AB0718 DIAGNOSTIC PERFORMANCE OF VARIOUS CLASSIFICATION CRITERIA IN LOW BACK PAIN PATIENTS WITH SUSPECTED AXIAL SPONDYLOARTHRITIS: A PRAGMATIC STUDY.

Abstract

Background:In routine practice, axial spondyloarthritis (SpA) can be a diagnostic challenge because there is potential overlap with osteoarthritic low back pain or diffuse polyalgic syndrome with axial pain as fibromyalgia. Internationally recognized classification criteria (AMOR, ASAS (1), ESSG (2)) are often used as diagnostic tools in clinical practice (3), but few studies have sought to establish their “real life” diagnostic performances.Objectives:This monocentric study aimed to evaluate the diagnostic performance of different sets of classification criteria for SpA in patients with chronic back pain hospitalized for suspected axial SpA. The second objective was to evaluate the impact of the sacroiliac MRI reading according to the ASAS MRI reading or expert musculoskeletal reading.Methods:Patients presenting with inflammatory low back pain who underwent standardized sacroiliac MRI protocol were consecutively included. The diagnoses obtained with the classification criteria (AMOR, ESSG, Modified AMOR and Modified ESSG (with sacroiliac MRI)) were compared to the gold standard diagnoses (made by a college of three experienced rheumatologists). Two readings of the sacroiliac MRI were performed (ASAS MRI reading and Expert MRI reading (including all inflammatory or structural T1 or T2 STIR abnormalities)). Diagnostic performance was measured for each set of classification criteria: sensitivity, specificity, predictive positive and negative values (PPV, NPV) and positive and negative likelihood ratios (PLR, PLN). The clinical, biological and MRI factors associated with axial SpA diagnosis were identified in a multivariate logistic regression model.Results:83 patients were included of which 27 had axial SpA according to the opinion of the college of rheumatologists. The AMOR modified criteria (with ASAS MRI reading) had the highest diagnostic performance values, with sensitivity = 96%, specificity = 71%, NPV = 97%, PPV = 54%, PLR = 3.3, and NLR = 0.06. Sacroiliac MRI reading demonstrated poor sensitivity (< 60%) for SpA (ASAS or expert reading), while expert reading of MRI had high specificity (98%), NPV = 90%, PPV = 78%, and PLR = 18. In multivariate logistic regression, the presence of enthesitis (OR = 3.51, IC95% [1.10;11.21]; p = 0.024), positive HLA B27 (OR = 4.22, IC95% [1.25;14.18]; p = 0.02)) or sacroiliitis on MRI (ASAS reading) (OR = 3.34, IC95% [1.04; 10.76]; p = 0.043) were independently associated with the final diagnosis of SpA.Conclusion:The results of this pragmatic study suggest that the Modified AMOR criteria with ASAS MRI reading can be used to rule out axial spondyloarthritis (NPV = 97%). However, the use of an MRI ASAS reading alone creates a risk of false positives because of the greater impact on ASAS criteria compared to AMOR or ESSG criteria. Further international studies are needed to decrease the rate of false positives in suspected cases of axial spondyloarthritis in routine practice.