Abstract

Background

   Effective biological markers able to monitor the response of Janus
   kinase inhibitor (JAKi) are lacking. Exosomal microRNAs (exomiRNAs) can
   alter their expression during treatment and are ideal biomarkers for
   therapeutic interventions. In this study, we explored potential
   biomarkers for monitoring tofacitinib treatment response in patients
   with RA.

Methods

   Peripheral blood mononuclear cells (PBMCs) were collected from 35
   healthy controls (HCs) and 74 patients with methotrexate
   (MTX)-resistant new-onset RA. We analyzed the profiles of exomiRNAs
   using next-generation sequencing (NGS) and verified them using
   quantitative real-time polymerase chain reaction (qRT-PCR). The
   functional roles of the selected exomiRNAs were analyzed using
   bioinformatics tools. Potential exomiRNAs were validated in
   MTX-resistant RA patients treated with tofacitinib for 3 months.

Results

   Fifty-six differentially expressed exomiRNAs were identified. High
   expressions of the exo-(miR-548ah-3p, miR-378 g, miR-27a-5p, and
   miR-30c-2-3p) were validated by qRT-PCR. Enrichment analysis indicated
   that these exomiRNAs may regulate immune cells and mediate immune
   responses. Exo-miR-27a-5p levels significantly decreased after
   tofacitinib treatment (p < 0.0001) and showed a strong correlation with
   the DAS28, RF and ESR. Receiver operating characteristic curve analysis
   showed that changes in the expression levels of exo-miR-27a-5p were
   significantly correlated with tofacitinib therapy (AUC = 0.92,
   p < 0.0001).

Conclusions

   This study suggests that circulating exo-miR-27a-5p is a novel
   non-invasive biomarker to monitor the response to tofacitinib
   treatment.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s41927-025-00502-1.

   Keywords: Rheumatoid arthritis, Circulating exosomes, MiRNAs,
   Biomarker, Tofacitinib

Key points

   • JAKi is frequently used to treat rheumatoid arthritis; however, no
   effective biological markers are available to monitor the treatment
   response.

   • Exo-miR-27a-5p is a novel non-invasive biomarker for monitoring the
   response to tofacitinib treatment.

Background

   Rheumatoid arthritis (RA) is a systemic rheumatic disease with
   persistent joint destruction as its main clinical manifestation; its
   prevalence in China is approximately 0.28% [[32]1]. Early diagnosis,
   standardized treatment, and treatment-to-target (T2T) are the clinical
   treatment approaches for patients with RA. Sustained intensive
   treatment is a better approach for achieving clinical remission and
   better clinical outcomes [[33]2]. However, rheumatoid factor (RF) and
   anti-citrullinated peptide antibodies (ACPAs) are not satisfactory
   markers for clinical use and are neither diagnostic biomarkers nor good
   predictors of responses to disease-modifying antirheumatic drugs
   (DMARDs) [[34]3–[35]5]. These findings indicate the need to identify
   new non-invasive biomarkers to improve personalized therapeutic
   decisions.

   In recent years, growing evidence has shown that microRNAs (miRNAs)
   play key roles in RA, including genetic susceptibility, immune
   inflammation, changes in synovial pathology, and bone destruction
   [[36]6, [37]7]. MiR-22 is associated with joint damage in patients with
   early RA, indicating that it is a circulating diagnostic biomarker
   [[38]8]. The miR-224, miR-760, miR-483-5p, miR-378, and miR-375 levels
   are positively correlated with the disease activity score (DAS)28 and
   may be prognostic biomarkers [[39]9]. Furthermore, some miRNAs can
   predict the response to TNF-a inhibitors, such as the levels of
   miR-99a, miR-143 miR-23a, and miR-197, which are associated with the
   response to Adalimumab (ADA) and Etanercept (ETN), respectively
   [[40]10].

