Consensus statements on the imaging of axial spondyloarthritis in Australia and New Zealand.

Abstract

Spondyloarthritis (SpA) describes a group of related inflammatory conditions, including ankylosing spondylitis (AS), psoriatic arthritis, reactive arthritis, SpA associated with inflammatory bowel disease and undifferentiated SpA.1 Classification criteria have been developed and validated by the Assessment of SpondyloArthritis international Society (ASAS) to distinguish axial-predominant SpA from peripherally predominant SpA (Fig. 1).1, 2 These criteria contribute to diagnosis, but are not ideal diagnostic criteria as they possess only moderate sensitivity.3 Diagnosis of axial SpA should be established by a rheumatologist, after careful consideration of these criteria and individual patient factors.4 The axial SpA classification addresses inflammatory disease of the axial skeleton, and consists of radiographic axial SpA and non-radiographic axial SpA. Radiographic axial SpA is equivalent to AS and requires definite radiographic sacroiliitis according to the modified New York (mNY) criteria and at least one clinical feature of SpA.4 Non-radiographic axial SpA requires either sacroiliitis on MRI and one clinical feature of SpA, or HLA-B27 and two clinical features of SpA. Non-radiographic axial SpA is associated with chronic symptoms and the risk of developing radiographic sacroiliitis.5-8 The natural history of non-radiographic axial Spa is unknown. Cases that progress to AS have been observed to do so at a mean duration of 10 years after symptom onset.6, 9 Early diagnosis of non-radiographic axial SpA relies upon early specialist review and MRI. ASAS has developed validated MRI criteria for sacroiliitis highly suggestive of SpA (Table 1). The definition of an ASAS-positive MRI includes bone marrow oedema (BMO) but not structural lesions such as sacroiliac joint (SIJ) erosions or fat deposition. Compared to the ASAS MRI criteria, diagnostic criteria that include structural lesions are either less sensitive, less specific or have not been validated.7, 10, 11 Function in AS, including spinal mobility, correlates with disease activity and syndesmophyte formation.12, 13 Treatment of axial SpA improves symptoms and function, but does not strongly affect bone formation. No therapeutic agent halts bone formation at the SIJ or vertebral corners. Continuous use of nonsteroidal anti-inflammatory drugs (NSAIDs) improves symptoms and may slow syndesmophyte formation to a limited degree.14-16 Tumour necrosis factor (TNF) inhibitors provide significant symptomatic benefit, but have not consistently been demonstrated to slow excessive bone formation. Recent research into axial SpA has improved diagnostic assessment, using MRI and has established MRI and radiography as the main tools for assessing disease activity, response to treatment and prognosis.17 Diagnosis of axial SpA is often dependent on the choice and interpretation of imaging in individuals with axial symptoms. These consensus statements were developed to provide an evidence-based approach to imaging in axial SpA (Table 2). A panel of 13 Australian and New Zealand radiologists and rheumatologists (including one rheumatology trainee) with expertise in axial SpA developed a set of study questions regarding imaging in axial SpA. The intention of the authors is to provide evidence-based principles to guide decision making. The group of authors was compiled by inviting clinicians with an interest in this topic known to ST, NS and MAB. Particular attention was given to clinical reasoning that should precede imaging, initial choice of imaging modality and interpretation of plain radiography and MRI. Eighteen questions were subjected to the systematic literature research phase (Note S1). Literature searches were performed, using the NCBI PubMed database (including Medline), including all articles present on 1 June 2015 published in English (Note S2).18 Available publications included those published in advance online. Titles and abstracts were reviewed for relevance to each clinical question. Each relevant article was reviewed in full text. Possibly relevant articles found by a related search or known to the panel were also reviewed. Results of the literature search were summarised and presented to the panel by ST in seminar form, including a group discussion of the evidence for each statement. Emphasis was placed on the papers most relevant to the clinical question and those with the highest level of evidence. Case reports were not examined. Recommendations relating to the natural history of axial SpA were based on results from published observational cohort studies. Each panel member was given access to the full text of each relevant paper. The consensus statements were developed and reviewed by