Added value of biomarkers compared to routine clinical parameters for the prediction of radiographic spinal progression in axial spondyloarthritis

Background: Structural damage in the spine determines the functional status and spinal mobility in axial spondyloarthritis (axSpA). Already present syndesmophytes, elevated C-reactive protein, cigarette smoking, and to a lesser extent male sex are routine clinical parameters predicting radiographic spinal progression. In the last years, several biomarkers with a predictive value for radiographic spinal progression were identified. It is, however, not known, if biomarkers have a meaningful added value over clinical parameters in prediction of radiographic spinal progression in axSpA. Objectives: The objective of the study was to examine whether adding biomarkers to the routine clinical parameters would improve prediction of radiographic spinal progression in axSpA. Methods: Altogether 117 patients with ankylosing spondylitis who completed a 2 year clinical and radiographic follow-up in the ENRADAS trial were included. Radiographic spinal progression was defined as a worsening of the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) by ≥2 points after 2 years. A clinical prediction model for radiographic spinal progression was constructed out of already present syndesmophytes, elevated CRP, cigarette smoking, and male sex. Serum biomarkers measured at baseline included: matrix metalloproteinase-3, vascular endothelial growth factor (VEGF), calprotectin, leptin, high molecular weight adiponectin (HMW-APN), osteoprotegerin, sclerostin, N-terminal telopeptide, procollagen type II N-terminal propeptide, and serum amyloid A. Results: Repeated cross-validation analyses revealed the following biomarker combination with added predictive value for radiographic progression compared to the clinical model (syndesmophytes, smoking, elevated CRP, and sex): Leptin +HWM-APN+VEGF. A combination of these biomarkers resulted in an Area Under the Curve (AUC) of AUCBiomarkers=0.731 (95%CI 0.614–0.848), thus numerically superior to the clinical model (AUCClinical=0.665 (95%CI 0.553–0.776)). A combination of both models resulted in improvement of the predictive value reflected by the AUCClinical+Biomarkers=0.768 (95%CI 0.666–0.871); though this improvement was not statistically significant compared to the clinical model in the permutation test (p=0.051). However, when only considering the part of receiver operating characteristic (ROC) curves with a specificity of ≥75%, the improvement becomes statistically significant (partial AUCClinical+Biomarkers=0.119 versus partial AUCClinical=0.053; p=0.010). Abstract FRI0154 – Figure 1 ROC curve analyses of biomarker and clinical model alone versus the combined model for the prediction of radiographic spinal progression (mSASSS worsening ≥ 2 after 2 years) in ankylosing spondylitis. Area under the curves (AUC) and respective 95% confidence intervals (CI) shown. *Leptin + HWM-APN + VEGF. Conclusions: Biomarkers are able to improve prediction of radiographic spinal progression in axSpA, especially if used in addition to the clinical parameters, but the added value seems to be rather small. Disclosure of Interest: None declared DOI: 10.1136/annrheumdis-2018-eular.3795 Citation: Ann Rheum Dis, volume 77, supplement Suppl, year 2018, page A620Session: Spondyloarthritis – etiology, pathogenesis and animal models