Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis Authors: Ifat Vigiser, Yoav Zeevi, Yoav Piura, Hadar Kolb, Maya Golan, Arnon Karni, Keren Regev Published: 2026 ABSTRACT BackgroundReliable biomarkers of inflammatory activity in multiple sclerosis (MS) are needed to complement clinical and MRI assessments. Circulating microRNAs (miRNAs) are promising candidates due to their serum stability and involvement in MS-related biological pathways.ObjectivesTo evaluate whether selected serum miRNAs can discriminate relapse from remission in treatment-naive patients with relapsing-remitting MS (RRMS).MethodsIn this prospective paired study, newly diagnosed, untreated RRMS patients were sampled during clinical relapse and subsequent remission. A targeted miRNA panel was quantified using quantitative PCR. Differential expressions were assessed using paired statistical testing with multiple comparison correction. Diagnostic performance was evaluated using ROC analysis and multivariate logistic regression. Pathway enrichment analysis was performed using Ingenuity Pathway Analysis.ResultsTwenty patients (40 paired samples) were included. Eight miRNAs were differentially expressed between relapse and remission after Benjamini–Hochberg correction. hsa-miR-122-5p showed the largest fold change (fold change = 3.0; adjusted p = 0.043). A three-miRNA panel (hsa-miR-122-5p, hsa-miR-15b-5p, hsa-miR-30e-5p) demonstrated high discriminative accuracy (area under the curve = 0.88; 95% CI 0.79–0.96). Pathway analysis highlighted inflammatory, metabolic, and hypoxia-related processes.ConclusionsCirculating miRNAs distinguished relapse from remission in treatment-naive RRMS patients, suggesting their potential as complementary biomarkers of disease activity. These findings should be considered exploratory and require validation in an independent, larger cohort. Background Multiple sclerosis (MS) is a chronic, immune-mediated disorder of the central nervous system (CNS) characterized by inflammation, demyelination, and neuroaxonal degeneration. 1 Accurate detection of inflammatory disease activity, particularly clinical relapses, is crucial for guiding treatment decisions. Relapse diagnosis is based on the appearance of new or worsening neurological symptoms lasting at least 24 h, in the absence of fever or infection. 2 Although MRI is an essential tool for identifying new lesions, lesion formation does not always coincide with symptom onset, and inflammatory activity may go undetected. 3 These limitations highlight the need for minimally invasive biomarkers that can reliably detect current inflammatory activity with high sensitivity and specificity. Serum neurofilament light chain (NfL) is the most widely studied fluid biomarker in MS. The NfL levels rise in association with relapses, MRI lesion activity, and long-term disability progression, reflecting axonal injury. However, NfL lacks disease specificity, being elevated in other neurological and systemic conditions, and is influenced by age, body mass index, and comorbidities.4,5 Additional biomarkers such as glial fibrillary acidic protein (GFAP) and chitinase-3-like-1 have shown associations with progressive disease and astrocytic pathology but face similar limitations of specificity and assay variability.5,6 Consequently, there is a need for complementary biomarkers that capture upstream immunological processes rather than downstream structural damage. MicroRNAs (miRNAs) are small, noncoding RNAs (∼19–23 nucleotides) that regulate gene expression posttranscriptionally, typically by binding to complementary sequences in target messenger RNAs, leading to degradation or translational repression. 7 They are key regulators of immune cell activation, cytokine production, blood–brain barrier integrity, mitochondrial dynamics, and glial–neuronal interactions.8–11 In MS, altered miRNA expression has been documented in active demyelinating lesions, cerebrospinal fluid, and peripheral blood, supporting their relevance to disease mechanisms.12–14 Circulating miRNAs are stable in serum and plasma due to protection within extracellular vesicles (EVs) or association with protein complexes, making them attractive biomarker candidates.7,15 Several studies have identified distinct circulating miRNA profiles in MS compared with healthy individuals and other neurological diseases, as well as differences between active and stable disease states.12–18 For example, hsa-miR-484 has been linked to mitochondrial fission via Fis1 regulation, a process that may contribute to oxidative stress in active lesions. 