   Exosomes are carriers of miRNAs, 30–150 nm in diameter, and are widely
   detected in plasma, serum, urine, and other body fluids; they also play
   a role in mediating communication between different tissues and cells,
   making them ideal biomarkers [[41]11–[42]13]. In recent studies,
   several circulating exomiRNAs have been identified as diagnostic
   biomarkers of RA, such as exo-miR-1915-3p, exo-miR-451a, and
   exo-miR-25-3p [[43]14, [44]15]. A few studies have suggested that
   exo-miR-122-5p, exo-miR-155-5p [[45]16], miR-1298-5p [[46]17], and
   miR-19b [[47]18] may act as novel biomarkers of the efficacy of
   tofacitinib. Although circulating exomiRNAs have been established as RA
   biomarkers, their role in predicting drug efficacy remains unclear. In
   this study, we explored circulating exomiRNAs as novel biomarkers of
   the therapeutic efficacy of tofacinib and analyzed their biological
   functions.

Methods

Study design

   Seventy-four patients with RA who met the criteria for the American
   College of Rheumatology/European League Against Rheumatism
   classification (2010) and sixty-one healthy controls (HCs) were
   recruited from volunteers during the same period. Patients with RA
   remained in high disease activity after methotrexate (MTX) treatment
   for more than 3 months and had a disease duration of less than 1 year.
   Patients were divided into four groups: sequencing, filtering,
   validation, and treatment. In the sequencing group, samples from HCs
   (n = 4) and patients with RA (n = 4) were analyzed using
   next-generation sequencing (NGS). In the filter group, samples from HCs
   (n = 5) and patients with RA (n = 5) were analyzed by quantitative
   real-time polymerase chain reaction (qRT-PCR) to screen for exomiRNAs.
   HCs (n = 26), and RA (n = 33) samples from the validation group were
   used to confirm exomiRNAs. In the treatment group, RA patients (n = 32)
   with a previous failure of MTX monotherapy would receive tofacitinib
   treatment for an additional 3 months. The clinical characteristics of
   the participants are shown in Table [48]1.

Table 1.

   Basic clinical characteristics of patients with rheumatoid arthritis
   and healthy controls
   Clinical variables Sequencing group Filter group Validation group
   Treatment group
   HC (4) RA (4) HC (5) RA (5) HC (26) RA (33) RA(32)
   Age (years) 44±11.5 55.5±4.4 49±18.8 56.8±14.1 54.9±9.1 57.5±8.5
   54.8±11.3
   Male/Female (n) 0/4 0/4 1/4 1/4 5/21 7/26 7/25
   RF (IU/mL) NA 166.6 (152.4–179.6) NA 161.3 (145–185.4) NA 172
   (22.4–394) 124.2 (3.5–203.6)
   ACPA (RU/mL) NA 1842 (1600–2000) NA 1289 (157.2–2000) NA 1053
   (117–1951) 647.9 (110.6–2000)
   CRP (mg/L) NA 49.3 (9.2–147.8) NA 15.75 (1.30–46.5) NA 23.3 (6.2–53)
   27.5 (1.2–92)
   ESR (mm/h) NA 82 (70.8–98) NA 88.8 (57–104) NA 69.7 (32–110) 62
   (11–120)
   DAS28 NA 6±0.5 NA 5.4±0.6 NA 5.9±0.5 5±0.9
   [49]Open in a new tab

Sample collection and exosome isolation

   Fresh blood samples were obtained from participants in EDTA collecting
   tubes (BD, USA). The PBMCs were isolated by the EasySep™ Direct Human
   PBMC Isolation Kit (STEMCELL, Beijing, China) following the protocol.
   Exosomes were isolated using different centrifugation procedures, as
   described previously [[50]19].

Circulating exosome identification

   We used three methods to identify exosomes: nanoparticle tracking
   analysis (NTA), transmission electron microscopy (TEM), and Western
   blotting identification of protein markers. All procedures were
   performed as described previously [[51]20]. Briefly, exosomes were
   dropped onto a copper grid for 1 min and phosphotungstic acid was added
   for 30 s. Images were captured at 80 kV using TEM (Hitachi, H7650,
   Japan). The particle size and concentration of the exosomes were
   measured by NTA (Particle Metrix, Germany). Exosome markers, including
   CD9 (1:1000, ab307085, Abcam), CD63 (1:2000, MA5-35208, Thermo Fisher),
   and HSP70 (1:1000, 4872, Cell Signaling Technology) were identified by
   Western blotting.