the panel. Expert opinion was only cited in the absence of relevant data. Further collaboration occurred with an online discussion board and email interactions. The panel was polled on the strength of recommendation for each consensus statement. Each author rated their agreement with each statement in an anonymous online poll. Agreement was rated on a scale of 1–10, where 1 indicated ‘not at all recommended’ and 10 indicated ‘completely recommended’. Due to the high strength of agreement for all statements, the statements did not require revision. The Australian National Health and Medical Research Council (NHMRC) level of evidence system for clinical practice guidelines was applied to each statement (Table 3).19, 20 Under this system, statements formed from expert opinion are referred to as ‘Good Practice Points’. The idea for this paper was conceived by the authors in response to an unmet need for guiding principles on imaging in axial SpA. It was not solicited, organised or convened by any specialty college, working group or corporation. This study was performed completely independently of a separate study in this field and was not influenced by it.21 Recommended use of imaging modalities in axial SpA Individuals with suspected axial SpA should be referred to a rheumatologist. Inflammatory back pain or any other feature of axial SpA (listed by the ASAS criteria for classification of axial SpA) for more than 3 months with onset before age 45 should be investigated by pelvic radiography. If radiography is negative for axial SpA and does not provide a clear alternate diagnosis, sacroiliac MRI should be performed. Diagnosis of axial SpA is an evolving field that presents challenges to experienced clinicians. The clinicians most skilled to diagnose this condition and related spinal conditions are rheumatologists. Suspected axial SpA should be reviewed by a rheumatologist even if diagnostic features of axial SpA are not present. Diagnosis can be made in the absence of sacroiliitis if multiple SpA features and HLA-B27 are present.4, 22 In individuals with symptoms longer than 3 months with onset before the age of 45, the features of inflammatory back pain, extra-articular manifestations of SpA, family history of definite SpA or a good response to NSAIDs are suggestive of axial SpA. In a cohort of Europeans, each feature had a positive predictive value (PPV) for expert-diagnosed axial SpA of 35–45%, while HLA-B27 had a PPV of 67%.23 Limited lumbar flexion and peripheral arthritis were also features that independently increased the likelihood of axial SpA. SpA features less useful for differentiating axial SpA from other causes of chronic back pain include enthesitis, dactylitis, uveitis, psoriasis and inflammatory bowel disease (IBD).2 Features of SpA are not uncommon in the general population, and if accompanied by Inflammatory Back Pain (IBP) should be assessed by a rheumatologist. The estimated prevalence of HLA-B27 in New Zealand Caucasians is similar to that in Britain, at 9.2% and 9.5% respectively.24 The ASAS criteria for classification of axial SpA uses these features and the presence of sacroiliitis on imaging to identify axial SpA (Table 2). These criteria are predictive of chronic sacroiliitis on radiography and MRI.5, 6 Investigation of suspected axial SpA should include MRI if plain radiography is non-diagnostic. An ASAS positive MRI is more sensitive than pelvic radiography for axial SpA with a sensitivity of 48% vs 29% in a large cohort.25 MRI also has the ability to diagnose diseases which cause symptoms similar to axial SpA.26 An Anteroposterior (AP) view of the pelvis is the recommended method of sacroiliac radiography. Equivocal cases should be investigated with MRI of the SIJ. Plain radiography of the pelvis remains the best initial imaging modality due to its wide availability, relatively low cost, and established relationship with disease outcomes. The addition of oblique sacroiliac views to AP views of the pelvis has been examined as a method of improving diagnostic certainty. A case series of AP films from 445 subjects found that addition of oblique sacroiliac views did not reclassify any subject as having unequivocal sacroiliitis or normal SIJ.27 Comparisons of PA and oblique sacroiliac views with CT demonstrate their sensitivity for sacroiliitis to be similar at 32% and 30%, respectively.28 Compared with CT, the sensitivity of PA, angled PA and oblique views combined was only 47%. Multiple radiographic views of the SIJ provide little benefit. If pelvic radiography is equivocal, sacroiliac MRI, which is more sensitive for axial SpA should be performed.25 Pelvic radiography is the recommended first-line imaging test for suspected axial SpA symptomatic for greater than 3 months. Choice of initial imaging modality in