Cellular miRNA-31 and miRNA-125b as predictors of remission and major disease improvement in patients with early rheumatoid arthritis

Abstract

Rheumatoid arthritis (RA) is an autoimmune disease that impairs quality of life, highlighting the need for novel biomarkers for early diagnosis and improve disease management. This study explored the cellular microRNAs (miRNAs) profile in early RA to identify potential predictors of therapeutic response. We measured a panel of 768 miRNAs in peripheral blood mononuclear cells (PBMCs) from two early RA cohorts (< 6 months), stratified by sustained SDAI/CDAI remission and non-response to conventional therapy over 12 months. Initial screening identified 13 differentially expressed miRNAs (≥ 1.5-fold change), which were validated in an independent cohort comprising 59 early RA patients and 54 healthy controls. MiRNA levels were quantified by qPCR, and predictive performance was assessed by receiver operating characteristic (ROC) curve. Among the 13 candidates, baseline levels of miRNA-31 and miRNA-125b predicted SDAI/CDAI remission at 6 months. Moreover, changes in their expression after 1 month of therapy predicted Boolean2.0 remission and achievement of major disease improvement (≥ 85% reduction in disease activity) at 6 months. Higher baseline expressions of both miRNAs were consistently observed in patients who achieved remission. These findings suggest that cellular expression of miRNA-31 and miRNA-125b may serve as promising prognostic biomarkers for 6-month remission in early RA.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-46990-0.

Introduction

Rheumatoid arthritis (RA) is a systemic autoimmune disease of unknown origin, characterized by inflammation of the synovial joints, progressive cartilage and bone destruction, and significant disability of affected patients¹. Genetic, epigenetic, environmental, and lifestyle factors, as well as alterations in the microbiome, play crucial roles in the development and progression of RA². Achieving optimal disease outcomes depends significantly on timely diagnosis and the prompt initiation of effective treatment². Despite recent advancements in therapeutic strategies for RA, a large proportion of patients still fail to achieve remission³. Therefore, the identification of novel biomarkers for early diagnosis and prediction of treatment response in patients with RA is needed⁴.

MicroRNAs (miRNAs) constitute a promising area of research in the identification of biomarkers for various diseases. MiRNAs represent class of small noncoding RNAs of ∼ 20–30 nucleotides in length, capable of modifying gene expression on an epigenetic level and thereby regulating a wide range of biological pathways, including the immune response⁵. They can be detected in a various tissues, biological materials or body fluids such as plasma or serum⁶. Altered expression of miRNAs has been observed in various immune-mediated inflammatory disorders, including RA, with their levels being influenced by pharmacological treatment⁷. Research indicates that miRNAs have the potential to differentiate between patients and healthy controls, associate with inflammatory pathways, and are linked with disease activity, progression, and treatment response in RA⁷. Thus, miRNAs hold promise as diagnostic and predictive biomarkers in the context of RA.

This study aims to expand on our previous analysis of cellular miRNA-125b in a cohort of patients with early RA, who were treatment-naïve and had less than six months of disease symptoms, including swelling in at least one small joint, and no alternative explanation for joint swelling by another disease^8,9. Here, we assess cellular miRNAs in patients and controls to identify a cluster of miRNAs that can more accurately predict the achievement of remission and major disease improvement after six months of treatment.

Results

Patients and baseline characteristics

The goal of the therapy was to achieve remission after six months of treatment. In addition, we assessed disease improvement at the six-month time point to explore the potential of miRNAs as predictive markers of therapeutic response. Remission was evaluated using three different indices—simplified disease activity index (SDAI), clinical disease activity index (CDAI), and the recently introduced Boolean2.0 criteria—in order to capture a comprehensive and clinically relevant spectrum of treatment responses, thereby enhancing the robustness and generalizability of our findings. Clinical definitions of remission according to all three indices, as well as definitions of disease improvement based on SDAI and CDAI, are provided in Supplementary Table 1A.