19 Members of the miR-320 family have been implicated in leukocyte trafficking and blood–brain barrier regulation,20,21 while hsa-miR-182-5p and hsa-miR-122-5p have been associated with modulation of inflammatory signaling and astrocyte microglia crosstalk.22–24 More recent studies suggest that EV-associated miRNAs may play an active role in immune dysregulation in MS, rather than serving solely as passive circulating markers. As circulating serum miRNAs may partially originate from EV-associated cargo, De Rosa et al. demonstrated that EVs released by regulatory T cells shuttle hsa-miR-142-3p to conventional T cells, thereby suppressing effector T-cell activity. In patients with relapsing-remitting MS (RRMS), reduced levels of EV-associated hsa-miR-142-3p were associated with impaired immune suppressive function and correlated with disability scores and gadolinium-enhancing lesions, supporting a potential role for EV-associated miRNAs in disease activity and loss of immune tolerance. Torres-Iglesias et al. further demonstrated that EV-derived miRNAs and EV subpopulations are associated with clinically relevant measures of disease activity and treatment response in MS. Specifically, CD3+ and neuronal-derived (L1CAM+) EVs correlated with relapses, MRI lesion activity, and progression of motor disability, while CD20+ EVs were associated with cognitive impairment. In addition, differential expressions of several EV-derived miRNAs distinguished treatment responders from nonresponders, and hsa-miR-186-5p expression correlated negatively with brain atrophy. These observations further highlight the potential relevance of circulating miRNAs to disease monitoring and treatment response in MS.25,26 Despite this promise, the translation of miRNA biomarkers into clinical relapse detection remains limited. Many previous studies have been cross-sectional, with heterogeneous patient populations, and enrolled participants already receiving disease-modifying therapies, which can alter circulating miRNA profiles.12,16,27 Moreover, variability in sample handling, detection platforms, and normalization strategies has hindered reproducibility across studies. Paired-sample designs in treatment-naive patients, particularly at the time of initial diagnosis, can help reduce these confounders and enable robust within-patient comparisons of relapse versus remission. In this study, we prospectively evaluated a biologically informed panel of MS-relevant circulating miRNAs (i.e., hsa-miR-326, hsa-miR-183-5p, hsa-miR-122-5p, hsa-miR-196b-5p, hsa-miR-182-5p, hsa-miR-423-3p, hsa-let-7d-3p, hsa-miR-15b-5p, hsa-miR-103a-3p, hsa-miR-191-5p, hsa-miR-484, hsa-miR-140-5p, hsa-miR-320a, hsa-miR-320c, hsa-miR-320b, hsa-miR-142-5p, hsa-miR-486-5p, hsa-miR-337-3p, hsa-miR-199a-5p, hsa-miR-30e-5p, hsa-miR-590-5p) in paired relapse and remission serum samples from treatment-naive patients at first diagnosis. The miRNA panel was defined using a hypothesis-driven approach, based on prior MS literature and biological relevance to key disease mechanisms, including inflammation, immune regulation, and neurodegeneration.14–18 Importantly, no high-throughput screening or large-scale discovery panel was performed in this cohort, and all miRNAs were prespecified prior to data collection and analysis, rather than selected from a larger tested pool. The final multivariate model was subsequently derived from this predefined panel through statistical model selection procedures within the study cohort. To support the rationale for panel composition, a summary of key studies is provided in Supplementary Table 1. The corresponding miRNA assay identifiers and technical assay information are provided in Supplementary Table 4. In addition, a subset of miRNAs (hsa-miR-103a-3p, hsa-let-7d-3p, hsa-miR-423-3p, and hsa-miR-191-5p) was included as candidate reference miRNAs based on their reported stability in serum and common use for normalization