Внеклеточная везиклерозовая микроРНК как биомаркер для терапической реакции при множественном склерозе.

Introduction

Multiple sclerosis (MS) is an immune-mediated inflammatory disease affecting the CNS. An estimated 2.8 million people worldwide suffer from MS, with the disease leading to varying degrees of disability that may profoundly affect their quality of life. 1 Around 50% of people with MS are diagnosed before the age of 30, a time at which they are beginning important life projects. As such, it is crucial that such individuals receive optimal support and the most appropriate treatments to control disease activity and prevent the associated disability.

Several disease-modifying treatments (DMTs) have been developed for MS over the past 20 years. 2 However, because the natural course of the disease is unpredictable, the benefit of each treatment in individual patients remains unknown, thus complicating clinical decision making considerably. Indeed, some patients initiate treatment but continue to experience relapses, new lesions in the CNS, or worsening disability. When this occurs, the response to treatment is considered suboptimal and the DMT must be switched to a more effective therapy, generally with a higher risk of serious adverse events. 3 These decisions are currently made on a case-by-case basis with the onset of lesions in the CNS and increased disability. 4 In this context, a biomarker that anticipates treatment response or treatment failure could help clinicians change the DMT before irreversible damage occurs.

In recent years, blood extracellular vesicles (EVs) have gained importance as potential biomarkers in MS. 5-8 These vesicles are small lipid bilayer spheres containing different molecules [DNA, messenger RNA, microRNA (miRNA), proteins, and lipids] and are involved in cell-to-cell communication in both physiologic and pathologic conditions. 5 Patients with MS show significant differences in EVs compared with healthy controls. 6-9 In addition, these EVs have also been found to correlate with relapses and with the appearance of new white matter lesions. 6-8 Some MS treatments have been described as modulating the levels of blood EVs, 10 although this field remains poorly explored. In addition, EVs contain large quantities of miRNA 11 regulating a multitude of biological processes involved in MS pathogenesis and have been found to be involved in resistance to some therapies. 12

In light of the above, in this study, we sought to analyze whether levels, size, and miRNAs derived from circulating EVs could provide relevant information on biological functions occurring in treatment response and whether EVs may also serve as a biomarker to guide early therapeutic response in patients with MS.

Methods

Study Design

This is an observational, longitudinal, prospective, and clinical study including patients with a confirmed diagnosis of relapsing-remitting MS, according to the McDonald criteria, 13 from May 2021 to June 2023. Patients meeting the following criteria were recruited for the study: (1) Inclusion criteria: men or women aged 18 years and older with relapsing-remitting MS initiating or changing ongoing DMT. (2) Exclusion criteria: patients with progressive MS, current drug dependence, and a severe concomitant disease; pregnant or breastfeeding patients; and patients participating in pharmacologic treatment trials. Informed consent was obtained from each participant. Using G power 3.1 software, it was estimated that for a power of 80% for “No Evidence of Disease Activity (NEDA)” and a significance value of 0.05, the sample size should be 80 patients with MS.

Demographic and Clinical Data

The following demographic data and clinical characteristics were collected from participants: sex, age, disease duration (time from the first onset of MS symptoms to the date of the baseline visit), baseline Expanded Disability Status Scale (EDSS) and Symbol Digit Modalities Test (SDMT) scores, baseline brain volume, and current treatments received. In those participants who had received previous DMT, the required washout period for each treatment was set according to standard clinical practice.

Visits and Outcome Measures

The study consisted of 3 visits per patient for correlative studies at specific time points, including PRE-treatment (before treatment initiation) and POST-treatment (3 and 12 months after treatment onset). The visits were those applied in routine clinical practice and, therefore, did not involve additional discomfort for the patient. The variables collected at each visit are presented in Figure 1A . These variables include specific clinical assessments established by general consensus among MS neurologists on the variables that should be evaluated to monitor response to DMT at 12 months. 14 All clinical assessments and measurements of outcomes were exclusively performed by experienced neurologists.

Flow Diagram and EV Characterization

(A) Flow diagram of the study illustrating the procedures applied to patients with MS and their distributions into groups. (B) Electron microscope image of an EV smaller than 200 nm. (C) Western blot image demonstrating the positivity of specific markers CD9, CD81, CD63, and Alix in the EV membrane from MOG + , CD20 + and CD3 + , and L1CAM + . Negative control samples are from plasma. The gel image has been cropped. (D) Size dispersion and concentration of MOG + , CD20 + and CD3 + , and L1CAM + EV samples analyzed by NTA. 9HPT = 9-Hole Peg Test; EDSS = Expanded Disability Status Scale; EVs = extracellular vesicles; kDa = kilodalton; NTA = nanoparticle-tracking analysis; MS = multiple sclerosis; NEDA = No Evidence of Disease Activity.