Table 1 summarizes the demographic and clinical characteristics of 59 treatment-naïve early RA patients and 55 healthy controls (HCs). The number of patients, along with their disease activity at baseline and after three and six months of follow-up, is provided in Supplementary Table 1B. Conventional synthetic disease modifying anti-rheumatic drugs (csDMARDs) were initiated in 55 patients. At the three-month follow-up, 48 patients were receiving glucocorticoids (GC) at a median daily dose of 5 mg of prednisone or equivalent. Additionally, three from 48 patients were treated exclusively with low dose GC due to liver toxicity or planned pregnancy.

Table 1.

Clinical characteristic of patients with early rheumatoid arthritis (early RA) and healthy controls (HCs).

		Early RA		HCs
Number		59		55
Sex, female/male		43/16		42/13
Age, years		54.93 ± 16.18		50.47 ± 15.39
Disease duration		< 6 months		–

Clinical characteristic	Baseline	3 months	6 months
RF IgM positivity (%)	52	NA	NA	NA
Anti-CCP positivity (%)	66	NA	NA	NA
SDAI	34.17 ± 17.71	9.75 ± 11.99	7.10 ± 7.94	NA
CDAI	40.68 ± 27.85	8.40 ± 10.97	6.57 ± 7.54	NA
CRP(mg/l)	12.71 (22.40)	3.38 (3.99)	2.69 (3.56)	1.00 (1.62)
ESR (mm/h)	28.00 (27.25)	15.50 (13)	14.00 (15)	NA

Treatment initiation	Baseline	3 months	6 months
Glucocorticoids (n)	46	48	38	–
Methotrexate (n)	32	49	48	–
Leflunomide (n)	0	1	1	–
Sulfasalazine (n)	2	5	3	–
Hydroxychloroquine (n)	2	2		–

Data are expressed as the mean ± SD or median (IQR) according to the distribution.

Anti-CCP, anti-citrullinated antibodies; CDAI, clinical disease activity index: CRP, C reactive protein: ESR, erythrocyte sedimentation rate; HCs, healthy controls; n, number of patients; NA, not analysed: RA, rheumatoid arthritis: RF, rheumatoid factor; SDAI, simplified disease activity index.

Prior to treatment initiation, over 80% of patients exhibited moderate to high disease activity, indicated by SDAI > 11, and CDAI > 10. After three months of therapy, more than 50% of patients achieved remission or demonstrated low disease activity (SDAI ≤ 11, and CDAI ≤ 10). Notably, clinical outcomes further improved at the six-month follow-up, with over 75% of patients achieving remission or low disease activity.

Explorative analysis of cellular miRNAs in early RA patients based on clinical response

For the exploratory phase of the study, we randomly selected four early RA patients who achieved sustained remission throughout the first year of treatment—specifically, they were in remission at the three, six, and 12-month assessments (responders, Rem)—and four early RA patients who consistently failed to respond to therapy, showing moderate to high disease activity at the same three, six, and 12-month assessments (non-responders, NR), as evaluated by the SDAI, CDAI, and Boolean 2.0 criteria.

We analysed the expression of 768 miRNAs in total RNA isolated from peripheral blood mononuclear cells (PBMCs) of treatment-naïve patients and compared the profiles between the groups (Fig. 1A). Among analysed cellular miRNAs, only 13 (miRNA-661, miRNA-10a, miRNA-23b, miRNA-27b, miRNA-31, miRNA-125b, miRNA-146a, miRNA-210, miRNA-342-3p, miRNA-452, miRNA-502-3p, miRNA-542-5p and miRNA-885-5p) were differentially expressed between responders and non-responders, exhibiting at least a 1.5-fold difference in expression (Fig. 1B).

Fig. 1.