and were not considered disease-associated targets. The detection and quantification methods used for measuring miRNA expression in this study build upon large-scale studies that initially employed high-throughput techniques, which were subsequently refined through various validation stages.11–13 We aimed to: (i) identify miRNAs with significant differential expressions between relapse and remission; (ii) assess the diagnostic performance of individual and combined miRNAs for relapse detection; and (iii) explore associated inflammatory pathways to better understand their mechanistic role in MS. Methods Study population and design We prospectively enrolled patients presenting with a first clinical relapse and a newly confirmed diagnosis of MS, according to the 2017 McDonald criteria, 28 at the Tel Aviv Sourasky Medical Center. Exclusion criteria included infection or vaccination within the preceding three months, presence of systemic comorbidities, or pregnancy. All patients were untreated with disease-modifying therapy at the time of enrollment. Relapse-phase samples were obtained before initiation of corticosteroid therapy, and all patients subsequently received high-dose intravenous methylprednisolone for relapse treatment. Follow-up (“remission”) samples were collected three months later, at least 30 days after the last corticosteroid dose. The study protocol was approved by the Institutional Review Board (approval No. 0194-13-TLV), and all participants provided written informed consent. Clinical assessment At each visit, patients underwent a standardized neurological examination and Expanded Disability Status Scale (EDSS) scoring by a certified examiner. Demographic and clinical data, including disease duration, relapse characteristics, and MRI findings, were recorded. Sample collection and RNA extraction Peripheral blood samples were allowed to clot for 30 min at room temperature, centrifuged at 3300 rpm for 7 min, and serum aliquots were stored at −80°C within 2 h of collection. Serum was transported to an external laboratory under controlled temperature conditions. Serum samples were thawed on ice and centrifuged at 3000 × g for 5 min at 4°C. An aliquot of 200 μL per sample was transferred to FluidX tubes, and 60 μL of Buffer RPL containing carrier RNA (1 μg per 60 μL Buffer RPL) and RNA spike-in template mixture was added. Samples were mixed for 1 min and incubated for 7 min at room temperature, followed by the addition of 20-μL Buffer RPP (rapid protein percipitation). Total RNA was extracted using the miRNeasy Serum/Plasma Advanced Kit (QIAGEN, Hilden, Germany) in an automated 96-well, bead-based format. RNA was eluted in a final volume of 50 μL. Because circulating serum miRNA yields are typically low and fragmented, conventional measures of RNA purity and integrity were not considered reliable; therefore, spike-in controls were used to monitor extraction and reverse transcription efficiency. Reverse transcription and quantitative PCR RNA was reverse-transcribed using the miRCURY LNA RT Kit (QIAGEN), and complementary DNA (cDNA) was diluted 1:50. Two microliters of RNA were reverse transcribed in 10 μL reactions. The resulting cDNA was analyzed in 10-μL PCR reactions using the miRCURY LNA SYBR Green master mix and the miRNA Ready-to-Use PCR panels. Each miRNA was assayed in technical duplicates. Each assay included a “no template” control; undetected signals were assigned Cq = 37. QIAGEN spike-in controls (UniSp2, UniSp4, and UniSp5) were used to monitor RNA extraction efficiency, while UniSp6 was used to assess cDNA synthesis efficiency. Expression values were normalized to the mean Cq of all miRNAs consistently detected across all samples. Only miRNAs detected in all samples were included in the normalization factor, and undetected values were excluded from its calculation. Negative controls, excluding template from the reverse transcription reaction, were included and processed in parallel. Paired samples were processed on the same plate to minimize batch effects. Amplification was performed on a LightCycler® 480 Real-Time PCR System (Roche) in 384-well plates. Amplification curves were analyzed using Roche LC software for Cq determination (second-derivative method) and melting