Isolation and sequencing of ExomiRNAs

   Total exomiRNAs were extracted by RNAzol^® RT RNA isolation reagent
   (GeneCopoeia, QP020, USA) according to the manufacturer’s protocol.
   ExomiRNAs were sequenced using a HiSeq 2500 (Illumina, USA) at RiboBio
   Co., Ltd. (RiboBio, China). Differentially expressed exomiRNAs were
   filtered using the following criteria (|log2FC| > 3, adj.
   p-value < 0.05) based on the false discovery rate (FDR), as shown in
   the Supplementary Table [52]1.

Quantitative real-time polymerase chain reaction (qRT-PCR)

   The expression of circulating exomiRNAs was measured by qRT‒PCR using
   the All-in-One™ miRNA qPCR Kit (GeneCopoeia, QP010, USA), as described
   previously [[53]21]. Cel-miR-39-3p (miRB0000010-3-1; RiboBio, China)
   was used as an external control. The expression of exomiRNAs was
   determined in three replicates and calculated using the 2^−ΔΔCt method.
   The primers used are listed in Supplementary Table [54]2.

Prediction of the potential target gene of ExomiRNAs

   To investigate the potential regulatory genes of exomiRNAs, we used
   three online databases, including TargetScan, miRDB, and mirDIP. The
   results extracted from the three databases were compared to obtain a
   set of potential regulatory target genes for exomiRNAs. We analyzed the
   data from three microarray datasets ([55]GSE55235, [56]GSE55457, and
   [57]GSE77298) from the Gene Expression Omnibus (GEO) database, as shown
   in Online Resource 1, Supplementary Table [58]3. Differentially
   expressed genes (DEGs) in RA were evaluated using a conservative
   threshold (|log2FC| > 0.5, adj. p. value < 0.05). Finally, we searched
   for genes in public databases, including CTD, DisGeNET, DrugBank,
   GeneCards, OMIM, and TTD. The “marker/mechanism and therapeutic” genes
   for direct evidence were selected from the CTD database. In the
   DisGeNET and GeneCards databases, genes with Score_gda and relevance
   scores higher than the average values were selected, and a defined
   Entrez Gene ID was required in the OMIM database. Potential regulatory
   target genes for exomiRNAs were obtained by comparing the prediction
   target sets with three GEO microarray datasets and six public
   databases.

Functional enrichment analysis

   We used Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and
   Genomes (KEGG) to annotate the potential targets of exomiRNAs through
   the clusterProfiler package (v.4.7.1). The selection criteria were a
   p-value < 0.05 and a q-value < 0.05. Based on the enrichment analysis
   results, a data file was then established to construct the interaction
   network between potential regulated targets of exomiRNAs and pathways
   using the Cytoscape software.

Construction of the protein–protein interaction (PPI) network

   The STRING database (v.11.0) was used to construct an interaction
   network of potential exomiRNA targets in RA. Parameter settings
   included Homo sapiens as the organism and a combined score > 0.4.
   Molecular complex detection (MCODE) in Cytoscape (v.3.7.1) was used to
   analyze the network module, and the protein–protein interaction (PPI)
   network was constructed using Gephi (v.0.9.2).

Statistical analysis

   Data are presented as the median (interquartile range) or
   mean ± standard deviation (SD). The Wilcoxon signed rank test for
   matched pairs and Mann–Whitney U test were used to calculate
   differences between patients with RA and HCs. The false discovery rate
   (FDR) was calculated using the BioLadder modules in the proteome data
   analysis framework online
   ([59]https://www.bioladder.cn/web/#/pro/index). To evaluate the
   sensitivity and specificity of circulating exomiRNAs as biomarkers for
   Janus kinase inhibitors (JAKis), the Wilson/Brown method was used in
   ROC curve analysis via the MedCalc statistical software. Statistical
   analyses were performed using R v.4.0.2 and Prism v.9.5 (GraphPad, San
   Diego, CA, USA). P-values < 0.05 were considered statistically
   significant.

Results

Characterization of circulating exosomes

   Circulating exosomes isolated from the HCs and participants with RA
   were identified using TEM, NTA, and Western blotting. According to the
   TEM and NTA results, the exosome size ranged from 30 to 150 nm in the
   HC and RA groups (Fig. [60]1a–d). Specific markers CD9, CD63, and HSP70
   were detected by Western blotting and were clearly observed in
   circulating exosomes of both groups (Fig. [61]1e).