early axial SpA is complicated by typically slow radiographic progression and variable degree of symptom severity. Symptom onset may precede definite sacroiliitis on plain radiography by years, and recollection of symptom duration can be poor. One retrospective cohort study found the mean time between symptom onset and definite AS by the mNY criteria to be 71 months.29 Three studies examining individuals with symptom duration less than 2 years found definite sacroiliitis present in 27–38% of subjects.6, 8, 30 These studies examined subjects of various ages, but in each the typical study participant was a male with symptom onset less than 2 years, who had their baseline radiography in their early thirties. Other cohorts of early axial SpA also demonstrate a mean age of onset between 30 and 35 years.29, 31 One of these studies examined subjects with a mean symptom duration of 4 months and found AS by mNY criteria in 27%. The utility of plain radiography in patients under 30 years of age has not been specifically studied in axial SpA. Definite radiographic sacroiliitis is a requirement for government-funded TNF inhibitor therapy in Australia but not New Zealand. These observational findings are consistent, but should be carefully applied to each patient. Individual patient factors such as differential diagnoses, age, comorbidities and features of SpA should be considered before choosing an initial imaging modality. Individuals with diagnosed or suspected non-radiographic axial SpA should have surveillance radiography of the SIJ repeated no more frequently than every 2 years. Spinal radiography at baseline provides prognostic information in individuals with AS, but repeat sacroiliac or spinal radiography has no role in clinical practice for the assessment of AS itself. Non-radiographic axial SpA often progresses to AS. Rates of progression differ between axial SpA cohorts. A German cohort of expert-diagnosed non-radiographic axial SpA reported a 10.5% rate of progression to AS after 2 years.32 An Asian non-radiographic axial SpA cohort observed that progression from onset of non-radiographic axial SpA to AS occurred over a mean of 6 years when pelvic radiography was repeated every 2–3 years.29 In a British non-radiographic axial SpA cohort (imaging arm), out of 24 subjects none progressed to AS after a mean follow-up of 7.7 years.6 An overlapping cohort of subjects with inflammatory back pain (IBP) found that 18% of subjects progressed to AS after 7.7 years.8 This slow rate of progression to AS supports the practice of repeating pelvic radiography no more frequently than every 2 years. The strongest risk factor for syndesmophyte formation is the presence of pre-existing syndesmophytes.33-35 The rate of syndesmophyte formation is poorly understood and varies between individuals.35 Repeat spinal radiography for surveillance of progression has no role in routine practice as syndesmophyte formation or ankylosis cannot be predicted from observed trends. The prognostic value of serial sacroiliac radiography has not been established and is not recommended. Further imaging of the spine or pelvis may be indicated for complications of AS, such as assessment of traumatic injury, or to assess new symptoms in AS patients. As an isolated measure, acute inflammation on spinal MRI demonstrates low correlation with activity of axial SpA. However in patients with a diagnosis of axial SpA acute inflammatory changes on sacroiliac and spinal MRI can be used to identify objective evidence of activity in conjunction with serum C-reactive protein (CRP). In axial SpA, inflammatory markers and BMO are objective findings of spinal and sacroiliac inflammation. Spinal BMO has a low to moderate correlation with CRP in both non-radiographic axial SpA and AS.36-40 Similar studies examining ESR are inconsistent.37, 38 In AS a single study found strong correlation between time-averaged CRP and spinal erosions but not spinal BMO.41 There is conflicting evidence of association between sacroiliac BMO and CRP.7, 37, 40 Symptom-based disease activity scores of axial SpA are not consistently associated with acute inflammation of the spine or SIJ.7, 36-40 The Ankylosing Spondylitis Disease Activity Score, using CRP (ASDAS-CRP) was found to have a low correlation with spinal BMO by one of three studies37, 39, 42 and a low to moderate correlation with sacroiliac BMO by one of two studies.37, 40 Spinal BMO, CRP and disease activity scores each provide a different perspective of disease activity. BMO and CRP provide objective evidence of acute inflammation, and when concordant, are important factors in decision-making. Symptom scores may be influenced by both acute inflammation and chronic structural changes but add an important functional and subjective component to assessment.37 Response