curve analysis. Ingenuity Pathway Analysis Differentially expressed miRNAs were analyzed in QIAGEN Ingenuity pathway analysis (IPA) using the eight miRNAs that remained significantly differentially expressed after Benjamini–Hochberg correction in the paired relapse-remission comparison to identify enriched canonical pathways and upstream regulators. Right-tailed Fisher's exact test determined the statistical significance of pathway enrichment. Upstream regulator predictions integrated known interactions between transcription factors, cytokines, miRNAs, and target mRNAs in the Ingenuity Knowledge Base. 29 Statistical analysis Based on previously reported effect size and variability for paired measurements, a power calculation was performed for the planned paired analysis. The calculation indicated that a minimum of 16 pairs is required to achieve the desired statistical power. Expression levels during relapse and remission were compared for each miRNA using paired t-tests. Undetected values were imputed as one unit lower than the smallest observed Cq across the dataset. To adjust for multiple testing, p-values were corrected using the Benjamini–Hochberg false discovery rate (FDR) procedure. ROC curves were generated for each miRNA, and corresponding area under the curve (AUC) values with 95% confidence intervals were calculated. A forward stepwise logistic regression model with AIC minimization was used to identify the optimal combination of miRNAs for distinguishing relapse from remission. To further assess model stability and reduce the risk of overfitting, a pair-matched conditional logistic regression model with bootstrap-based internal validation (1000 iterations) was additionally applied. Model performance was assessed using the AUC. Spearman's rank correlation was used to determine associations between miRNA expression and disability measures (EDSS and MSSS). To address the nonlinear nature of EDSS, we additionally calculated MSSS, integrating disability and disease duration using the global reference table. 30 MSSS was calculated at the remission timepoint (Timepoint B) to minimize relapse-related fluctuations. Exploratory analyses were performed to assess potential sex-related differences in miRNA expression at the remission timepoint, with p-values corrected for multiple comparisons using the Benjamini–Hochberg FDR procedure. A sample size estimation for a paired comparison indicated that approximately 16 pairs would provide 80% power to detect a moderate-to-large within-subject standardized effect size (dz ≈ 0.70) at a two-sided α level of 0.05. Given the exploratory nature of the study, this estimation was used as a general reference point rather than a strict design constraint. Analyses were performed in R (www.r-project.org) and Stata/IC version 14. Results Patient characteristics The study cohort included 20 patients with newly diagnosed RRMS. The mean age at diagnosis was 34.1 years (SD ± 10.49; median 32, IQR 26–40), and 50% were female. Mean EDSS at diagnosis was 2.1 (SD ± 1.79; median 1.5, IQR 1.0–2.5). Intrathecal oligoclonal bands were detected in 16/20 patients (80%). Gadolinium-enhancing lesions were observed in 14/20 patients (70%) on baseline brain MRI. Lesion distribution included supratentorial lesions in 16/20 (80%), infratentorial lesions in 14/20 (70%), and spinal cord lesions in 17/20 (85%) (Table 1). Table 1.Baseline demographic and clinical characteristics of the study cohort.Demographic characteristicsn = 20Mean age of diagnosis (SD), years34.1 (10.49)Females, n (%)10 (50) Clinical characteristics Mean EDSS score at diagnosis (SD)*2.1 (1.79)Patients with positive intrathecal OCB, n (%)16 (80)MRI profile at diagnosis, n (%)Patients with gadolinium-enhancing lesions14 (70)Patients with supra-tentorial lesions16 (80)Patients with infra-tentorial lesions14 (70)Patients with spinal cord lesions17 (85)OCB, oligoclonal band. Baseline demographic and clinical characteristics of the study cohort. OCB, oligoclonal band. Differential expressions of miRNAs Of the 21 miRNAs analyzed, 12 showed nominal significance (raw p < 0.05), including hsa-miR-122-5p, hsa-miR-182-5p, hsa-miR-484, hsa-miR-320a, hsa-miR-320b, hsa-miR-320c, hsa-miR-423-3p, hsa-miR-15b-5p, hsa-miR-326, hsa-miR-30e-5p, hsa-miR-183-5p, and