Fig. 1.

   [62]Fig. 1
   [63]Open in a new tab

   Identification of circulating exosomes. (a, b) Size and concentration
   of circulating exosomes from patients with rheumatoid arthritis (RA)
   and healthy controls (HCs) analyzed by nanoparticle tracking analysis
   (NTA), respectively; (c, d) Circulating exosomes of patients with RA
   and HCs verified by transmission electron microscopy (TEM),
   respectively; scale bar: 200 nm. (e) Specific exosome markers CD9, CD63
   and HSP70 analyzed by Western blotting

Differential expression of circulating ExomiRNAs from patients with RA and
HCs

   We evaluated the expression of exomiRNAs using NGS in PBMCs from
   patients with RA and HCs. The levels of 45 exomiRNAs were significantly
   elevated, whereas those of 11 exomiRNAs were decreased (Fig. [64]2a,
   b). Based on the p-value and fold changes, we selected 11 exomiRNAs
   (exo-miR-92b-5p, exo-miR-582-3p, exo-miR-548ah-3p, exo-miR-500a-3p,
   exo-miR-450a-3p, exo-miR-378 g, exo-miR-27a-5p, exo- miR-1285-3p,
   exo-miR-10399-3p, exo-miR-30c-2-3p, and exo-miR-760) for preliminary
   screening in the filtered group using qRT-PCR (Fig. [65]2c–m).

Fig. 2.

   [66]Fig. 2
   [67]Open in a new tab

   Screening of differentially expressed circulating exomiRNAs in patients
   with RA and HCs. (a) Volcano map of differentially expressed
   circulating exomiRNAs in the sequencing group; Green: downregulated
   exomiRNAs in the sequencing group; Gray: not significant. (b) Heatmap
   of differentially expressed circulating exomiRNAs. (c–m) qRT-PCR
   detected the expression of 11 candidate circulating exomiRNAs in the
   selected group. The values are normalized to Cel-miRNA-39-3p and shown
   on a log10 scale on the 𝑦-axis. All data are represented as the
   mean ± SD. ***p < 0.001, ** p < 0.01, * p < 0.05

   The expression levels of exo-miR-548ah-3p, exo-miR-378 g,
   exo-miR-27a-5p, and exo-miR-30c-2-3p) in the PBMCs of patients with RA
   were significantly different from those of the HCs. Similar findings
   were observed in the qRT-PCR results of the validation group
   (Fig. [68]3a–d). The ROC areas of exo-miR-27a-5p, exo-miR-30c-2-3p,
   exo-miR-548ah-3p, exo-miR-378 g were 0.79 (95% CI: 0.67– 0.91), 0.80
   (95% CI: 0.68–0.91), 0.82 (95% CI: 0.71– 0.93), and 0.74 (95% CI:
   0.59–0.88), respectively (Fig. [69]3e). We analyzed the profiles of
   four potential exo-miRNAs as a whole to distinguish the healthy
   controls and RA patients. The results showed that the estimated AUC was
   0.97 (95%CI: 0.89–0.99, p < 0.0001, cut off value: 0.69, YI: 0.81,
   sensitivity: 84.85%, specificity: 96.15%, as shown in Fig. [70]3e).

Fig. 3.

   [71]Fig. 3
   [72]Open in a new tab

   Validation of candidate targets of circulating exomiRNAs. (a–d)
   Relative levels of four circulating exomiRNAs in the validation group.
   (e) Receiver operating characteristic (ROC) curve of four circulating
   exomiRNAs for RA. The values are normalized to Cel-miRNA-39-3p and
   shown on a log10 scale on the 𝑦-axis. All data are shown as mean ± SD.
   ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05

