Title: RNA transcripts from peripheral blood mononuclear cells as predictors of clinical responsiveness in rheumatoid arthritis subjects treated with abatacept

Authors: Matthew Henkel; Fang Du; Donald M. Jones; Erich R. Wilkerson; William Horne; Jay Kolls; Marc C. Levesque; Mandy J. McGeachy

Background/Purpose:
Biologics, including abatacept (CTLA4Ig), improve outcomes for many RA patients. However, approximately 40-50% of RA patients fail to respond to abatacept, and there are currently no biomarkers that predict responsiveness. Analysis of a defined panel of peripheral blood mononuclear cell (PBMC) gene transcripts is a non-invasive and technically feasible approach for routine clinical use to predict abatacept responsiveness. Advances in transcriptome profiling techniques (RNA-Seq) now allow high-throughput “deep-sequencing” of relatively small amounts of input RNA. Direct detection of transcripts by RNA-Seq offers several advantages over conventional microarrays: high sensitivity, low background signal, high dynamic range and accuracy of transcript quantification and high reproducibility.

Methods:
We analyzed RNA samples derived from PBMC from 6 subjects treated with abatacept ± oral DMARDs and ± prednisone who were enrolled in the Rheumatoid Arthritis Comparative Effectiveness Research (RACER) registry at the University of Pittsburgh. Five of 6 subjects were anti-CCP positive and all 6 RA subjects had active disease at baseline (mean DAS28-CRP ± SD; 4.4 ± 0.7) despite recent TNF inhibitor therapy. Based on DAS28-CRP scores at baseline and at 6 months after abatacept initiation, 3 of the RA subjects were deemed responders (DDAS28-CRP = -1.3 ± 1.0) and 3 non-responders (DDAS28 = 0.1 ± 0.8). PBMC RNA samples from the 6 RA subjects were analyzed by RNA-Seq prior to receiving abatacept and approximately 2 months (6 to 10 weeks) after abatacept initiation. We identified genes that differed at baseline between abatacept responders and non-responders, and for responders and non-responders, genes that changed between baseline and 2 months by ≥ 1.3 fold with p < 0.05 (t test).

Results:
There was relatively little overlap between responders and non-responders when analyzing RNA transcript changes from baseline to 2 months (< 10 transcripts). A substantially larger proportion of transcripts were significantly altered (increased or decreased) from baseline to 2 months in responders (6339 transcripts) compared to non-responders (117 transcripts). We analyzed expression of genes related to T and B cell function, analyzing baseline predictors of response (different at baseline between responder and non-responder groups) and 2 month predictors of response (different at 2 months versus baseline). We found that PBMC RNA transcripts for IgG isotypes and IL-17 were good 2-month predictors of a 6-month clinical response, but baseline levels of these transcripts did not predict efficacy. In contrast, IL6R transcripts were a good baseline predictor of efficacy but did not change from baseline to 2 months.

Conclusion:
These data support the sensitivity of RNA-Seq as an assay for responses to biologic therapies in PBMC from RA patients. These RNA-Seq results, with only three subjects per group demonstrate the great potential of this technique to elucidate both mechanistic and biomarker-related pathways altered in PBMC by therapy. Future studies will validate these results in prospectively collected samples and expand these analyses to other biologic therapies.