Screening and selection of microRNAs (miRNAs) in peripheral blood mononuclear cells (PBMCs) from 4 early RA patients achieving sustained remission (Rem) in contrast to 4 early RA long-term non-responders (NR). (A) The expression heat map of 768 miRNAs using expression miRNA array x-fold data. (B) Baseline expression of thirteen selected miRNAs in remission group compared to NR group of patients. Threshold for selection was 1.5-fold (dotted line) diverse expression between groups.

The effect of treatment on the expression of selected cellular miRNAs

Following the initial screening, we compared the expressions of selected 13 miRNAs between all 59 treatment-naïve early RA patients at baseline (M0), as well as after one month (M1) and three months (M3) of therapy and compared to a cohort of 55 HCs. Our aim was to highlight differences between patients and HCs at each time point, rather than to assess longitudinal changes within individual patients from M0 to M1 or M3 to evaluate the effect of treatment on miRNA expression (Supplementary Fig. 1A–M). Baseline expressions of miRNA-31, miRNA-125b and miRNA-342-3p were significantly lower in RA patients compared to HCs (p < 0.01). However, their expression levels at M1 and M3 did not reach statistical significance (p > 0.05). The expression pattern of miRNA-125b was consistent with our previous findings⁸. Similarly, miRNA-10a, miRNA-23b, miRNA-27b, miRNA-210 and miRNA-452 also exhibited significantly lower baseline expression compared to HCs (p < 0.001); however, their levels remained unchanged following treatment initiation. The expression of miRNA-542-5p and miRNA-661 was comparable to that of HCs at baseline, showed a significant increase after one month of treatment initiation (p < 0.001 and p < 0.01, respectively), but returned to levels comparable with HCs after three months of therapy. The expression of miRNA-885-5p was significantly higher than in healthy controls both before and after treatment (p < 0.01), and therapy did not alter its levels. Likewise, baseline expression of miRNA-146a did not differ from HCs but significantly decreased after three months of treatment (p < 0.001). Finally, the baseline expression of cellular miRNA-502-3p did not significantly differ from HCs and remained unchanged throughout the treatment period.

Predictive value of differentially expressed miRNAs for treatment response

To assess the predictive potential of differentially expressed miRNA in achieving treatment target, we conducted receiver operating characteristic (ROC) curve analysis based on age-adjusted expression values and area under the curve (AUC) with 95% confidence interval (CI). Among the 13 miRNAs analysed, only miRNA-31 and miRNA-125b demonstrated the ability to distinguish patients who achieved remission within six months from those who did not.

Baseline expression of miRNA-31 was predictive of remission as defined by SDAI (AUC = 0.665 [95% CI: 0.521 – 0.809]; p = 0.033), but not by CDAI (AUC = 0.615 [95% CI: 0.463 – 0.767]; p = 0.141) or the Boolean2.0 criteria (AUC = 0.617 [95% CI: 0.469 – 0.766]; p = 0.132) (Fig. 2A). Similarly, baseline expression of miRNA-125b predicted remission according to SDAI (AUC = 0.682 [95% CI: 0.536 – 0.829]; p = 0.019), CDAI (AUC = 0.670 [95% CI: 0.523 – 0.816]; p = 0.031) and Boolean2.0 criteria (AUC = 0.661 [95% CI: 0.515 – 0.806]; p = 0.039) (Fig. 2A).

Fig. 2.

