Article Text


The term ‘non-radiographic axial spondyloarthritis’ is much more important to classify than to diagnose patients with axial spondyloarthritis
  1. Atul Deodhar1,
  2. Vibeke Strand2,
  3. Jonathan Kay3,
  4. Juergen Braun4
  1. 1Division of Arthritis & Rheumatic Diseases, Oregon Health & Science University, Portland, Oregon, USA
  2. 2Division of Immunology/Rheumatology, Stanford University School of Medicine, Palo Alto, California, USA
  3. 3Division of Rheumatology, Department of Medicine, UMass Memorial Medical Center and University of Massachusetts Medical School, Worcester, Massachusetts, USA
  4. 4Rheumazentrum Ruhrgebiet, Ruhr-University Bochum, Herne, Germany
  1. Correspondence to Professor Dr J Braun, Rheumazentrum Ruhrgebiet, Ruhr-University Bochum, Claudiusstr. 45, Herne 44649, Germany; j.braun{at}


The term axial spondyloarthritis (axSpA) now is used frequently to describe patients with predominantly axial symptoms who fit into the spectrum of a well-recognised rheumatic disease that continues to be known as ankylosing spondylitis (AS). The 2009 Assessment of SpondyloArthritis international Society (ASAS) classification criteria, developed to identify patients with early or atypical disease which could not be classified by the 1984 modified New York (mNY) criteria for AS, have led to a differentiation between non-radiographic axial spondyloarthritis (nr-axSpA) and radiographic axSpA, which is largely synonymous with AS. The main reason to distinguish between these ends of the spectrum of axSpA was that tumor necrosis factor (TNF) inhibitors (TNFi) approved for AS could obtain additional labelling for nr-axSpA and be used to treat all patients manifesting clinical features of axSpA. These two terms are distinguished by the degree of ‘radiographic sacroiliitis’ assessed by conventional radiography, according to the 1984 mNY criteria for AS. Since this differentiation has been shown to be not very reliable, we argue that the terms nr-axSpA and AS should only be used for classification of patients with axSpA and not as separate diagnoses. Therefore, we propose that only the term axSpA be used to diagnose patients, unless there is a meaningful medical reason to differentiate nr-axSpA from AS. The available data justify performing randomised controlled trials designed to obtain regulatory approval for therapeutic agents in patients across the entire spectrum of axSpA.

Statistics from

Approximately 2–3 million individuals in the USA suffer from axial spondyloarthritis (axSpA) according to National Health and Nutrition Examination Survey (NHANES) data.1 The Assessment of SpondyloArthritis international Society (ASAS) classification criteria for axSpA2 ,3 were developed because the modified New York (mNY) criteria for ankylosing spondylitis (AS4) and the European Spondylarthropathy Study Group (ESSG) criteria5 were not sufficiently sensitive to identify patients with early and/or ‘atypical’ disease. The development of new classification criteria was necessary to broaden the indications for treatment, specifically with tumor necrosis factor (TNF) inhibitors (TNFi), which had been demonstrated to be efficacious in AS.6–9 Furthermore, MRI had been demonstrated convincingly to be able to detect sacroiliitis and spondylitis,10 ,11 and definitions for a ‘positive MRI’ in axSpA have been proposed.12 ,13 Using the ASAS classification criteria, axSpA has been divided into ‘radiographic’ (AS) and ‘non-radiographic’ axial spondyloarthritis (nr-axSpA) forms, depending on whether or not definitive structural changes are evident in the sacroiliac joints (SIJ) on plain radiographs, recognising that SIJ involvement is considered to be a characteristic feature of axSpA. The implications of this have been reviewed recently.14

Since structural changes of the SIJ are present in almost all patients with AS,4 many rheumatologists use this radiographic manifestation as a necessary criterion for diagnosis, despite published evidence that AS may manifest less commonly with structural changes only in the spine.15 However, the presence of syndesmophytes, although helpful when diagnosing individual cases of AS,15 has never been considered suitable as a classification criterion because the differentiation between syndesmophytes and degenerative osteophytes may be reportedly difficult.16 At present, CT is the best imaging technique to detect structural changes in the SIJ,17 but its widespread use has been limited because of the higher radiation dose required, which is a concern when evaluating young adults and women of childbearing age. However, this may change in future with the availability of better low-dose CT techniques.18