hsa-let-7d-3p. After Benjamini–Hochberg correction, eight miRNAs remained significant (adjusted p < 0.05). The largest fold change was observed for hsa-miR-122-5p (fold change = 3.0; raw p = 0.006; adj p = 0.043), while hsa-miR-182-5p demonstrated the lowest p-value (fold change = 1.9; raw p = 0.004; adj p = 0.015). Box plots illustrating expression changes between relapse and remission are provided in Figure 1. Boxplots of eight microRNAs show significant differences between relapse and remission in multiple sclerosis. Boxplots depicting log2-transformed normalized serum expression levels of eight microRNAs in patients with multiple sclerosis during relapse and remission. These microRNAs represent the subset that remained statistically significant following Benjamini–Hochberg (BH) correction for multiple comparisons (adjusted p < 0.05). Data are shown for paired serum samples obtained from 20 patients during relapse and remission. The plot illustrates both the magnitude and direction of expression changes between the two clinical states. Diagnostic performance of individual miRNAs Eleven individual miRNAs demonstrated AUC values ranging from 0.65 to 0.82. The strongest single predictor was hsa-miR-484 (AUC = 0.823; 95% CI 0.71–0.94; p = 0.004), followed by hsa-miR-320a-3p (AUC = 0.758; 95% CI 0.63–0.88; p = 0.011), hsa-miR-122-5p (AUC = 0.753; 95% CI 0.62–0.88; p = 0.010), and hsa-miR-182-5p (AUC = 0.753; 95% CI 0.62–0.88; p = 0.010) (Table 2, Figure 2). Receiver operating characteristic (ROC) curves of individual microRNAs (miRNAs) distinguishing relapse from remission in MS. Receiver operating characteristic curves illustrating the discriminatory performance of 11 serum miRNAs with statistically significant area under the curve (AUC) values for distinguishing relapse from remission in patients with multiple sclerosis. AUC values and corresponding 95% confidence intervals are presented for each miRNA. Data are based on paired serum samples obtained from 20 patients during relapse and remission. Table 2.Mean expression levels of selected circulating microRNAs during relapse and remission.miRNAMean expression (relapse)Relapse SDMean expression (remission)Remission SDFold-changeDirection of change (remission)T test p valueBH adjusted p valueAUCp value (AUC)hsa-miR-182-5p−6.7250.869−5.8260.9861.864Upregulated in Remission0.004 0 . 043 0.7520.011hsa-miR-484−0.6360.246−0.3580.2251.213Upregulated in Remission0.0040.0430.8230.004hsa-miR-122-5p−1.3381.9030.2521.3053.012Upregulated in Remission0.0060.0430.7520.010hsa-miR-320a-3p1.5760.4001.9910.4711.333Upregulated in Remission0.0100.0500.7570.011hsa-miR-320c−0.3520.4430.0330.4311.306Upregulated in Remission0.0120.0500.7450.016hsa-miR-30e-5p−0.3670.411−0.0370.2981.257Upregulated in Remission0.0170.0500.7380.014hsa-miR-326−4.2620.584−3.7210.6801.455Upregulated in Remission0.0180.0500.7330.019hsa-let-7d-3p−2.3690.337−2.1100.4291.197Upregulated in Remission0.0190.0500.6850.050hsa-miR-423-3p−1.6600.353−1.9220.2830.834Downregulated in Remission0.0250.0530.7030.023hsa-miR-15b-5p0.6301.1561.2940.2861.585Upregulated in Remission0.0250.0530.6550.069hsa-miR-320b−0.9690.421−0.6620.4611.237Upregulated in Remission0.0340.0660.7000.044hsa-miR-183-5p−5.7730.832−5.2340.8441.453Upregulated in Remission0.0410.0720.6720.058hsa-miR-486-5p3.7940.7564.2470.4961.369Upregulated in Remission0.0510.0830.6670.041hsa-miR-337-3p−6.6381.289−6.0161.1251.539Upregulated in Remission0.0680.1010.6670.124hsa-miR-103a-3p1.3810.3511.5650.2481.136Upregulated in Remission0.1000.1340.6500.072hsa-miR-196b-5p−7.6190.898−7.0730.9601.460Upregulated in Remission0.1060.1340.6720.078hsa-miR-191-5p0.2790.1860.3570.1371.056Upregulated in Remission0.1080.1340.5880.147hsa-miR-199a-5p−3.7680.752−4.0660.7110.814Downregulated in Remission0.2230.2610.6250.204hsa-miR-590-5p−2.5350.379−2.4220.3241.081Upregulated in Remission0.3100.3420.6100.313hsa-miR-140-5p−2.9580.518−3.0650.4010.929Downregulated in Remission0.4090.4290.5950.460hsa-miR-142-5p0.7580.4870.7780.4371.014Upregulated in Remission0.8890.8890.5070.886The mean expression levels and standard deviations (SDs) of a predefined panel of circulating microRNAs (miRNAs) measured during relapse and remission in treatment-naive patients