Receiver operating characteristic (ROC) curves illustrate the predictive value of miRNA-31 and miRNA-125b expression—individually and in combination—for remission achievement and major disease improvement after six months of anti-rheumatic therapy. (A) Baseline (M0) expression of miRNA-31 and miRNA-125b predicted achievement of remission by SDAI index. Baseline (M0) expression of miRNA-31did not predict remission achievement using CDAI index and Boolean2.0 score in contrary to miRNA-125b. Combination of initial expression (M0) of both miRNAs (miRNA-31 & miRNA-125b) predicted the remission achievement using SDAI and CDAI indexes and also Boolean2.0 score. (B) Furthermore, miRNA-31 (M1-M0) expression after one-month therapy predicted also remission achievement by Boolean2.0 score and major disease improvement by SDAI index. Additionally, change (M1-M0) in miRNA-125b expression predicted also major disease improvement by SDAI and CDAI indexes. Combination of miRNA (miRNA-31 & miRNA-125b) changed expressions (M1-M0) after one-month therapy also predicted significantly achievement of remission or major disease improvement. MiRNA-31 and miRNA-125b are better predictors of remission or major disease improvement than clinical inflammatory markers (CRP) or disease activity indices (SDAI, CDAI, Boolean 2.0) at baseline (M0), prior to therapy initiation. Disease activity, improvements and response at sixth month was evaluated by simplified disease activity index (SDAI), clinical disease activity index (CDAI) and Boolean2.0 score (Supplementary Table 1A, B). 23 patients achieved remission using SDAI and Boolean2. score and 22 patients achieved remission using CDAI index. Major disease improvement was achieved by 28 patients using SDAI and 33 patients using CDAI index at six months of treatment.

In line with these findings, we also calculated the changes in expression levels of both miRNAs during the first month of therapy (M1–M0) and used the same dataset for ROC analyses to evaluate their predictive and correlational potential. Changes in miRNA-31 expression, and to a lesser extent changes in miRNA-125b expression, after one month of treatment predicted remission at six months based on the Boolean2.0 criteria (miRNA-31: AUC = 0.734 [95% CI: 0.598–0.863]; p = 0.004; miRNA-125b: AUC = 0.659 [95% CI: 0.503–0.816]; p = 0.052). Furthermore, changes in expression of both miRNAs predicted major disease improvement (≥ 85% reduction in disease activity) according to SDAI (miRNA-31: AUC = 0.720 [95% CI: 0.583–0.857]; p = 0.006; miRNA-125b: AUC = 0.779 [95% CI: 0.646–0.912]; p = 0.001). Interestingly, changes in miRNA-125b, but not miRNA-31, expression also predicted major disease improvement according to CDAI (miRNA-31: AUC = 0.638 [95% CI: 0.481 – 0.794]; p = 0.095; miRNA-125b: AUC = 0.720 [95% CI: 0.572–0.868]; p = 0.008) at six months (Fig. 2B).

To further enhance predictive accuracy, we examined whether combining miRNA-31 and miRNA-125b expression levels could improve outcome prediction. The combination of baseline expression levels significantly predicted remission according to SDAI (AUC = 0.694 [95% CI: 0.547–0.842]; p = 0.012), CDAI (AUC = 0.663 [95% CI: 0.513–0.814]; p = 0.037) and Boolean2.0 criteria (AUC = 0.668 [95% CI: 0.522–0.814]; p = 0.031) (Fig. 2A). Moreover, combining changes in expression of both miRNAs after one month significantly predicted remission at six months according to Boolean2.0 criteria (AUC = 0.728 [95% CI: 0.592–0.863]; p = 0.005). These combinations also significantly predicted major disease improvement based on SDAI (AUC = 0.777 [95% CI: 0.650–0.904]; p = 0.001) and CDAI (AUC = 0.717 [95% CI: 0.571–0.863]; p = 0.009) (Fig. 2B). Besides, expression of miRNA-31 and miRNA-125b are better predictors of remission or major disease improvement than clinical inflammatory marker (CRP) or disease activity indices (SDAI, CDAI, Boolean 2.0) at baseline (M0), prior to therapy initiation (Fig. 2A, B). We also performed ROC analysis in a selected subgroup of early RA patients who initiated therapy with GC plus MTX (n = 39) (Supplementary Fig. 2). In this cohort, baseline expression levels (M0) of both miRNAs, when analysed in combination, improved the prediction of remission at six months of therapy according to SDAI (AUC = 0.750 [95% CI: 0.591–0.909]; p = 009, CDAI (AUC = 0.745 [95% CI: 0.585–0.904]; p = 0.010) and Boolean2.0 criteria (AUC = 0.711 [95% CI: 0.540–0.882]; p = 0.025) (Supplementary Fig. 2A). Combining changes in expression of both miRNAs after one month (M1-M0) significantly predicted major disease improvement based on SDAI (AUC = 0.769 [95% CI: 0.650–0.904]; p = 0.010) but not CDAI (AUC = 0.687 [95% CI: 0.500–0.874]; p = 0.092) (Supplementary Fig. 2B). Furthermore, these combinations also significantly predicted remission using Boolean2.0 score (AUC = 0.785 [95% CI: 0.625–0.946]; p = 0.005).