to treatment correlates with changes in inflammatory markers and acute inflammation on sacroiliac and spinal MRI. Spinal BMO is a fluctuating measure of disease activity that improves with treatment and is concordant with other measures of disease activity. Three studies have observed a reduction in spinal BMO and CRP (or ASDAS-CRP) after commencement of a TNF inhibitor.39, 43, 44 Change in sacroiliac BMO has shown moderate correlation with change in symptom-based activity or functional scores (BASDAI, BASFI or ASDAS-CRP).43, 44 These studies found no association between sacroiliac BMO and CRP.43, 44 Response to treatment is often reflected by reduced signs of acute inflammation, particularly in the spine. This observation is not universal and validated criteria for treatment response have not been developed. Sacroiliac MRI lesions at high risk of progression from non-radiographic axial SpA to AS include BMO, erosions and fatty lesions. Vertebral corner fatty lesions, especially when associated with BMO, are high risk sites for syndesmophyte formation. Presence of syndesmophytes on spinal radiography strongly predicts future syndesmophyte formation. Observation of axial SpA cohorts has identified risk factors for progression from non-radiographic axial SpA to AS. Presence of moderate or severe BMO, multiple erosions or multiple fatty lesions in the SIJ are predictive of progression to radiographic sacroiliitis.7, 8 Baseline MRI with mild or no sacroiliitis was predictive of no radiographic sacroiliitis at follow-up (mean follow-up 7.7 years), irrespective of HLA-B27 status. The events that precede syndesmophyte formation are poorly understood. Acute inflammation of vertebral corners, termed Corner Inflammatory Lesions (CIL) and post-inflammatory fat deposition, Corner Fatty Lesions (CFL), are associated with syndesmophyte formation.45-47 Sites of partially resolved corner inflammation, especially those with concurrent fat deposition possess the highest risk of syndesmophyte formation with an estimated Odds Ratio of 3.3–7.6.45-47 The frequency of syndesmophyte formation at sites of previous CIL has been reported as high as 43%, compared to 2.4% at vertebral corners where no CIL was observed.48 However, in these observational studies, most syndesmophytes arose from corners with no visible lesion on previous MRI, which leaves the pathogenesis of syndesmophytes poorly defined. For this reason, there is no validated method of predicting syndesmophyte formation based on MRI signs. The main predictor of future syndesmophytes seen on plain radiography is existing syndesmophytes.33-35 Erosions and sclerosis have been associated with syndesmophyte formation by one study of spinal radiography.33 However, these features are uncommon, and in this study most syndesmophytes arose from corners with no prior corner lesion. Computed tomography and bone scintigraphy are not recommended for diagnosis or assessment of axial SpA, but may be useful in identifying other causes of spinal pain. Magnetic resonance imaging is more useful than CT for investigation of suspected axial SpA. MRI is inferior for the detection of structural lesions, such as sacroiliac erosions, with a relative sensitivity of 61%.49 However, the significance of erosions or sclerosis visible on CT or MRI, but not visible on plain radiography has not been established. No validated diagnostic or classification criteria for non-radiographic axial SpA or AS include these features. Validated methods of diagnosing axial SpA by MRI assess BMO but not erosions.6, 22 Sacroiliac BMO is a more useful sign than erosions in non-radiographic axial SpA. Sacroiliac BMO (with or without erosions) is 4–8 fold more common than erosions without BMO.10 The ability to detect BMO makes MRI more useful than CT in the diagnosis of axial SpA. Bone scintigraphy has a poor balance of sensitivity and specificity for the diagnosis of axial SpA. They are more sensitive than plain radiography but less sensitive than MRI.50-52 Unilateral sacroiliitis on bone scintigraphy has high specificity (93%) but poor sensitivity (25%), while bilateral sacroiliitis has poor sensitivity (40%) and specificity (58%) for axial SpA.53 Increased scintigraphic activity outside the SIJ is common in axial SpA but has not been assessed for diagnosis of this condition. Plain radiography and MRI are the best imaging modalities for suspected axial SpA and involve limited radiation exposure. In certain circumstances, including suspected fracture, metabolic bone disease or when MRI is contraindicated, CT and bone scintigraphy may provide useful information. The recommended MRI technique for imaging axial SpA is the combination of Short Tau Inversion Recovery (STIR) and T1 without gadolinium. The recommended planes for MRI are axial and oblique