The distinction between AS and nr-axSpA has recently gained attention because TNFi approved for AS are effective in treating symptoms of axSpA in patients without radiographic evidence of sacroiliitis9 and now have sought regulatory approval to treat patients with nr-axSpA.8–12 However, the validity of the X-ray changes required to fulfil mNY criteria has recently been challenged because grading of structural changes in the SIJ is poorly reproducible and inconsistent among rheumatologists and radiologists when neither erosions nor ankylosis is evident. Importantly, training the readers did not improve reproducibility.19 Over the past 2 years, there has been much discussion regarding whether central reading is better than local interpretation of SIJ radiographs. Indeed, these methodological challenges have recently been well described.17 ,18 Although central reading may be more reliable, interpretation of radiographs at local sites, closer to where patients are cared for in daily practice, may be better. Local readers can take into account the clinical characteristics of individual patients when interpreting their imaging studies, especially the potentially important factors of age14 ,15 or of confounding diagnoses such as erosive osteochondrosis (Modic lesions) at the L4/L5 level. Disagreement in interpreting SIJ imaging, both between local rheumatologists or radiologists and central trained readers and between different central trained readers, has been reported to occur in 20–30% of cases.20 ,21 The experience in the first randomised controlled trials (RCTs) to study nr-axSpA10 was that the trained central readers revised the interpretation of SIJ radiographs by local readers in 185 (40%) of cases—an unfortunate, but typical, example.

In addition, the term nr-axSpA is confusing to many, since it allows patients with axSpA to be classified as ‘non-radiographic’ in the presence of syndesmophytes, as long as the SIJ radiographs reveal no definite changes. Typically, syndesmophytes occur in later stages of disease, whereas structural changes of the SIJ occur earlier. A patient with unilateral grade 2 sacroiliitis (with erosions and sclerosis) is also labelled as ‘non-radiographic’, because these changes do not fulfil the definition of ‘definite’ sacroiliitis according to the mNY criteria. Patients without definite radiographic changes in the SIJ may have spinal inflammation evident on MRI22—at present, these patients are not classified as having nr-axSpA, since MRI findings restricted to the spine,23 ,24 which are often due to Modic lesions that are degenerative in nature,25 ,26 create substantial ‘background noise.’ These issues require further study.

Although the rates of response to TNFi in AS and nr-axSpA are almost identical,8 no TNFi has yet received regulatory approval for treatment of nr-axSpA in the USA. While rejecting approval of two TNFi for treatment of nr-axSpA, the US Food and Drug Administration (FDA) expressed doubts about the specificity of the ASAS criteria and the natural history of nr-axSpA, hinting that these patients may remit spontaneously and thus would not require treatment with TNFi. However, in 2012, the European Medicines Agency (EMA) decided differently and recommended that adalimumab be approved for treatment of nr-axSpA, in addition to AS.10 Subsequently, certolizumab pegol,8 etanercept11 and golimumab12 were approved in the European Union (EU) for treatment of patients classified as nr-axSpA. In contrast to AS, the use of TNFi in the EU to treat nr-axSpA is restricted to patients with evidence of osteitis (bone marrow oedema) on MRI or elevated C-reactive protein (CRP).27

There are other differences between how FDA and EMA view future approvals of products for an indication for treatment of nr-axSpA. FDA has recently requested 52-week placebo-controlled trials, that is, ‘standard-of-care controlled’ trials, for products seeking this indication. Their rationale is that such trials will provide information about the natural history of the disease, particularly the proportion of patients with nr-axSpA who may remit spontaneously over the course of a year. In contrast, EMA will not allow a 52-week placebo-controlled trial, since patients who have already failed nonsteroidal anti-inflammatory drugs (NSAID) therapy would be randomised to continue receiving only NSAIDs instead of an active agent. In contrast, to investigate whether patients with nr-axSpA can achieve a state of drug-free remission, EMA has requested a ‘withdrawal’ trial design, in which patients with nr-axSpA in remission for a predefined duration of time are randomised to receive either placebo or active study agent. So far, neither agency has requested an RCT that would enrol patients both with nr-axSpA and with AS, classified together as ‘axial spondyloarthritis’. However, the main reason why this was handled differently in the past is historical—all TNFi were first approved for active AS only.