Cellular miRNAs expression and disease activity

In addition to the ROC analysis, we also assessed expression levels of miRNA-31 and miRNA-125b across patient groups stratified by disease activity status at six months of therapy (remission, low, moderate, and high disease activity). Patients who achieved remission exhibited significantly higher expression levels of both miRNA-31 and miRNA-125b during therapy compared to those with low, moderate or high disease activity (Fig. 3A, B). We also assessed this comparison in a subgroup of patients who initiated treatment with GC plus MTX (Supplementary Fig. 3A, B). Similarly, both miRNAs had higher baseline expression in patients who achieved remission compared to those with low, moderate and high disease activity.

Fig. 3.

(A, B) Patients who responded to anti-rheumatic therapy exhibited higher expressions of miRNA-31 and miRNA-125b compared to non-responders. Treatment response was defined as remission achieved at six months, assessed by simplified disease activity index (SDAI; 23 patients in remission, 36 non-responders), clinical disease activity index (CDAI; 22 patients in remission, 37 non-responders) and Boolean2.0 criteria (23 patients in remission, 36 non-responders). Expressions of both miRNAs were determined at baseline (M0), and after one (M1) and three (M3) months of treatment. Data are shown as median with interquartile range. *p < 0.05; **p < 0.01, ns, non-significant.

We further examined the potential influence of demographic parameters on miRNA-31 and miRNA-125b expression. Both miRNAs demonstrated a significant association with age; therefore, age-adjusted correlation (partial correlation) analyses were performed to account for this potential confounder. Subsequently, we examined whether the expression levels or the changes during the first month in expression of either cellular miRNA, using the same dataset, were associated with disease activity.

Both miRNA-31 and miRNA-125b expression levels were strongly positively correlated with each other (r = 0.530, p < 0.0001) (Supplementary Table 2). Baseline expression of miRNA-31 negatively correlated with baseline disease activity as measured by SDAI (r = − 0.357, p = 0.006), but not CDAI (r = − 0.229, p = 0.084). Additionally, higher baseline miRNA-31 expression levels were positively correlated with changes in disease activity in both SDAI (r = 0.404, p = 0.002) and CDAI (r = 0.288, p = 0.030) after six months of therapy (Supplementary Table 2). Similarly, changes in miRNA-125b expression after one month of treatment (M1-M0) positively correlated with SDAI (r = 0.312, p = 0.023) and CDAI (r = 0.327, p = 0.017) improvement after three months of treatment (Supplementary Table 2). Moreover, changes in miRNA-31 and miRNA-125b expressions after one month (M1-M0) were negatively correlated with disease activity according to SDAI (miRNA-31: r = − 0.368, p = 0.007; miRNA-125b: r = − 0.319, p = 0.021) and CDAI (miRNA-31: r = − 0.318, p = 0.022; miRNA-125b: r = − 0.331, p = 0.017) at six months, and positively correlated with the degree of SDAI (miRNA-31: r = 0.487, p < 0.001; miRNA-125b: r = 0.532, p < 0.0001) and CDAI (miRNA-31: r = 0.332, p = 0.016; miRNA-125b: r = 471, p < 0.001) disease activity improvement at the six month follow-up (Supplementary Table 2).

