BACKGROUND AND INTRODUCTION
The ASAS group has developed classification criteria for axial spondyloarthritis (axSpA) that allow the inclusion of patients with an early form of disease that is not yet clearly visible on plain radiography1. These criteria include both an imaging ‘arm’, that incorporates either radiographic sacroiliitis or magnetic resonance imaging (MRI) features of active inflammation in the sacroiliac joints (SIJ) plus one clinical feature of spondyloarthritis (SpA), and a clinical ‘arm’ that requires the presence of HLA B27 plus 2 clinical features of SpA. The sensitivity and specificity of these criteria for axSpA were 83% and 84%, respectively, when assessed in a prospective cohort of patients referred to a rheumatologist with undiagnosed chronic back pain and suspicion of SpA.
The gold standard for evaluation of these criteria was the rheumatologist expert opinion of the SpA diagnosis after incorporating lab and imaging data with clinical evaluation. Part of this cohort (40%) was used to test candidate criteria that had been selected after a paper scoring exercise as well as existing classification criteria and the remainder (60%) was used for validation. ASAS has also proposed a consensus based definition of a positive MRI for classification of axSpA based on the finding of increased signal on fat-saturated MRI sequences, such as short tau inversion recovery (STIR), in subchondral bone that is “highly suggestive” of SpA2. This is indicative of bone marrow edema (BME) associated with inflammation and a positive MRI is defined according to the presence of two lesions on a single coronal slice or a single lesion on two consecutive slices.
Since their introduction, these classification criteria have come under scrutiny because their implementation in population studies of patients with chronic low back pain has resulted in substantial differences in the prevalence estimates of axSpA3,4. Genetic studies have failed to demonstrate similar disease associations with risk loci comparable to studies that have included patients according to the modified New York criteria (mNY)5. Additional cohort studies have demonstrated an excess of females, low prevalence of Human Leukocyte Antigen B27 (HLA B27), and lower response rates to tumor necrosis factor inhibitor (TNFi) therapy raising concerns regarding misclassification6-9.
A number of limitations beset the imaging component of the original ASAS study, including the fact that not all participants in the cohort studied underwent MRI assessment. The imaging arm has been further scrutinized because evaluation of patients with non- specific back pain has demonstrated a positive MRI in 20%10, 11. In addition, the reliability of detection of radiographic sacroiliitis, especially for grade 2 disease, is poor12,13. Yet the imaging arm requires only one SpA feature in addition to positive imaging to classify a patient as having axSpA. Lastly, MRI, and especially the STIR sequence, has been incorporated into the diagnostic decision making process in studies that have aimed to test the performance of the classification criteria. This has lead to circularity and the potential for overestimation of the role of MRI in classification14.
The clinical arm has been scrutinized because some clinical features e.g. inflammatory back pain and response to non-steroidal anti-inflammatory (NSAID) agents may be readily observed in non-specific causes of back pain. A five year follow up of the ASAS classification cohort has reported high positive predictive value of the criteria for axSpA although direct follow up of patients in the clinic was limited to only 35% of patients15. It has been suggested that further education regarding appropriate ascertainment of SpA features and evaluation of imaging might help reduce misclassification. In particular, there appears to be a lack of awareness and/or acceptance of these criteria in the United States (US). Expertise in MRI interpretation varies widely even amongst musculoskeletal radiologists. Conversely, the criteria have also been inappropriately used as a basis for diagnostic evaluation of SpA. The FDA has expressed concern that such inappropriate application of the criteria and/or their perceived lack of specificity might lead to the inappropriate institution of expensive therapy in patients with conditions resembling SpA16. This has been cited as a basis for its failure to grant a treatment indication for non- radiographic axSpA (nr-axSpA) .