Prediction and screening for ExomiRNA targets

   Based on online databases, we predicted that exo-miR-27a-5p,
   exo-miR-30c-2-3p, exo-miR-378 g, and exo-miR-548ah-3p had 186, 78, 358,
   and 48 potential target genes, respectively. A total of 631 target
   genes regulated by the four exomiRNAs were identified after merging the
   data (see Online Resource 1, Supplementary Fig. [73]1a–d). Furthermore,
   we analyzed DEGs from the GEO database, and 2757 DEGs associated with
   RA were obtained after merging the data (see Online Resource 1,
   Supplementary Fig. [74]2a–f). A total of 106 potential target genes of
   the four exomiRNAs were obtained after merging the profiles of
   three-GEO, and -online databases. Furthermore, we selected and
   identified 1731 RA-related genes by merging them from six public
   databases (CTD, DisGeNet, DrugBank, GeneCards, OMIM, and TTD) and 190,
   472, 586, 841, 314, and 130 genes, respectively (see Online Resource 1,
   Supplementary Fig. [75]2g). After combining the results of the above
   two methods, 133 target genes of the exomiRNAs were identified (see
   Online Resource 1, Supplementary Fig. [76]2h). Potential target genes
   were subjected to functional enrichment analysis.

Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway
enrichment analysis

   The clusterProfiler R package was used to analyze 133 target genes
   regulated by circulating exomiRNAs. The results from the gene ontology
   (GO) analysis showed that these target genes were mainly enriched in
   169 biological processes, including regulation of hemopoiesis
   (GO:1903706), lymphocyte differentiation (GO:0030098), mononuclear cell
   differentiation (GO:1903131), myeloid cell differentiation
   (GO:0030099), gland development (GO:0048732), embryonic organ
   development (GO:0048568), cell‒cell adhesion regulation (GO:0022407),
   alveolar-type cell migration (GO:0001667), regulation of leukocyte
   differentiation (GO:1902105), and B-cell activation (GO:0042113). Two
   molecular functions (MFs) were also enriched: protein serine/threonine
   kinase activity (GO:0004674) and protein serine kinase activity
   (GO:0106310) (Fig. [77]4a). Kyoto Encyclopedia of Genes and Genomes
   (KEGG) analysis revealed that the 133 RA-related genes regulated by
   exomiRNAs were enriched in 84 pathways, including in the human
   papillomavirus infection (hsa05165), influenza A (hsa05164),
   Alzheimer’s disease (hsa05010), JAK/STAT signaling pathway (hsa04630),
   Kaposi sarcoma-associated herpesvirus infection (hsa05167),
   Epstein–Barr virus infection (hsa05169), and proteoglycans in cancer
   (hsa05205) (Fig. [78]4b). The enrichment network revealed the
   relationship between potential exomiRNA target genes and pathways
   (Fig. [79]4c).

Fig. 4.

   [80]Fig. 4
   [81]Open in a new tab

   Enrichment analyses of the biological functions of exomiRNAs. (a) Gene
   ontology (GO) annotation of the biological processes (BPs) and
   molecular functions (MFs) for targets of exomiRNAs. (b) Kyoto
   Encyclopedia of Genes and Genomes (KEGG) analysis of possible signaling
   pathways of exomiRNAs. (c) Enrichment analysis network between target
   genes of exomiRNAs and pathways. (d) PPI network showing the 133
   exomiRNA targets. (e) MCODE analysis showing the network of the 9 hub
   genes. (f) Regulatory network between exomiRNAs and 9 hub genes. The
   brown circles represent hub genes and the blue triangles represent
   exomiRNAs

PPI network analysis of the hub genes of ExomiRNAs

   The 133 target genes regulated by exomiRNAs were imported into the
   STRING database and a PPI network was constructed. Seven genes showed
   high degree values: AKT1 (34), FN1 (23), STAT1 (14), IL2 (12), CASP8
   (10), HSPA4 (10), and PIK3CA (10) (Fig. [82]4d). Furthermore, the MCODE
   analysis revealed that a submodule was present in the PPI network,
   which included AKT1, FN1, STAT1, IL2, CASP8, HSPA4, ATG7, CCR5, and
   PTPN2 (Fig. [83]4e). Interestingly, five of nine genes had higher
   degree values (degree ≥ 10) in the PPI network. This module is a key
   submodule of the exosome-derived miRNA regulatory target gene PPI
   network (Fig. [84]4f).

