Introduction: With the development of increasing numbers of potential therapeutic agents in inflammatory disease comes the need for effective biomarkers to help screen for drug efficacy and optimal dosing regimens early in the clinical trial process. This need has been recognized by the Outcome Measures in Rheumatology Clinical Trials (OMERACT) group, which has established guidelines for biomarker validation. To seek a candidate synovial biomarker of treatment response in psoriatic arthritis (PsA), we determined whether changes in immunohistochemical markers of synovial inflammation correlate with changes in disease activity scores assessing 28 joints (ΔDAS28) or magnetic resonance imaging synovitis scores (ΔMRI) in patients with PsA treated with a biologic agent.
Methods: Twenty-five consecutive patients with PsA underwent arthroscopic synovial biopsies and MRI scans of an inflamed knee joint at baseline and 12 weeks after starting treatment with either anakinra (first 10 patients) or etanercept (subsequent 15 patients) in two sequential studies of identical design. DAS28 scores were measured at both time points. Immunohistochemical staining for CD3, CD68 and Factor VIII (FVIII) was performed on synovial samples and scored by digital image analysis (DIA). MRI scans performed at baseline and at 12 weeks were scored for synovitis semi-quantitatively. The ΔDAS28 of the European League Against Rheumatism good response definition (>1.2) was chosen to divide patients into responder and non-responder groups. Differences between groups (Mann Whitney U test) and correlations between ΔDAS28 with change in immunohistochemical and MRI synovitis scores (Spearman's rho test) were calculated.
Results: Paired synovial samples and MRI scans were available for 21 patients (8 anakinra, 13 etanercept) and 23 patients (8 anakinra, 15 etanercept) respectively. Change in CD3 (ΔCD3) and CD68 expression in the synovial sublining layer (ΔCD68sl) was significantly greater in the disease responders compared to non-responders following treatment (P = 0.005 and 0.013 respectively). ΔCD3, but not ΔCD68 or ΔFVIII, correlated with both ΔDAS28 (r = 0.49, P = 0.025) and ΔMRI (r = 0.58, P = 0.009).
Conclusions: The correlation of ΔCD3 with ΔDAS28 and ΔMRI following biologic treatment in this cohort contributes to the validation of ΔCD3 as a synovial biomarker of disease response in PsA, and supports the further evaluation of ΔCD3 for predictive properties of future clinical outcomes.