Background Predicting response to biologic therapies in RA remains a clinical challenge. The anti-CD20 monoclonal antibody Rituximab effectively depletes peripheral blood B-cells, however response rates are approximately 60%. We hypothesized that expression of CD20 in diseased tissue may be an important predictor of response, since up to 40% of patients have few/no B-cells present in the synovium.

Objectives To ascertain whether a synovial B-cell gene signature can enhance prediction of responsiveness to Rituximab therapy and identify genes involved in response/non-response to treatment.

Methods Synovial tissue was obtained using ultrasound-guided synovial biopsies from 20 patients with active RA who were treated with Rituximab therapy after failure of conventional DMARD and anti-TNF therapy. High-throughput quantitative real-time PCR for 190 genes was performed in collaboration with MedImmune (MedImmune, LLC) using the Fluidigm platform. Samples were classified as B-cell “rich” or “poor” according to levels of MS4A1 expression. Using an empirical Bayes statistical model, a further 17 significant (p<0.05) genes were identified to create a baseline B-cell gene expression signature. Hierarchical clustering and receiver operating characteristic curve analysis was used to assess ability to predict EULAR criteria response at 16 weeks. Gene expression was compared pre- and post-treatment in responders versus non-responders and correlated with delta DAS-28. Logistic regression with backward and forward selections using AIC was applied to identify further predictive models.

Results Seventeen genes were identified that clustered with MS4A1 expression to create the B-cell signature. Fifteen of these are directly involved in B-cell and plasma-cell signaling or differentiation pathways and immunoglobulin synthesis. Two genes identified within this signature play a pivotal role in antigen presentation and TH17 differentiation. The baseline B-cell signature had an area under ROC curve (AUC) of 0.87 (95%CI 0.76–1.0) to predict response to Rituximab therapy at 16 weeks. Three additional genes were identified segregating responders from non-responders; that when combined with the baseline B-cell signature the AUC improved to 0.95. Additionally, ten genes were identified that significantly correlated with delta DAS-28. High expression of genes involved in the formation of ectopic lymphoid structures and the TH17 lineage correlated with non-response to Rituximab therapy (all p<0.05).

Conclusions This study shows that a B-cell rich gene signature has high sensitivity to predict response to Rituximab therapy. High expression of TH17 related genes correlates with non-response to B-cell depletion, indicating that in a subset of patients, these pathways may play a more pivotal role in disease pathogenesis. The observed high expression of ELS associated genes in non-response may relate to variable synovial B-cell depletion; disruption of these follicles may be the key to response, particularly with repeat cycles. These findings warrant confirmation in a larger cohort of patients; however they strongly support the notion that a stratified approach to treatment relies upon the identification of biomarkers, both in the synovium and peripheral blood that reflect heterogenous molecular disease pathways in RA.

Disclosure of Interest None declared