Impact of Tofacitinib therapy on the expression of potential ExomiRNAs

   Bioinformatics analysis indicated a possible association between
   circulating exo-(miR-27a-5p, miR-30c-2-3p, miR-378 g, and miR-548ah-3p)
   and the JAK/STAT pathway. Consequently, tofacitinib was selected as the
   therapeutic drug, and alterations in exosomal miRNAs in patients with
   RA were monitored before and after treatment. The results indicated a
   significant decrease in exo-miR-27a-5p levels following tofacitinib
   treatment (p < 0.001), but without statistical variance when compared
   with HCs (Fig. [85]5a). the expression levels of exo-miR-548ah-3p,
   exo-miR-378 g, and exo-miR-30c-2-3p were not significantly different
   (p > 0.05) after JAKi therapy (as shown in Online Resource 1,
   Supplementary Fig. [86]3a–c). The altered expression of exo-miR-27a-5p
   was strongly correlated with DAS28, RF, and ESR after tofacitinib
   therapy (Fig. [87]5b–d). However, the remaining three exomiRNAs were
   not correlated with RF or ESR (p > 0.05); only exo-miR-378 g was
   correlated with the DAS28 score (p = 0.001, see Online Resource 1,
   Supplementary Fig. [88]4a–c). The ROC curve was performed to evaluate
   the value of exosomal miR-27a-5p expression levels in predicting
   therapeutic response to JAK inhibitor (JAKi) therapy. The estimated AUC
   was 0.92 (95% CI: 0.82–0.97, p < 0.0001, cut off value: 0.93, YI: 0.75,
   sensitivity: 90.62%, specificity: 84.37%, shown in Fig. [89]5e). The
   results of ROC curve indicated that changes in the exo-miR-27a-5p
   expression levels were significantly correlated with JAKi therapy
   outcomes.

Fig. 5.

   [90]Fig. 5
   [91]Open in a new tab

   Impact of tofacitinib therapy on the expression of potential exomiRNAs.
   (a) Levels of exo-miR-27a-5p before and after tofacitinib therapy.
   (b–d) Correlation analysis of exo-miR-27a-5p expression with DAS28,
   ESR, and RF before and after tofacitinib treatment. (e) ROC analysis of
   the sensitivity and specificity between changes in the expression of
   exo-miR-27a-5p and tofacitinib therapy. Values are normalized to
   CelmiRNA-39-3p and are shown on a log10 scale on the 𝑦-axis. All data
   are shown as the mean ± SD. ****p < 0.0001, ***p < 0.001, ** p < 0.01,
   * p < 0.05

Discussion

   Tofacitinib is a sustained intensive DMARD for clinical RA treatment
   with good efficacy and safety. Evidence suggests that the
   downregulation of tumor necrosis factor-α (TNFα), interleukin- 6
   (IL-6), C-X-C motif chemokine ligand 1(CXCL1), matrix
   metalloproteinase-1(MMP-1), IL-17 A, and IL-17 C mRNA after tofacitinib
   therapy is associated with clinical remission of RA [[92]22–[93]24],
   indicating that they can be biomarkers for assessing the response to
   JAKi therapy. The lower the multi-biomarker disease activity (MBDA)
   scores of patients with RA receiving tofacitinib treatment, the better
   their clinical outcomes [[94]25], which indicates that the
   musculoskeletal ultrasound (MSUS) can be a predictive tool for the
   response to JAKi. However, the efficacy of JAKi therapy is difficult to
   assess and predict.

   In our research, the age of the HCs was lower than that of RA patients
   in the group of sequencing and filtering. Recently, some evidence that
   RA might be related to age has been reported [[95]26]. The younger
   patients with RA exhibited higher clinical remission rates and lower
   radiological progression than older patients [[96]27]. However, the
   relationship between age and RA remained controversial. The prevalence
   of ACPA-negative among older populations with RA was higher, whereas
   there was no significant correlation with age in the ACPA-positive
   subgroups [[97]28]. Furthermore, the clinical remission timelines
   showed no statistically significant differences between older and
   younger patients with RA. These findings did not support the definitive
   relationship between age and RA. Consequently, age is unlikely to
   constitute a confounding variable and should not affect the outcomes of
   our experiment.