A joint meeting of the ASAS and the SpA Research and Treatment Network (SPARTAN) executive boards has recommended that the ASAS classification criteria undergo further validation in a prospective North American cohort of patients presenting with undiagnosed chronic back pain to rheumatologists in the US and Canada. It was also considered desirable to include patients from other parts of the world although feasibility concerns were acknowledged. A consensus of board members determined that it would be appropriate to aim for classification criteria that have a specificity of at least 90% for axSpA in view of the high background prevalence of non-specific back pain. Sensitivity of at least 75% was also considered appropriate. A major differentiating feature of the proposed study design compared to prior cohort studies would be the institution of a process of central image review that would aim to provide radiologic diagnostic input directly to the investigator rheumatologist in later formulating the gold standard expert opinion. This would serve to enhance confidence in the gold standard determination of the diagnosis of axSpA. This review process would also serve to educate both rheumatologists and radiologists in those aspects of image evaluation that are considered relevant to the diagnosis of axSpA.
With effective medication now available for several years, and effectiveness being particularly apparent in early disease, reducing diagnostic delay is a priority. In particular, it has been shown that patients with early nr-axSpA have comparable severity of symptoms to those with established disease17. But many of these patients do not have an opportunity to receive effective treatment because access to treatment is based on regulatory approval, which is currently not available in the US for non-radiographic disease. There is also emerging evidence that early diagnosis may be an essential prerequisite for effective disease modification. Consequently, there is impetus for further education in the appropriate use of imaging as well as the correct application of the criteria so that sufficient confidence in the performance of the criteria emerges to allow regulatory acceptance of the nr-axSpA indication and thereby early institution of treatment in countries where this is not currently possible.
STUDY OBJECTIVES
Primary
To test (“validate”) the performance of the current ASAS classification criteria in a prospective cohort of patients presenting to a rheumatologist in North America with undiagnosed back pain of ≥3 months duration with onset ≤45 years of age.
Secondary
- To test candidate enhancements to the imaging arm of the current criteria based on more stringent requirements for conducting MRI assessment and definition(s) for a positive MRI.
- To determine which clinical parameters best discriminate between patients diagnosed clinically with axSpA versus non-specific back pain and categorize them into major and minor SpA features.
- To test candidate enhancements to the clinical arm of the current criteria incorporating a new framework of major and minor clinical features of SpA.
- To compare the ascertainment of sacroiliitis by MRI and the confidence of this ascertainment by investigators at sites across North America before and after interactive educational intervention. The educational intervention will consist of formal reporting and extensive DICOM-image annotation provided by musculoskeletal radiologists and rheumatologists with extensive expertise (>10years experience) in the assessment of axSpA (central readers) and local investigators.
INVESTIGATION PLAN OVERVIEW
The overall study design will follow a similar format to that used to develop the ASAS axial SpA classification criteria. All consecutive patients referred to a rheumatologist with undiagnosed back pain of ≥3 months duration with onset ≤45 years of age will comprise the prospective cohort. The sample size of 500 is aimed at ensuring that a sufficient number of patients will have axSpA and also permit the option to conduct 5-year follow up to determine predictive validity of the classification criteria. Details of the clinical assessment, including history and physical examination, will be recorded by the rheumatologist in the electronic case report form (eCRF). This will be accessed through an online portal.
The rheumatologist will complete 5 global assessments to determine the presence/absence of axSpA. The first will be completed immediately after the first clinical assessment at the end of the patient encounter and will incorporate details of the history and physical exam. This is aimed at ascertaining which clinical features are considered most important in formulating the rheumatologist’s opinion regarding the diagnosis of axSpA. The second will be conducted once the C-Reactive Protein (CRP), and HLA-B27 data are made available. The third will be completed after the pelvic radiograph has been reviewed by the rheumatologist. The fourth will be completed after the rheumatologist has reviewed the report of the pelvic MRI scan provided by the local radiologist. The fifth will be completed after central review of the anonymized radiograph and pelvic MRI scan and feedback to both the rheumatologist and radiologist. Patients will then be followed according to the discretion of the rheumatologist and appropriate standards of clinical practice. Patient contact information will be obtained and patient consent obtained through the informed consent to allow contact with the patient after 5 years of follow up.