coronal planes for the SIJ and the sagittal plane for the spine. The ideal MRI sequence for diagnosis of axial SpA should provide high sensitivity and specificity, high inter-reader agreement and should possess a suitable cost, availability and safety profile. Imaging findings indicative of axial SpA include structural and acute inflammatory lesions. Commonly used MRI sequences include those sensitive for structural changes, such as T1, and sequences sensitive for acute inflammation including T1 with gadolinium, STIR, and Diffusion Weighted Imaging (DWI). Very few studies directly compare sensitivity and specificity of various MRI sequences for axial SpA. Only one study compares multiple MRI sequences with a validated clinical gold standard in a single cohort.54 This study compared T2 Fat Saturated, STIR, DWI and Dynamic Contrast-enhanced MR (DCE-MR) with the gold standard diagnosis of axial SpA by the ASAS criteria (clinical and imaging arms). STIR was found to have the highest sensitivity, while specificity was 66.6% for all methods other than DCE-MR (100%). Enthesitis of the SIJ, if present, is a characteristic feature of axial SpA. It involves inflammation of the intraarticular or periarticular ligaments, or the joint capsule and is demonstrated by STIR hyperintensity and post-contrast enhancement.4 T1 with contrast is more sensitive for SIJ enthesitis than STIR, which implies it may be more sensitive for axial SpA.55, 56 However, multiple studies have found no improved sensitivity of T1 with contrast compared to STIR.56-59 The largest study assessed 127 subjects with IBP for axial SpA. STIR and T1 with gadolinium displayed complete agreement for the finding of an ASAS positive MRI.56 Sacroiliac enthesitis was always accompanied by significant BMO on STIR, so all patients with enthesitis had a positive STIR MRI. The use of gadolinium is therefore not indicated if there is a high suspicion of axial SpA. No published data are available comparing the utility of different MRI imaging planes in axial SpA. The standard protocol involves imaging the SIJ along oblique coronal (semicoronal) and axial planes with 3–4 mm slices, and imaging of the spine in the sagittal plane.4 Axial plane slices may be used for assessment of posterior spinal elements. Use of validated MRI criteria for axial SpA aids consistency between local and expert readers. Equivocal imaging should be referred to an expert reader. Numerous scoring criteria have been developed to aid diagnosis of axial SpA. Inter-reader reliability studies of various scoring systems have found reliability to be moderate to excellent.6, 60, 61 Few studies have compared global MRI assessment with a criteria based score. The SPARCC MRI score, a criteria based 72 point score of sacroiliac BMO, has been compared with global assessment for a baseline activity score and change in activity. The SPARCC score had higher inter-reader reliability for sensitivity to change than global assessment, but was no different for baseline scores.60 The most comprehensive and clinically informative study of inter-reader reliability examined 582 subjects with IBP, using T1 and STIR.62 The ASAS MRI criteria were applied by a local radiologist or rheumatologist and two calibrated expert readers. Agreement between the local and expert readers was good (Cohen's kappa 0.7), with no difference in agreement between local radiologists and local rheumatologists. The expert readers classified 12% of local readings as false negatives and 17% as false positives. After incorporating plain radiographs and the ASAS criteria for classification of axial SpA, only 8% of cases were reclassified by expert review. This finding reflects the utility of the ASAS criteria, which are more reproducible and more robust to minor variation than global assessment of MRI for the diagnosis of axial SpA. For the diagnosis of axial SpA, spinal MRI does not add value to sacroiliac MRI. Spinal MRI is often useful to diagnose other causes of back pain, especially when sacroiliac MRI is negative for axial SpA. The sensitivity of sacroiliac MRI for diagnosis of axial SpA could be improved by adding spinal MRI. In a cohort of symptomatically active non-radiographic SpA, vertebral BMO characteristic of axial SpA was present without significant sacroiliac BMO in 49%.63 Spinal MRI for diagnosis of axial SpA has not been studied in as much detail as sacroiliac MRI, but characteristic acute and chronic features of axial SpA have been identified.64 Vertebral corner lesions, including CIL, CFL and erosions are common in axial SpA. However, these lesions may also be seen in degenerative spinal disease, where presence of disc degeneration provides a useful differentiating feature. Vertebral corner lesions are more frequent in axial SpA compared to non-specific back