   In this study, 56 differentially expressed exomiRNAs were identified by
   NGS sequencing analysis, and exo-(miR-548ah-3p, miR-378 g, miR-27a-5p,
   and miR-30c-2-3p) were elevated in the PBMCs of MTX-resistance patients
   with RA. Their biological functions were found to regulate the
   activation and signal transduction of immune cells and the
   immune–inflammatory response via the JAK/STAT signaling pathway, the
   B-cell receptor pathway, the Toll-like receptor pathway, Fc gamma
   R-mediated phagocytosis, and the apoptosis-related pathway. In further
   research, we observed the changes in circulating exosomes from
   MTX-resistant patients with RA before and after three months of JAKi
   treatment to evaluate whether these exosomes could serve as predictive
   efficacy markers of JAKi.

   Previous studies have shown that exo-(miR-548ah-3p, miR-378 g, and
   miR-30c-2-3p) may be associated with the immune response and
   inflammation. MiR-378 g is one of the least studied members of the
   miR-378 family, and its function in RA has not yet been studied.
   Researchers found that miR-378 g promotes osteogenic differentiation by
   targeting nicotinamide N-methyltransferase (NNMT) during bone
   metabolism [[98]29]. In addition, miR-378 g regulates the Treg/Th17
   imbalance via CCL5 [[99]30], and exo-miR-378 g regulates macrophage
   polarization, T-cell activation, and inflammatory cytokine
   transcription [[100]31]. Numerous studies have found that miR-30c-2-3p
   regulated the activity of mitogen-activated protein kinase (MAPK)
   [[101]32] and the NF-κB pathway [[102]33] involved in inflammatory
   processes, apoptosis, and cell proliferation. Exo-miR-30c-2-3p can be
   captured and is highly expressed in plasma exosomes, serving as a
   biomarker for neurodegenerative diseases [[103]34]. Some studies have
   shown that exo-miR-30c-2-3p mediates the expression of AKT1S1 and
   EIF4B, which play significant roles in PI3K/AKT signaling [[104]35]. In
   this study, these exo-(miR-548ah-3p, miR-378 g, and miR-30c-2-3p)
   showed high expression only in the peripheral blood of patients with
   RA, suggesting poor correlation with clinical indicators and the
   prediction efficacy of JAKi.

   The involvement of miR-27a-5p in the progression of RA has not been
   reported previously. However, it is encapsulated in extracellular
   vesicles and participates in apoptosis and inflammation [[105]36,
   [106]37]. Some studies have shown that miR-27a-5p also inhibits M2-like
   macrophage polarization through the Ppm1l and PI3K/Akt pathways
   [[107]38, [108]39]. These studies indicate that exo-miR-27a-5p plays an
   essential role in the immune and inflammatory processes of different
   diseases and is an essential pathogenic mechanisms of RA. Based on the
   PPI network, protein tyrosine phosphatase N2 (PTPN2) may be a potential
   target of miR-27a-5p, which is a negative regulator of the T-cell
   receptor (TCR) and JAK/STAT signaling pathways [[109]40]. Furthermore,
   low expression of PTPN2 accelerates IL-6 production and enhances the
   T-cell-mediated immune response [[110]41, [111]42]. Based on these
   findings, we concluded that the JAK/STAT pathway may play a key role in
   the biological functions of miR-27a-5p, demonstrating its potential as
   a biomarker of JAKi therapy.

   In conclusion, this study demonstrated that exo-miR-548ah-3p,
   exo-miR-378 g, exo-miR-27a-5p, and exo-miR-30c-2-3p expressions were
   elevated in the PBMCs of MTX-resistant RA patients. Furthermore,
   circulating exo-miR-27a-5p may act as a novel biomarker in monitor
   clinical response of tofacitinib treatment. However, there are some
   limitations in our study, such as the limited sample size and lack of
   confirmation using histological samples, which should be incorporated
   into future research plans.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [112]Supplementary Material 1^ (238.4KB, tif)
   [113]Supplementary Material 2^ (5.8MB, tif)
   [114]Supplementary Material 3^ (13.5MB, tif)
   [115]Supplementary Material 4^ (15.6MB, tif)
   [116]Supplementary Material 5^ (18.5KB, docx)

Acknowledgements