 2S2. E-mail: walter.maksymowych@ualberta.ca Diagnostic evaluation and management of inflammatory arthritis based primarily in the axial skeleton is arguably more problematic than in most other forms of inflammatory arthritis. Physical findings are typically confined to later stages of disease and are insensitive to change. Laboratory investigations are essentially limited to acute phase reactants and HLA-B27, which alone have limited diagnostic value. Plain radiographic abnormalities occur late, are insensitive to change, and depict structural abnormalities only1. Consequently, diagnostic evaluation and longterm management is still largely dependent on historical inquiry and subjective patient self-report. The advent of magnetic resonance imaging (MRI) has proven to be a milestone in the evaluation of spondyloarthritis (SpA) through its ability to depict objective features of active inflammation, thereby facilitating earlier diagnosis and ongoing management, and permitting quantitative assessment of extent and severity of spinal inflammation. There is clearly increasing awareness of the clinical utility of MRI, as reported in a recent survey of Canadian rheumatologists, although most are unfamiliar with imaging techniques, the spectrum of abnormalities observed, and pitfalls in assessment.
The availability of numerous MR sequences and approaches to imaging has the potential to confuse and deter clinicians in the interpretation of MR images. We recommend that rheumatologists focus on 2 MR sequences that provide most of the imaging details required for the assessment of SpA, require the least imaging time, and incur the least cost. In T1-weighted images (T1WI), normal bone marrow fat is bright, while bone is dark, and these contrasting features contribute to the enhanced anatomical resolution provided by T1 compared to other MR sequences. Loss of the fat signal within bone marrow is a nonspecific finding and may be observed following fat replacement by most pathological processes such as sclerotic bone, fibrosis, tumor, cyst, and inflammation. However, the presence of acute inflammation in the marrow may be obscured by the signal from marrow fat. The short tau inversion recovery (STIR) sequence is a T2-weighted sequence that suppresses the fat signal so that the marrow now appears dark, and any fluid within the marrow due to active inflammation is relatively bright (Figure 1). The signal detected by the STIR sequence is a nonspecific indicator of increased free water content (both intra- and extracellular), so that the differential diagnosis for abnormal signal on this sequence includes any pathology associated with fluid accumulation in bone marrow (e.g., tumor, type I Modic change related to disc degeneration, infection, and trauma). Images of the spine are typically obtained in the sagittal orientation, and STIR images can be readily discerned from T1WI by the presence of high signal emanating from the cerebral spinal fluid in the spinal canal. Images of the sacroiliac (SI) joints are typically obtained in a tilted coronal orientation along the long axis of the sacral bone. Consecutive slices in this plane allow visualization of the cartilaginous portion of the SI joint, which is convex-shaped with the apex facing antero-inferiorly.
One of the earliest features of sacroiliitis is the presence of subchondral bone marrow edema in the postero-inferior region of the synovial portion of the joint, as seen on STIR images (Figure 2). This reflects the subchondral marrow inflammation reported on histopathological examination of early sacroiliitis2. The lesion develops into an erosion as defined by loss of the marrow fat signal on T1WI together with loss of subarticular bone. Healing of erosion/inflammation or progression to ankylosis is typically associated with repopulation of the marrow space by fat (Figure 3). The spectrum of abnormalities is easiest to interpret when T1WI and corresponding STIR images are viewed concurrently. Characteristic abnormalities observed in the spine may also be seen on STIR images, which depict increased marrow signal at the anterior and posterior corners of the vertebrae, adjacent to the insertion of the annulus fibrosis, corresponding to the Romanus lesion observed histopathologically (Figure 1). A lesion that had been underestimated prior to the advent of MRI is the occurrence of spondylodiscitis characterized as inflammatory marrow signal within vertebrae adjacent to the central portion of the vertebral