Background/Purpose: In rheumatoid arthritis (RA), initiating effective treatment as soon as possible within the so-called therapeutic “window of opportunity” is the strategy, and remission is a primary goal. Recent work from our group demonstrated that pre-treatment serum type I IFN- β/α ratio >1.3 can predict non-response to anti-TNF-alpha therapy in RA patients. The cellular mechanisms that underlie the IFN-β/α ratio that predicts response are not known. Effects of IFN on single cells and immune cell subtypes may be masked in whole blood or mixed cell populations.

Methods: To better understand the underpinnings of the pre-treatment IFN-β/α ratio, we used single cell expression analysis to investigate whether monocyte gene expression differs significantly between RA patients according to their pre-TNF-α inhibitor serum IFN-β/α ratio. Single classical (CL) and single non-classical (NC) blood-derived monocytes were isolated from 15 seropositive RA subjects prior to biologic therapy. Subjects were grouped by pre-TNF-α inhibitor serum IFN-β/α ratio into two groups, IFN-β/α >1.3 (n=6) and IFN-β/α <1.3 (n=9). We performed unsupervised hierarchical clustering of 87 target genes, and compared groups by Mann-Whitney and Fisher’s. Genes that differed in categorical analysis were tested in logistic regression models.

Results: JAK1 and IL1A were strong differentiators between patients with IFN-β/α < 1.3 and IFN-β/α > 1.3, the groups which correspond to response/non-response to anti-TNF agents. In categorical analyses, in NC cells alone, expression (OR, p) of STAT2 (19.2, <0.0001), ILT7 (10.4, 0.02)), PKR (8.9, 0.03), TLR7 (3.1, 0.03), and IRAK1 (3.4, 0.04) was more likely in the non-response group. In CL monocytes alone, expression of IFIT2 (8.9, 0.04) and CD36 (9.7, 0.04) was more likely. In multivariate logistic regression, IL1A, CD32a, IL-8, TYK2, and IRAK1 expression in monocytes (CL+NC) aligned with treatment response groups. Each was also strongly statistically significant in regression models of monocyte subsets. In comparison to the mixed monocyte model, IL-8 and IRAK1 in NC and CXCR3 in CL cells demonstrated even stronger alignment with response groups. STAT2 was strongly predictive of response group in NC cells alone. CXCL9 was strongly predictive of response group in CL cells alone. Models from monocyte subsets provided higher area under the curve in ROC analysis in comparison to the mixed monocyte model.

Conclusion: Within-cell co-expression patterns demonstrate biological differences in monocyte subsets of RA patients with an IFN-β/α > 1.3, the ratio of type I IFNs which predicts non-response to anti-TNF therapy. Differentiation by gene expression was strongest among the response/non-response patient groups when monocyte subsets were analyzed separately, and distinct expression signatures were identified, suggesting that further study of monocyte subsets will likely illuminate molecular differences that determine treatment response to TNF-α inhibition in RA. This work will help to develop a more individualized approach to therapy in RA based upon the underlying biology of disease in a given patient.

« Back to 2016 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/distinct-single-cell-gene-expression-signatures-of-monocyte-subsets-differentiate-between-tnf-alpha-inhibitor-treatment-response-groups-in-rheumatoid-arthritis/