Are FDA's concerns justified? Longitudinal observation of patients with undifferentiated spondyloarthritis (some of whom likely would be classified as nr-axSpA) has demonstrated that only 36–59% progress to AS, even after 10 or more years of follow-up.28 ,29 Patients with nr-axSpA who do not progress to AS do not necessarily remit—they just do not develop radiographic changes despite persistence of inflammatory back pain. Data from Spondyloarthropathy Epidemiology and Burden (SPEED)-2, a study conducted in the USA, showed that an even smaller proportion of patients diagnosed with nr-axSpA 120 (28%) progressed to AS over up to 11 years.30 Comparing 120 patients with nr-axSpA with 166 patients with AS in this study, few in either group remitted spontaneously (9.2% vs 7.8%) and fewer with nr-axSpA remitted with treatment (28% vs 42%). In both groups, 189 (66%) patients reported at least one flare, with mean durations of 18 and 23 days.30 Of 326 patients with nr-axSpA in the Devenir des Spondylarthropathies Indifférenciées Récentes (DESIR) cohort, of whom 192 (59%) were women with mean age 34 years and mean disease duration 8 months at baseline, only 16 (5%) progressed to fulfil mNY criteria at 2 years.31 In multivariate analyses, predictors of radiographic progression were smoking, the presence of HLA-B27 and inflammatory changes on MRI. In the most recent paper on this topic,32 the clinical disease course over 2 years was similar among 145 patients with nr-axSpA and 158 patients with AS who were not treated with TNFi. Importantly, the proportion of patients with Bath ankylosing spondylitis disease activity (BASDAI) scores ≥4 and an elevated CRP at baseline, but who achieved low disease activity at two or more time points during the 2-year follow-up period did not differ significantly between patients with nr-axSpA and patients with AS. Spontaneous remission occurred infrequently and the proportions of patients who developed ankylosing spondylitis disease activity score (ASDAS) inactive disease also did not differ significantly between the two groups.32 These data clearly argue against major differences in the course of the first years of the disease.

Patients with nr-axSpA and AS have been shown to respond similarly to TNFi treatment. A large trial of certolizumab pegol that enrolled both patients with nr-axSpA and patients with AS demonstrated similar rates of response to treatment in both groups.8 Although this was the only clinical trial to include both patients with nr-axSpA and patients with AS in the same study, major differences between the two groups have not been reported until now, with the exception of gender and degree of inflammation: more women have nr-axSpA and patients with nr-axSpA have significantly lower CRP levels.33 However, this may simply be explained by the fact that men with AS develop more radiographic damage and that women with nr-axSpA have lower degrees of inflammation.34 Of note, in several other rheumatic diseases, patients classified under a single diagnosis may exhibit prognostic variability among subgroups, such as patients with rheumatoid arthritis with or without rheumatoid factor or anti-citrullinated protein antibodies and patients with systemic lupus erythematosus with or without nephritis.

The specificity of the ASAS classification criteria for axSpA could potentially be improved by adding ‘exclusion of other diagnoses’ in the stem of the criteria, as was done in the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) rheumatoid arthritis (RA) classification criteria.35 For example, the presence of certain conditions considered in the differential diagnosis of axSpA, such as degenerative disc disease, infectious sacroiliitis and spondylitis, lymphoma and fractures,36 ,37 could be excluded because each may cause bone marrow oedema in the axial skeleton. More relative weight could be assigned to objective extra-articular features of spondyloarthritis such as acute anterior uveitis, inflammatory bowel disease or psoriasis. However, the advantage of this proposal needs to be proven.