endplate, which occurs in up to 15% of patients (Figure 4)3. Acute lesions may also occur in the posterior elements of the spine including the facet joints, pedicles, and ligamentary insertions at spinous and transverse processes (Figure 1). We have also documented a high frequency of acute changes in the costovertebral joints that may represent the major portion of the inflammation occurring in the thoracic spine (Figure 5)4. Syndesmophytes tend to be less well visualized on MRI than by radiography, and a systematic comparison showed no advantages of MRI over radiographs for the assessment of chronic lesions5. Pitfalls to the use of MRI can be broadly subdivided into those that are primarily technical and those that relate to interpretation. MRI is subject to numerous artefacts that are invariably present on these types of scans. Motion artefact is a particular problem and this includes any patient movement (including breathing, swallowing) and physiological motion (e.g., arterial, venous, or cerebrospinal fluid pulsation artefact). These commonly simulate low-grade inflammation and so are particularly relevant to patients with SpA (Figure 6). The spine is typically assessed in 2 halves to reduce imaging time, an upper cervico-thoracic and a lower thoraco-lumbar. This results in a relatively large field of view so that the cervical vertebrae appear small, limiting anatomical resolution and detectability of lesions. Detection of abnormalities also depends on viewing conditions, which are much superior when the image is displayed on a computer monitor (in contrast to hard copy film), allowing the observer to manipulate the display properties. The setting of the thresholds and grey-scale (windowing) for viewing markedly affects the visualization of subtle lesions, and is particularly important in the assessment of STIR images, where signal to noise ratios are usually limited. On suboptimal quality scans, it may be quite difficult to reliably determine what constitutes abnormal increased signal intensity referable to background. This is particularly germane to the development of scoring systems for axial inflammation, where explicit definition of what constitutes abnormal inflammatory signal in the SI joint and spine, such as in the SpA Research Consortium of Canada (SPARCC) MRI index6,7, is essential for observer consistency.
Pitfalls in interpretation are often related to a lack of systematic assessment, limitations in anatomical resolution, failure to apply consistent definitions of abnormalities, and fixed postural abnormalities (e.g., scoliosis) in the patient. Examination of the lateral and posterior segments of the spine that may harbor inflammation in the costovertebral joints, facet joints, pedicles, and entheses may be difficult due to the complexity of the anatomy. Training is essential if readers of MRI are to avoid frequent misinterpretation of scans, and even radiologists with considerable experience in MRI are frequently unaware of the nuances of interpretation in this field. A systematic approach to detection of lesions in SpA is essential, and a diagnostic imaging protocol has been posted on our website [www.arthritisdoctors.org]. We have recently shown that when the spine is assessed systematically, a major portion of the inflammation in the thoracic spine is primarily observed laterally, such as in the costovertebral joints4. Some MRI reports have primarily focused on descriptions of individual lesions in a single dimension and described lesions primarily in vertebral bodies8. This may be sufficient in framing a scoring system but is clearly insufficient for diagnostic evaluation. It is important to emphasize that interpretation of abnormalities on MRI (e.g., erosions) may be rather different from radiography, and few reports clearly define such abnormalities. Where this has been done, an erosion has been defined as loss of marrow fat signal together with loss of the overlying subchondral bone on a T1WI9. However, because bone is dark, loss of subchondral bone may not be readily apparent and sometimes may even be more evident on STIR images. Of particular relevance to the interpretation of spinal abnormalities, degenerative discal lesions are common in the age group with SpA and may include loss of the discovertebral endplate with adjacent bone marrow edema that may be indistinguishable from the spondylodiscitis of SpA. Unfortunately, there have been no reports that systematically analyzed the sensitivity and specificity