Background The treatment of rheumatoid arthritis (RA) has changed considerably over the last couple of decades, particularly since the introduction of biological therapies; however, a considerable proportion of patients fail to respond satisfactorily (30–40%). Early and effective treatment is known to correlate with better long-term outcomes and there is an unmet clinical need for predictors of response which can be used to stratify patients prior to treatment; so that those prone to non-response can be offered alternative options. We hypothesis that gene expression profiles of RA patients about to commence treatment with the biological drug etanercept will provide such a biomarker.

Objectives To identify gene expression biomarkers which discriminate between good-responders (R) and non-responders (NR) to etanercept in RA patients.

Methods 72 RA patients about to commence treatment with etanercept were chosen for analysis from the Biologics in Rheumatoid Arthritis Genetics and Genomics Study Syndicate (BRAGGSS) longitudinal cohort. Using the established EULAR classification criteria this equated to 34 R (28-joint disease activity score (DAS28) of ≤3.2 and improvement of >1.2) and 38 NR (DAS28 >5.1 and improvement <0.6 or between 0.6–1.2) patients at 3-months follow-up. Total RNA was extracted from pre-treatment whole blood previously collected into TEMPUS/PaxGene blood tubes upon recruitment to BRAGGSS. Following biotin labelling and RNA amplification (Illumina TotalPrep RNA Amplification Kit), cRNA was hybridised to Illumina HumanHT-12 v4 Expression BeadChips targeting over 47,000 probes. GenomeStudio software summarised the bead level data, followed by QC and differential expression analysis using the Bioconductor package Limma.

Results Using a fold-change (FC) cut-off of >1.2 between R and NR and p-value <0.05, 82 probes were differentially expressed; 33 up-regulated in NR and 49 up-regulated in R prior to treatment with etanercept. Interestingly, of the probes up-regulated in NR, an interferon (IFN) signature was observed, including the type-I IFNs, IFITM3 (FC 1.53, p-value=0.01), MX1 (FC 1.45, p-value=0.02) and IFI44L (FC 1.28, p-value=0.04) and a type-II IFN, BTN3A2 (FC 1.38, p-value 9.6 x 10^–05). Of note, expression of PTPRC, a previously reported treatment response gene, was found to be up-regulated in R as compared to NR prior to treatment (FC 1.24, p-value=0.03).

Conclusions The preliminary results suggest that an IFN signature prior to etanercept treatment is indicative of non-response in RA patients. These results are encouraging in that an IFN signature has consistently been associated with RA and clinical outcome to biological therapies. However, to the best of our knowledge this is the first study to observe an IFN signature correlating with response to etanercept. More in-depth analysis is ongoing using packages such as WGCNA (weighted correlation network analysis) and Matrix eQTL to further explore these associations.

Acknowledgement SLS was a PhD student funded by an investigator-initiated award to AB from Pfizer (award number WS1940162).

Disclosure of Interest None declared