Diagnostic criteria consist of a combination of signs, symptoms and test results with both high sensitivity and high specificity, which establish a diagnosis to guide care of the individual patient. Classification criteria, on the other hand, are developed to ensure homogeneity of groups of patients enrolled in research studies and thus must have very high specificity, which may result in a loss of sensitivity.38 Even though the rheumatologist may refer to the same SpA features described in the ASAS classification criteria such as inflammatory back pain,39 these criteria should not be used to establish the diagnosis.40 A diagnosis of axSpA can be justified on clinical grounds, even when the classification criteria are not fulfilled.40

Does it matter to differentiate between nr-axSpA and AS in daily clinical practice? Neither is there a need nor is it appropriate to make an artificial distinction between nr-axSpA and AS, as both are part of the spectrum of axSpA. It is much more important to distinguish between early and established disease based on the duration of symptoms rather than on the absence or presence of structural changes in the SIJ. After 2 years of inflammatory back pain, about 20% of patients with axSpA already have radiographic changes in the SIJ.21 ,41 Thus, if trials are designed to assess a therapeutic intervention early in the course of the disease, enrolling patients with nr-axSpA will exclude 20–30% who have definitive sacroiliitis in the SIJ—specifically those patients with more aggressive disease. If the aim of a clinical trial is to treat patients with axSpA early in the course of their disease, the duration of symptoms should be a more important inclusion criterion than the presence or absence of definite sacroiliitis.

Epidemiological studies of axSpA have revealed that rheumatologists in the USA diagnose what could be classified as nr-axSpA as AS, even if strict mNY criteria are not met. In the SPEED-1 study of patients in US rheumatology practices with chronic back pain, 514 met ASAS criteria for axSpA (63%), 390 (76%) of whom were diagnosed as having axSpA.42 Another multicentre study, the PRevalence Of axial SpA (PROSpA), assessed patients with chronic low back pain of at least 3 months duration, with onset before 45 years of age and at least one of the following: the presence of HLA-B27, current inflammatory back pain and MRI and/or radiographic evidence of sacroiliitis.43 Of the 238 patients independently classified with nr-axSpA by ASAS criteria, 164 (69%) were diagnosed as having axSpA by their rheumatologists. These patients had characteristics similar to those of the 108 other patients with AS (who met mNY criteria). Of interest, in the absence of an International Statistical Classification of Diseases and Related Health Problems (ICD)-9 or ICD-10 code for nr-axSpA, US rheumatologists may have used the ICD-9 code 720.0 (AS) or now use the ICD-10 codes M45.0–M45.9 so that the patient's insurance carrier would cover the cost of a TNFi.

Taken together, these data indicate there is no need to differentiate between a diagnosis of nr-axSpA and that of AS in clinical practice, since the only purpose for having these two labels is classification. Established recommendations for management of axSpA have already widened the disease spectrum considerably.44 Nevertheless, there are exceptions when a formal differentiation between nr-axSpA and AS may be necessary, and this may be important in daily clinical practice: (i) to specify an approved indication for TNFi therapy, when off-label use of biologics must be avoided (eg, infliximab is approved for AS but not for nr-axSpA in Europe) and (ii) to clarify the presence of structural changes that are required for patients to receive coverage from their insurance carrier to use a TNFi. Thus, nr-axSpA and AS represent the ends of the spectrum of axSpA, which should be considered as a single disease process. Patients with nr-axSpA more often are women and have a shorter duration of disease than patients with AS. However, despite the absence of radiographic evidence of sacroiliitis, patients with nr-axSpA have a substantial burden of illness, with self-reported disease activity and functional impairments comparable with those of patients with structural changes consistent with AS.45 Accordingly, we propose that RCTs designed to obtain regulatory approval for therapeutic agents be conducted in patients across the entire spectrum of axSpA.


View Abstract


  • Handling editor Tore K Kvien

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.