of lesions in the spine. Inflammation in bone may also be detected using T2-weighted spin-echo sequences with fat suppression or with T1WI with fat suppression following gadolinium contrast enhancement (T1FS-postGd). Contrast-enhanced sequences do offer specific advantages that may be evident in early inflammatory lesions; for example, inflammatory changes in soft tissues and at sites of enthesitis may be seen first or be easier to identify with T1FS-postGd2. However, comparative studies have yet to be conducted, and the method of fat suppression required is problematic as it is frequently subject to significant artefact when scanning with a large field of view in the spine (Figure 7). Consequently, most approaches to the quantitative assessment of acute MRI lesions are primarily focused on the evaluation of STIR images and can generally be subdivided into 2 related methods (ASspiMRI, Berlin score) that score individual lesions in a single dimension (anterior-posterior extent for the spine)7 and one method (SPARCC MRI index) that scores lesions in 3 dimensions6. Inflammatory lesions within bone are often highly asymmetrical and therefore are more precisely quantified with a method that systematically assesses lesions in several dimensions. With the former methods, evaluation of the vertebral body of all 23 vertebral segments is mandatory, necessitating the scoring of regions subject to artefact and/or limited anatomical resolution. The median number of affected vertebral segments is typically 5–6 in a cohort of severely affected patients that would be regarded as eligible for trials of anti-TNF therapies10. In patients with many levels affected, the less conspicuous lesions are often subtle, and therefore the scoring of these lesions is less responsive to therapy. It is therefore not surprising that a comparison of the SPARCC scoring method for all 23 DVU versus only the 6 most severely affected discovertebral unit shows similar reliability but better responsiveness for the latter approach10. One can conclude that MRI is now established as an essential component of the clinician's toolkit, although in need of further educational initiatives. For the clinical researcher, much greater scrutiny is required to address issues of standardization of methodology and pitfalls in assessment. However, there is also a need to recognize and incorporate the unique opportunities afforded by this imaging modality in the evaluation of lesions in patients with SpA.
2. Muche B, Bollow M, Francois RJ, Sieper J, Hamm B, Braun J. Anatomic structures involved in early- and late-stage sacroiliitis in spondylarthritis: a detailed analysis by contrast-enhanced magnetic resonance imaging. Arthritis Rheum 2003;48:1374-84. [MEDLINE] 3. Braun J, Sieper J, Bollow M. Imaging of sacroiliitis. Clin Rheumatol 2000;19:51-7. [MEDLINE] 4. Rennie WJ, Dhillon SS, Conner-Spady B, Maksymowych WP, Lambert RGW. Standard MRI assessment of spinal inflammatory lesions in AS may omit a significant component of inflammation in the thoracic spine. Ann Rheum Dis 2006;65 Suppl 2:534. 5. Braun J, Baraliakos X, Golder W, et al. Analysing chronic spinal changes in ankylosing spondylitis: a systematic comparison of conventional x-rays with magnetic resonance imaging using established and new scoring systems. Ann Rheum Dis 2004;63:1046-55. [MEDLINE] 6. Maksymowych WP, Inman RD, Salonen D, et al. Spondyloarthritis Research Consortium of Canada magnetic resonance imaging index for assessment of spinal inflammation in ankylosing spondylitis. Arthritis Rheum 2005;53:502-9. [MEDLINE] 7. Maksymowych WP, Inman RD, Salonen D, et al. Spondyloarthritis Research Consortium of Canada magnetic resonance imaging index for assessment of sacroiliac joint inflammation in ankylosing spondylitis. Arthritis Rheum 2005;53:703-9. [MEDLINE] 8. Braun J, Baraliakos X, Golder W, et al. Magnetic resonance imaging examinations of the spine in patients with ankylosing spondylitis, before and after successful therapy with infliximab. Evaluation of a new scoring system. Arthritis Rheum 2003;48:1126-36. [MEDLINE] 9. Bird P, Conaghan P, Ejbjerg B, et al. The development of the EULAR-OMERACT rheumatoid arthritis MRI reference image atlas. Ann Rheum Dis 2005;64 Suppl 1:i8-10. 10. Maksymowych WP, Dhillon SS, Park R, Salonen D, Inman RD, Lambert RGW. Validation of the Spondyloarthritis Research Consortium of Canada (SPARCC) MRI Spinal Inflammation Index: Is it necessary to score the entire spine? Arthritis Rheum (in press).
|