Discussion

In this study, we investigated dynamic changes in cellular miRNA expression in treatment-naïve early RA patients and their association with clinical outcomes following conventional therapy. We identified specific cellular miRNAs, particularly miRNA-31 and miRNA-125b that distinguished RA patients achieving sustained remission from those with persistent disease activity. These finding highlight their potential as prognostic biomarkers for treatment response in early RA.

Our exploratory analysis revealed several miRNAs with distinct expression patterns differentiating responders from non-responders. Notably, miRNA-31, miRNA-125b and miRNA-342-3p were significantly downregulated at baseline in RA patients, but their expression levels increased following treatment, reaching levels similar to those in HCs. This pattern suggests that their regulatory roles are modulated by therapeutic interventions. Reduced expression of miRNA-31 and miRNA-125b has also been reported in other immune-mediated inflammatory diseases, such as juvenile idiopathic arthritis, psoriasis, and also in established RA^10–12, supporting their potential anti-inflammatory role.

In contrast, miRNAs such as miRNA-10a or miRNA-452 were differentially expressed in RA patients at baseline, but did not significantly change with therapy, suggesting that they may reflect disease-specific features independent of treatment effects. Meanwhile, other miRNAs (e.g., miRNA-542-5p, miRNA-661, miRNA-885-5p and miRNA-146a) showed stable baseline expression compared to controls but showed variable modulation following treatment. These patterns may reflect complex interactions with immune pathways not directly targeted by conventional therapy.

To evaluate the predictive potential of these miRNAs, we performed ROC analyses and demonstrated that baseline expression of miRNA-31 and miRNA-125b could predict remission and major disease improvement at six months. Importantly, early changes in their expression (after one month of therapy) also correlated with clinical outcomes, suggesting their utility as early markers of treatment response. These findings extend our previous results, where baseline cellular expression of miRNA-125b predicted therapeutic response at three months⁸.

Correlation analyses explored the clinical relevance of these miRNAs. While miRNA-31 and miRNA-125b were positively correlated with each other, only miRNA-31 showed a significant negative association with baseline disease activity determined by SDAI. Notably, changes in the expression of both miRNAs after one month were negatively associated with disease activity scores at six months of therapy. Higher baseline expression levels were observed in patients who ultimately achieved remission, and early changes in expression were positively associated with clinical improvement at follow-up. These results suggest that higher initial expression of miRNA-31 and miRNA-125b, as well as their changes in expression after one month in PBMCs, may predict a more favourable therapeutic response.

Functionally, miRNA-31 and miRNA-125b regulate diverse gene networks involved in inflammation, including the nuclear factor kappa B (NF-κB) signalling pathway. For example, loss of miRNA-31 activates NF-kB inducing kinase (NIK) triggering non-canonical NF-kB signalling in adult T-cell leukaemia¹⁴. It also regulates migration of pro-inflammatory Th1 cells¹⁵, attenuates cartilage damage and suppress the release of pro-inflammatory cytokines (IL-1β, TNF) in osteoarthritic models¹⁶ and reduces pyroptosis in fibroblast-like synoviocytes¹¹. These findings further support the anti-inflammatory and chondroprotective role of miRNA-31, highlighting its potential therapeutic relevance in musculoskeletal diseases. Similarly, miRNA-125b modulates macrophage-driven inflammation by targeting CCL4 chemokine and the negative costimulatory molecule B7-H4^20,21, although its effect varies depending on the immune cell type and disease context^17–19.

Another potential explanation for the upregulation of miRNA-31 and miRNA-125b following anti-rheumatic therapy could be an indirect effect of GC and csDMARDs through suppression of NF-κB signalling, rather than a direct influence on miRNA transcription. Although no direct regulation of miRNA-31 by GC has been reported, dexamethasone has been shown to transiently increase miRNA-125b expression during the treatment of B-cell multiple myeloma cells²². Similarly, no direct effect of methotrexate on either miRNA has been described. Therefore, it is more likely that the increased expression of these miRNAs reflects an overall positive therapeutic response, rather than a drug-specific effect^23–26.