with relapsing-remitting multiple sclerosis (RRMS). Fold-change was calculated as the ratio between mean expression levels in remission and relapse. Direction of change refers to relative upregulation or downregulation during remission compared with relapse. Statistical significance was determined using paired t-tests, and multiple comparisons were adjusted using the Benjamini–Hochberg (BH) procedure. BH-adjusted p values < 0.05 were considered statistically significant. Receiver operating characteristic (ROC) analysis was performed for each miRNA, and the corresponding area under the curve (AUC) and p values are presented. Mean expression levels of selected circulating microRNAs during relapse and remission. The mean expression levels and standard deviations (SDs) of a predefined panel of circulating microRNAs (miRNAs) measured during relapse and remission in treatment-naive patients with relapsing-remitting multiple sclerosis (RRMS). Fold-change was calculated as the ratio between mean expression levels in remission and relapse. Direction of change refers to relative upregulation or downregulation during remission compared with relapse. Statistical significance was determined using paired t-tests, and multiple comparisons were adjusted using the Benjamini–Hochberg (BH) procedure. BH-adjusted p values < 0.05 were considered statistically significant. Receiver operating characteristic (ROC) analysis was performed for each miRNA, and the corresponding area under the curve (AUC) and p values are presented. Multivariate model for relapse detection Forward stepwise logistic regression identified an optimal three-miRNA model comprising hsa-miR-122-5p, hsa-miR-15b-5p, and hsa-miR-30e-5p. This model achieved an AUC of 0.88 (95% CI 0.79–0.96; p < 0.001). ROC curves for combined miRNA predictors are shown in Figure 3. Although hsa-miR-15b-5p showed modest individual discriminatory performance, it contributed to the final multivariate model; its individual ROC curve is presented in Supplementary Figure 1. Additional analyses using a pair-matched conditional logistic regression model with bootstrap-based internal validation (1000 iterations) demonstrated comparable performance (AUC = 0.86; bootstrap median AUC = 0.87, 95% CI: 0.74–0.95). Receiver operating characteristic (ROC) curve of the multivariate three-microRNA (miRNA) model distinguishing relapse from remission in multiple sclerosis. Univariate analysis – disability No significant correlations were observed between disability measures (EDSS or MSSS) and the expression of any studied miRNA after correction for multiple testing (all p > 0.05). Exploratory analysis – sex differences Exploratory analyses assessing potential sex-related differences in miRNA expression at the remission timepoint (B) revealed no significant differences between males and females after correction for multiple testing. Canonical pathways and upstream regulator analysis Ingenuity pathway analysis of the eight miRNAs that remained significantly differentially expressed after Benjamini–Hochberg correction in the paired relapse-remission comparison revealed significant enrichment of immune and inflammatory pathways. The top five included ID1 signaling (−log(p) = 13.3; ratio = 0.0735), HEY1 signaling (−log(p) = 10.8; ratio = 0.0745), p53 signaling (−log(p) = 10.4; ratio = 0.101), IL-4 and IL-13 signaling (−log(p) = 9.9; ratio = 0.0901), and RAR activation (−log(p) = 9.63; ratio = 0.0369) (Supplementary Table 1). The IPA Upstream Regulator Analysis identified hsa-miR-122-5p (p = 5.27 × 10−118), hsa-miR-21-5p (p = 1.10 × 10−89), hsa-miR-21 (p = 3.85 × 10−42), hsa-miR-199a-5p (p = 3.82 × 10−22), and TNF (p = 8.82 × 10−22) as the top predicted upstream regulators (Supplementary Table 2). Discussion In this prospective, paired-sample study of treatment-naive patients with newly diagnosed RRMS, we identified a set of circulating miRNAs whose expression differed significantly between relapse and remission. Three of these, hsa-miR-122-5p, hsa-miR-15b-5p, and hsa-miR-30e-5p, formed a multivariate panel that achieved high accuracy in discriminating relapse from remission, with an AUC of 0.88, outperforming individual miRNAs. To further evaluate the stability of these findings, we performed