Measuring response to treatment in axial spondyloarthritis using quantitative imaging biomarkers: a prospective observational cohort study

Objective Objective assessments of disease activity and response to treatment in axial spondyloarthritis (axSpA) remain a challenge; quantitative imaging biomarkers (QIBs) of inflammation could enhance assessments of disease activity and therapeutic response. We aimed to determine the responsiveness of QIBs obtained from diffusion-weighted imaging (DW-MRI) and chemical shift-encoded MRI (CSE-MRI) using the partially automated Bone Edema and Adiposity Characterisation with Histograms (BEACH) software tool in axSpA patients undergoing biologic therapy. Methods We conducted a prospective longitudinal cohort study, including 30 patients with axSpA undergoing biologic therapy. Patients were scanned before and after biologic therapy using conventional MRI, DWI and CSE-MRI at 3T. Apparent diffusion coefficient (ADC) and proton density fat fraction (PDFF) were assessed using the BEACH tool (https://github.com/TJPBray/ BEACH), and conventional MR images were assessed using established visual scoring methods by expert radiologists. Responsiveness – the ability of the MRI measurements to capture changes in disease occurring as a result of biologic therapy – was assessed using the standardized response mean (SRM). Inter-reader relia-bility of the ADC and PDFF maps was assessed using Bland-Altman limits of agreement analysis and the intra-class correlation coefficient. Results Responsiveness to therapy was moderate for ADC-based parameters (SRM 0.50) and comparable to established visual scoring methods for bone marrow oedema (SRM 0.53). Interobserver variability was lower for QIBs compared with conventional visual scores methods. Conclusions QIBs measured using the BEACH tool are sensitive to changes in inflammation in axSpA following biologic therapy, with similar responsiveness and lower interobserver variability to visual scoring by expert radiologists. Advances in knowledge QIBs measured using the partially automated BEACH tool offer an objective measure of response to biologic therapy in axSpA.