The ability of cellular miRNA-31 and miRNA-125b to predict remission in early RA aligns with their previously observed prognostic value in other diseases, particularly in various types of cancer^27–30. One limitation of our study is that we were unable to replicate our findings in circulating miRNA levels. Specifically, miRNA-31 was undetectable, and miRNA-125b did not predict clinical outcomes⁸. This discrepancy may be attributed to the heterogeneous origin of circulating miRNAs, which are released from various cell types and tissues¹³. In contrast, cellular miRNAs from PBMCs may better reflect disease-specific patterns in RA³¹. Similar results from the ROC/AUC analyses and from comparisons of miRNA expression between remission and non-remission groups were also observed in the subgroup of patients who received initial GC plus MTX treatment. However, this subgroup analysis has certain limitations, including the smaller sample size, which reduces statistical power. In addition, the GC plus MTX treatment approach does not fully reflect the variability and broader therapeutic strategies commonly applied in routine clinical practice. We also recognize that the exploratory phase included only four patients per group, which limited statistical power and increased the risk of false discoveries. This phase was intended as a preliminary step to identify potential biomarkers, and the findings were interpreted cautiously. It should also be noted that the Rem/NR patient groups in the exploratory phase were based on 12-month clinical outcomes, which may limit the generalizability of the biomarker findings for predicting shorter-term responses. Importantly, promising results were validated in a larger cohort to improve robustness. Nonetheless, further studies with larger and independent RA cohorts are needed to confirm these results and fully determine their clinical utility.

Conclusion

Extending our previous findings, this study demonstrates that higher baseline expression of cellular miRNA-31 and miRNA-125b is associated with remission at six months in patients with early RA. These results suggest a potential role for these miRNAs as prognostic indicators of early treatment response. Nevertheless, since the present analysis is based solely on internal validation, these findings require confirmation in larger, independent cohorts of patients with early RA before they can be considered for clinical implementation or to inform treatment decisions.

Methods

Patients and healthy subjects

This observational, case–control study comprised a total of 114 subjects. A cohort of 59 patients (43 females/16 males; mean age 55 ± 17 years) with early treatment-naïve RA (disease duration < 6 months; fulfilling the 2010 American College of Rheumatology (ACR)/European Alliance of Associations for Rheumatology (EULAR) classification criteria for RA³⁰ were compared to a cohort of 55 age- and sex-matched HCs (42 females/13 males; mean age 50 ± 15 years). Healthy subjects had no known history of systemic autoimmune disease or cancer. Disease activity, including remission, was assessed at baseline, three and six months after treatment initiation using SDAI, CDAI and Boolean2.0 score for remission (Supplementary Table 1B). For screening purposes, patients were categorized as NR if they experienced with long-term moderate to high disease activity and as responders if they achieved sustained remission (Rem). Additionally, we assessed major, moderate, mild, and no improvement of disease according to SDAI and CDAI at three and six months of therapy (Supplementary Table 1B). The treatment target was defined as achieving remission after six months. In line with our previous publication⁸, this study was conducted using the same cohort of patients and healthy controls.

RNA isolation and quantification

Isolation of PBMC from the whole blood was performed using standard Ficoll density gradient centrifugation (Greiner Bio-One, Leipzig, Germany). The resulting cell pellets were snap-frozen and stored at – 80 °C until further analysis. Total RNA was extracted using the miRNeasy Mini Kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. Any residual DNA was removed with the RNase-Free DNase Set (Qiagen), as recommended by the manufacturer⁸.