pain and healthy controls, however, attempts to create diagnostic criteria for axial SpA based on spinal MRI signs alone have not successfully created criteria with high sensitivity and specificity.65, 66 The sensitivity of sacroiliac MRI for axial SpA improves when assessed with spinal MRI, but the false-positive rate dramatically increases. One study found that diagnosis of axial SpA by global assessment of spinal MRI had false-positive rates of 25% in healthy controls and 24–32% in subjects with non-specific back pain.65 This was attributed to the presence of CIL and CFL in subjects without axial SpA, who presumably developed these lesions as a result of degenerative disease. Consequently, at this point in time adding spinal sequences to sacroiliac MRI does not aid diagnosis.10, 65 Spinal inflammation is a prominent imaging feature of axial SpA, however diagnosis by MRI must be based on sacroiliac imaging findings until validated spinal diagnostic criteria exist. The imaging arm of the ASAS criteria for classification of axial SpA is a validated method of identifying individuals with axial SpA. Individuals who have a positive MRI for axial SpA (as per the ASAS criteria for classification of axial SpA) have a high risk of recurrent sacroiliitis and progression to AS. The imaging and HLA-B27 arms of the ASAS criteria for classification of axial SpA provide a practical approach to identify axial SpA. When applied together, they also demonstrate moderate sensitivity for axial SpA (estimated at 68%) which is superior to other diagnostic criteria.67 The inclusion of MRI findings in the diagnostic criteria increases sensitivity and add objective evidence to the diagnosis of axial SpA.6 An ASAS positive MRI is a valuable predictor of chronic sacroiliitis. A Chinese observational study followed 668 subjects with axial SpA for 2 years.5 Almost all subjects (95.5%) with axial SpA by the ASAS imaging arm still met the criteria for axial SpA at follow-up. A European study performed sacroiliac MRI on IBP patients at baseline, after 1 year and after 2 years.68-70 Subjects with an ASAS negative MRI at baseline who were HLA-B27 negative had <5% likelihood of a positive MRI during follow-up. HLA-B27 positive individuals had a likelihood of 27%. Significant sacroiliitis on MRI is also strongly predictive of progression to AS. In a separate cohort followed up for 8 years, moderate or severe sacroiliitis had a PPV for progression to AS of 80% in HLA-B27 positive axial SpA cases, and 50% in HLA-B27 negative cases.8 This finding reinforces the predictive value of the two arms of the ASAS criteria, which require either sacroiliitis or HLA-B27. Diagnostic criteria for axial SpA should involve a criteria-based definition of sacroiliitis and clinical criteria. Each of these features adds sensitivity. In one IBP cohort, diagnosis using an ASAS positive MRI had a superior sensitivity for axial SpA than global assessment (79% vs 66%).6 The sensitivity increased further to 83% using the imaging arm of the ASAS criteria, which requires one clinical feature of SpA. MRI lesions suggestive of axial SpA that contribute to diagnosis include BMO, erosions, fatty bone marrow lesions and subchondral sclerosis. High numbers of Corner Inflammatory Lesions are specific but not sensitive for axial SpA. Features of axial SpA on MRI include acute inflammatory and chronic (structural) lesions.64 Despite the presence of many structural lesions in axial SpA, the only validated MRI criteria for axial SpA examine only sacroiliac BMO.71 The acute and chronic sacroiliac lesions of axial SpA can be present in non-specific back pain and in healthy controls.10, 66 Diagnosis is based on the presence of multiple BMO lesions in the SIJ, which has a moderate sensitivity and specificity for axial SpA. The presence of multiple erosions, multiple fatty marrow lesions, or combinations of fatty marrow lesions, erosions and BMO in the SIJ are specific, but not sensitive.11 The number of lesions also provides prognostic information as multiple BMO lesions, multiple erosions or multiple fatty marrow lesions confers risk of progression to AS.7 Data on the significance of other features of axial SpA seen on MRI such as sclerosis and joint space alterations are limited.66 Sacroiliac enthesitis is a feature of axial SpA, but is believed to occur only in the presence of significant sacroiliac BMO.56 Spinal features of axial SpA include acute inflammation across entire vertebrae, at vertebral corners, vertebral discs and zygoapophyseal joints and at the entheses of spinal ligaments. Structural lesions of the spine may include fat deposition in bone marrow, cortical erosions, sclerosis, syndesmophytes and vertebral ankylosis.64 Structural lesions occur in 98% of patients with AS, but also in 26–28% of patients