additional analyses using a pair-matched conditional logistic regression model with bootstrap-based internal validation, which yielded comparable performance estimates. Pathway analysis linked these miRNAs to processes central to MS pathophysiology, including immune cell trafficking, cytokine signaling, mitochondrial dynamics, and hypoxia response. Dysregulated miRNAs identified in this study have been previously associated with biological pathways relevant to inflammatory activity in MS. hsa-miR-122-5p, predominantly expressed in the liver, has been associated with lipid metabolism and systemic inflammatory responses; its downregulation during relapse may reflect shifts in energy metabolism and immune activation.23,31 hsa-miR-15b-5p has been associated with the regulation of apoptosis and immune cell proliferation,32,33 while hsa-miR-30e-5p has been implicated in autophagy and inflammatory signaling in multiple cell types, including microglia. 34 The observed downregulation of hsa-miR-182-5p and hsa-miR-484 during relapse is also consistent with prior studies linking these miRNAs to mitochondrial homeostasis and neuroinflammatory control.20,24 Together, these observations support the possibility that relapse-associated changes in circulating miRNAs may reflect not only local CNS inflammation but also systemic immunometabolism adaptations. The NfL is currently the most widely studied fluid biomarker for MS activity, showing robust associations with relapses, MRI lesion formation, and disability progression.4,5 However, NfL reflects axonal injury, a downstream consequence of inflammation, rather than the upstream immune processes driving relapses. Moreover, NfL can be elevated in other neurological conditions, reducing disease specificity. In contrast, miRNAs may offer complementary information by capturing early immune activation and metabolic shifts preceding or accompanying tissue damage. The three-miRNA model identified in our study achieved an AUC of 0.88, which is numerically higher than AUC values previously reported for serum NfL in relapse detection across independent cohorts.35–37 However, these comparisons are indirect and should be interpreted cautiously given differences in study design, patient populations, and assay methodologies. Nevertheless, these observations support further evaluation of combined miRNA-protein biomarker strategies to enhance the sensitivity and specificity of relapse detection in MS. Our study design, using paired relapse and remission samples from treatment-naive patients at first diagnosis, minimized interindividual variability and avoided confounding effects of prior DMT (disease modifying treatments) exposure. The selection of miRNAs was biologically informed, focusing on candidates with previous evidence of involvement in MS or related inflammatory pathways. We applied correction for multiple comparisons and used multivariate modeling to evaluate the combined discriminatory performance of selected miRNAs. Additional bootstrap-based internal validation demonstrated comparable model performance; however, the multivariate model should still be considered exploratory and requires validation in independent cohorts. We further explored the relationship between miRNA expression and clinical disability. No significant associations were observed between miRNA expression and EDSS. To address the known limitations of EDSS as a nonlinear scale, we additionally evaluated MSSS. However, given the early disease stage of this treatment-naive cohort at initial diagnosis, MSSS values likely exhibited limited variability, and no significant associations were identified. Nonetheless, several limitations should be acknowledged. First, the sample size, while adequate to detect moderate effect sizes, limits the ability to assess performance across subgroups (e.g., relapse phenotype, MRI activity patterns). In addition, although exploratory analyses did not identify significant sex-related differences in miRNA expression, the sample size limits the ability to detect more subtle sex-specific effects. Second, our cohort was drawn from a single center, which may limit the generalizability of our findings. Third, the study focused on a predefined panel of miRNAs rather than an