First, we performed exploratory screening of miRNA expression from stored PBMC samples in a group of 4 patients with stable remission (12 months, Rem) and a group of 4 patients with long-lasting moderate to high disease activity (12 months, NR) by TaqMan Array Human MicroRNA Cards (Thermo Fisher Scientific, Waltham, MA USA). Screening of miRNAs in PBMC samples started with reverse transcription using Megaplex™ RT Primers, Human Pool A + B v2.1 (Thermo Fisher Scientific) with TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific), followed by preamplification reaction using Megaplex PreAmp Primers (Thermo Fisher Scientific) and TaqMan® PreAmp Master Mix (Thermo Fisher Scientific), finished with quantitative polymerase chain reation (qPCR) reaction using TaqMan Array Human MicroRNA A + B Cards (Thermo Fisher Scientific) with TaqMan Universal PCR Master Mix, no AmpErase UNG (Thermo Fisher Scientific) according to the company´s protocol.

Based on the baseline screening, 13 miRNAs with at least 1.5-fold difference in expression between the two patient groups were selected and validated in a cohort comprising 59 early RA patients and 55 HCs. The isolated total RNA was first reverse transcribed using dedicated TaqMan microRNA Assays (Thermo Fisher Scientific) with TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific) followed by qPCR reaction using specific TaqMan microRNA Assays (Thermo Fisher Scientific) with TaqMan Universal PCR Master Mix, no AmpErase UNG (Thermo Fisher Scientific). All reverse transcription reactions were conducted using the Thermocycler MyCycler (Biorad, Hercules, CA, USA), while quantification reactions were performed using the 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The average value of all expressed miRNAs in TaqMan Array Human MicroRNA A + B Cards was used for normalization of screened miRNAs. Small nucleolar miRNA RNU44 was used for normalization of selected cellular miRNAs.

Statistical analysis

D´Agostino-Pearson omnibus K2, were used to assess normality of our data. Kruskal–Wallis test (with Dunn´s correction test) and Mann-Witney U test were performed for the comparison of differences between groups. For correlation analyses, Spearman´s correlation test was used and age-adjusted associations were assessed using partial Spearman’s correlation. Prediction of remission achievement was analysed by ROC curve based on the age-adjusted expression values and AUC with 95% CI was calculated. We employed a binary logistic regression model to integrate the expression levels of miRNA-31 and miRNA-125b as a composite biomarker for prediction. This combined probability model was then applied in ROC analysis to evaluate its ability to distinguish patients who achieved remission or exhibited major clinical improvement from those with low, moderate, or high disease activity, or with only mild, moderate, or no improvement. Data were shown as the mean with standard deviation (SD) or median with interquartile range (IQR). Results with P-values < 0.05 were considered as statistically significant. For statistical analyses IBM SPSS Statistics (IBM Corp., Armonk, NY) and GraphPad Prism, version 6 (GraphPad Software, San Diego, CA, USA) was used.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

V.H. performed isolation of miRNAs from PBMCs and plasma samples, qPCR analyses and drafted the manuscript. H.M., M.F., J.V. and L.Š. were involved in enrolling the patients and made clinical assessments. K.P., T.K. and L.V. contributed to data preparation and interpretation. M.K. carried out the statistical analysis. L.Š. made substantial contributions to study concept and design, and revised the final draft of the manuscript. M.F., J.V., K.P. and L.Š. revised the manuscript critically for important intellectual content. All authors read and approved the final version of the manuscript.

Funding

This work was supported by projects SVV 260 638, MHCR 023728 and BBMRI.cz, reg. no. LM2023033.

Data availability

Data will be available upon reasonable request to the corresponding author. All data relevant to the study are included in the article.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

The study was approved by the local ethics committee (no. 3294/2012) of the Institute of Rheumatology in Prague, Czech Republic. The study was approved by the local ethics committee of the Institute of Rheumatology in Prague, Czech Republic, and is registered under ethical approval number 3294/2012. Written informed consent was obtained from all participants. All experiments were performed in accordance with relevant guidelines and regulations.