with non-specific back pain and healthy controls.72 Structural abnormalities were omitted from the definition of an ‘ASAS positive MRI’ because the clinical significance of lesions visible on MRI, but not plain radiography has not been established.71 Diagnostic criteria requiring multiple CIL or CFL are specific for axial SpA, but lack sensitivity.64, 66 Candidate diagnostic criteria based on multiple CIL and/or CFL are specific but not sensitive and are not clinically useful.10, 66 Multiple CIL are believed to be more clinically useful than multiple CFL, but are yet to be validated as a sign of axial SpA. ‘Non-radiographic axial spondyloarthritis’ is an appropriate term to use when communicating with General Practitioners (GPs) and patients. ‘Ankylosing spondylitis’ is the preferred term to ‘radiographic axial spondyloarthritis’. ‘Axial spondyloarthritis’ is a new term which GPs are unlikely to be familiar with. A diagnosis of non-radiographic axial SpA provides a management pathway and prognostic information for individuals who do not meet the mNY criteria for AS. As axial SpA is usually diagnosed before the age of forty, avoidance of this classification may delay treatment and increase morbidity. Information on axial SpA is readily available, and is sought online by 77% of individuals with AS.73 An Internet search for ‘axial spondyloarthritis’ returns patient information from reputable sources, including the Spondylitis Association of America and the National AS Society (UK).74 It can be inferred that in Australia and New Zealand, GPs and most newly diagnosed patients (who by definition are younger than 45) will access high-quality patient information online. It is therefore reasonable to use the term ‘axial spondyloarthritis’ in correspondence, imaging reports and when consulting with patients. The term ‘Ankylosing Spondylitis’ is preferable to ‘radiographic axial spondyloarthritis’ as it is already in widespread use. These consensus statements provide evidence-based guidance within a challenging area of diagnostic medicine. The evidence is sufficient to guide practice in most situations. Some clinical questions remain difficult to answer due to disease heterogeneity, the transience of acute inflammatory signs and the use of expert opinion as a gold standard. Recent advances in imaging and classification of axial SpA enable early diagnosis and intervention. Radiography remains the first-line imaging modality, and leads to a diagnosis of AS in one-third of new cases. Sacroiliac MRI is a useful adjunct to radiography in suspected axial SpA. It is more sensitive than radiography and can identify numerous causes of spinal pain and can assess disease activity and response to treatment. An ASAS positive MRI, when interpreted with clinical features and HLA-B27 can also be used to predict chronic sacroiliitis and development of AS. It is however, slightly limited by a moderate sensitivity and does not assess vertebrae or signs of chronic inflammation. The imaging arm of the ASAS criteria for axial SpA is a powerful tool for assessment of diagnosis and prognosis. It should be used as the primary tool for diagnostic assessment, to permit rapid diagnosis and intervention. We thank Dr Murray Hargrave for assisting with the literature search. ST and MH were supported by an educational grant from AbbVie, UCB, Janssen and Pfizer. No other author received funding from any other source (government or corporate) related to this paper. The educational grant was donated to the University of Queensland Diamantina Institute then forwarded to ST and MH. ST and MH had no communication whatsoever with AbbVie, UCB, Janssen or Pfizer regarding this work. NS, PB, IL, LS, S Stebbings, AT, SW, JZ and MAB have participated in Abbvie advisory board meetings. NS, LS, SW, PB and MAB have participated in advisory boards for Pfizer. SW, PB and MAB have participated in advisory board meetings for Janssen. PB has participated in advisory board meetings for Celgene. SW and MAB have participated in advisory boards for UCB. SW has participated in advisory boards for Menarini. PR has consulted for and given talks for Pfizer, UCB, AbbVie and Janssen. JZ has consulted for and given talks for Pfizer, UCB, AbbvVie, Celgene and Janssen. JZ has conducted clinical trials for Pfizer, UCB, AbbVie, Janssen, BMS, Roche, Daiichi-Sanyo, Sanofi-Aventis, Novartis, Celgene and Carbylan. MAB has received research funding from Janssen, AbbVie, UCB, Leo Pharma and Complete Genomics and has been an invited speaker for Abbvie, UCB, Pfizer and UCB. Table S1. NHMRC Evidence Hierarchy for diagnostic accuracy and prognostic studies (Source: National Health and Medical Research Council1. Note S1. Research questions. Note S2. PubMed Search terms. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.