unbiased discovery approach; while this increases interpretability, it may miss other relevant markers. In addition, some circulating miRNAs included in the model, such as miR-122-5p, are not CNS-specific and may be influenced by systemic or non-neurological conditions, potentially limiting biological specificity. Although pathway analysis supported the biological plausibility of the identified miRNAs, no mechanistic validation was performed in the present study. Functional studies using cell-specific experimental systems and in vitro modulation of miRNA expression will be required to determine whether the observed associations reflect causal biological pathways or secondary disease-related phenomena. Such mechanistic investigations may further clarify the role of these miRNAs in MS-associated immune regulation and neuroinflammatory processes. The study did not include an independent validation cohort, and the identified miRNA signatures and multivariate model should therefore be further evaluated in larger, independent cohorts. At the same time, the paired within-subject design, comparing relapse and remission samples from the same individuals, reduces interindividual variability and strengthens internal validity. This study was exploratory and not powered based on a formal a priori sample size calculation. While the sample size is expected to be sufficient to detect moderate to large within subject effects, smaller effects may have been missed. These results support the potential utility of miRNA panels as complementary biomarkers for detecting inflammatory activity in MS. Beyond relapse detection, longitudinal monitoring of circulating miRNAs could provide insights into treatment response, subclinical disease activity, and the mechanism of progression. Future studies should include larger, multicenter cohorts, incorporate direct comparisons with established biomarkers such as NfL and GFAP, and explore integration of miRNA data with advanced imaging and immunophenotyping. Validation in real-world clinical settings will be critical to determine feasibility, cost-effectiveness, and added value over current monitoring strategies. In conclusion, specific circulating miRNAs, particularly the combination of hsa-miR-122-5p, hsa-miR-15b-5p, and hsa-miR-30e-5p, can distinguish relapse from remission in treatment-naive RRMS at first diagnosis, with high accuracy and strong biological relevance. These findings provide a basis for further development of miRNA-based tools to complement existing clinical and imaging assessments in MS. Supplemental Material sj-docx-1-mso-10.1177_20552173261471723 - Supplemental material for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis Supplemental material, sj-docx-1-mso-10.1177_20552173261471723 for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis by Ifat Vigiser, Yoav Zeevi, Yoav Piura, Hadar Kolb, Maya Golan, Arnon Karni and Keren Regev in Multiple Sclerosis Journal – Experimental, Translational and Clinical sj-docx-2-mso-10.1177_20552173261471723 - Supplemental material for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis Supplemental material, sj-docx-2-mso-10.1177_20552173261471723 for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis by Ifat Vigiser, Yoav Zeevi, Yoav Piura, Hadar Kolb, Maya Golan, Arnon Karni and Keren Regev in Multiple Sclerosis Journal – Experimental, Translational and Clinical sj-docx-3-mso-10.1177_20552173261471723 - Supplemental material for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis Supplemental material, sj-docx-3-mso-10.1177_20552173261471723 for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis by Ifat Vigiser, Yoav Zeevi, Yoav Piura, Hadar Kolb, Maya Golan, Arnon Karni and Keren Regev in Multiple Sclerosis Journal – Experimental, Translational and Clinical sj-docx-4-mso-10.1177_20552173261471723 - Supplemental material for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis Supplemental material, sj-docx-4-mso-10.1177_20552173261471723 for Circulating microRNA expression profiles as biomarkers of disease activity in multiple sclerosis by Ifat Vigiser, Yoav Zeevi, Yoav Piura, Hadar Kolb, Maya Golan, Arnon Karni and Keren Regev in Multiple Sclerosis Journal – Experimental, Translational and Clinical