Micro-RNA content of circulating extracellular vesicles in early rheumatoid arthritis as biomarkers and mediators of methotrexate efficacy

Abstract

Objectives: Extracellular vesicles (EVs) are abundant in body fluids, contributing to intercellular signalling by transferring cargo that includes microRNAs (miRs)-themselves implicated in pathobiology. For the first time we evaluated the potential of EV miRs to contribute diagnostic information in early RA, predict methotrexate (MTX) efficacy or shed light on the drug's mechanism of action.

Methods: Seven hundred and ninety-eight miRs isolated from serum-derived EVs of 46 patients with untreated RA, 23 with untreated polymyalgia rheumatica (PMR; inflammatory disease control group) and 12 in whom significant inflammatory disease had been excluded (non-inflammatory controls; NICs) were profiled (NanoString); the same measurements were made for RA patients after 6 months' MTX treatment. Analyses took multiple testing into account.

Results: Twenty-eight EV miRs were robustly differentially expressed between early RA (but not PMR) patients and NICs after correction for age and sex, suggesting discriminatory value. Cross-validated partial least squares-discriminant analysis also indicated the predictive potential of a distinct baseline EV miR signature with respect to MTX-induced remission at 6 months. The change in expression of 13 miRs over the course of MTX treatment differed significantly between responders and non-responders, and four of those exhibiting increased relative abundance amongst responders have known roles in regulating the pathogenic potential of synovial fibroblasts, namely miR-212-3p, miR-338-5p, miR-410-3p and miR-537.

Conclusion: Our data highlight the potential of serum EV miRs as diagnostic and therapeutic biomarkers, highlighting a novel potential mechanism by which MTX may exert its therapeutic effect in early RA that warrants further investigation.

Keywords: DMARDs; biomarkers; microRNA; rheumatoid arthritis.

Figure 1.

An EV miR signature for discriminating untreated early RA from control groups. (A and B) Volcano plots displaying miRs significantly downregulated (red; upper left) and up-regulated (blue; uppper right) in RA (A) or PMR (B) compared to NIC control group. Each data point is a single miR and horizontal lines represent significance thresholds for α = 5% (Wald test) with and without multiple test correction. Vertical lines represent a fold change of ±1.25. (C) Venn diagram summarizing the number of significantly differentially expressed miRs between both disease states and disease controls according to whether the findings are or are not robust to multiple test correction (above and below horizontal line, respectively; see Supplementary Table S2 and S3, available at Rheumatology online for complete listings). (D) Heatmap displaying Z-scores for 28 uniquely RA-associated miRs. Direct comparison between RA and PMR groups. Clustering by Euclidian distance. *Adj P: adjusted P-value employing false discovery rate (FDR); EV: extracellular vesicle; FC: fold change; miR: microRNA; PMR: polymyalgia rheumatica

Figure 2.

A pre-treatment EV miR signature for predicting early methotrexate response in newly diagnosed RA. (A) Volcano plots displaying miRs significantly downregulated (red) and up-regulated (blue) at baseline in RA patients who responded to MTX compared with those who did not. Each data point is a single miR and horizontal lines represent nominal significance threshold for α = 5% (Wald test; no multiple test correction). Vertical lines represent a fold change of ±1.25. (B) PLS-DA plot demonstrating statistical difference between MTX responders (orange) and non-responders (blue) at baseline. EV: extracellular vesicle; FC: fold change; miR: microRNA; MTX: methotrexate; PLS-DA: partial least squares discriminatory analysis

Figure 3.

Differential dynamic changes in EV miR expression according to methotrexate responsiveness. (A) Scatter plot displaying linear model output testing responder effect for all 798 miRs included in the analysis. Negative log-transformed P-values are plotted against the rank of the size-of-effect estimate for each miR. Each data point is a single miR and horizontal lines represent significance thresholds for α = 5% with and without multiple test correction for effect difference between responders and non-responders, with the direction of difference represented by colour as indicated in the key. Response-regulated miRs subject to age: sex interactions (see text) are circled. (B) Plots displaying log fold-change pre- and post-treatment for the 13 miRs for which there was a significantly different change in abundance over time between responders and non-responders. EV: extracellular vesicle; LFC: log fold change; LM: linear model; miR: microRNA