Ethics Approval and Consent to Participate

This study was approved by La Paz University Hospital Research Ethics Committee (PI-4675). All data management was conducted within the framework of Spanish Law 14/2007 of 03 July on Biomedical Research, ensuring the confidentiality of all personal data. The study was performed in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). Informed consent was obtained from each participant.

Treatment Response

Response to treatment was evaluated at 12-month follow-up in accordance with the ECTRIMS guidelines 4 and NEDA-3 criteria, defined as patients with no relapses, no MRI activity, and no disability progression in the past 12 months. Conversely, patients who had experienced at least one of the 3 clinical conditions were categorized as evidence of disease activity (EDA)-3. NEDA-3 refers to patients who respond to DMT compared with those with EDA-3 status who do not (responders and non responders, respectively). However, in our in-depth study, we have gone one step further by including NEDA-4, defined as patients with NEDA-3 status with no sign of cognitive progression. By contrast, EDA-4 refers to patients who have experienced NEDA-3 or EDA-3 and cognitive progression. NEDA-5 refers to patients identified as NEDA-4 with brain atrophy ≤0.4% in the past 12 months. EDA-5 refers to patients who have experienced NEDA-4 or EDA-4 and brain atrophy.

Blood Samples and EV Isolation

Blood samples were collected at PRE-treatment and POST-treatment (3 months after treatment initiation) in 9-mL EDTA tubes. These samples were centrifuged at 1,800 rpm for 15 minutes to obtain cell-free plasma and stored at −80°C until analysis. For EV isolation, approximately 1 mL of plasma per patient sample contained in a cryovial was thawed rapidly in a 37°C water bath. The plasma was then transferred to a 1.5mL Eppendorf tube and centrifuged at room temperature for 20 minutes at 2,000g. For total circulating EV isolation, we used a precipitation method (ExoQuick EV isolation kit; System Biosciences). For cell origin–based isolation (EVs derived from neurons, oligodendrocytes, and B and T cells), we used a two-step isolation procedure for different cell-derived EV populations involving a combination of precipitation and immune isolation. For precipitation, we used a precipitation method (ExoQuick EV isolation kit; System Biosciences), followed by immune isolation with biotinylated antibodies against EV surface markers to isolate subpopulations. Primary biotinylated anti-L1CAM antibody (Thermo Fisher Scientific) was used to obtain enriched blood EVs of neuronal origin, biotinylated anti-MOG antibody (R&D System) for those from oligodendrocytes, biotinylated anti-CD20 antibody (Merck Millipore, Germany) for B cells, and biotinylated anti-CD3 antibody (Merck Millipore, Germany) for T cells. Streptavidin agarose beads (Thermo Fisher Scientific) were used to precipitate EVs. 6 The EVs contained in this resuspension were stored at −80°C until further use.

EV Characterization, Concentration, and Size Assessment

For verification of purified EVs, vesicles from L1CAM + , MOG + , CD20 + , and CD3 + were thawed on ice. These EVs were characterized regarding the following: (1) EV morphology, analyzed by transmission electron microscopy (JEOL JEM1010), as described previously by our group 6 ( Figure 1B ); (2) presence of specific EV markers analyzed by Western blot (WB) using anti-CD9 (Thermo Fisher Scientific), anti-CD81 (1:250, Abcam, United Kingdom), anti-CD63 (1:250, Abcam, United Kingdom), and anti-Alix (1:250, Cell Signal) antibodies, and anti-albumin as purity control, followed by goat anti-mouse or anti-rabbit horseradish peroxidase antibodies (1:750, Invitrogen) ( Figure 1C ); blots were visualized using Pierce enhanced chemiluminescence (Thermo Fisher Scientific) and an UVITEC Cambridge imaging system; (3) EV size and concentration, determined by nanoparticle-tracking analysis (NTA) using the NanoSight NS500 nanoparticle analyzer (Malvern Instruments, United Kingdom), which is equipped with fast video capture and particle-tracking software, following the manufacturer's instructions ( Figure 1D ). For each individual patient, triplicate measurements of EV levels and sizes were obtained using NTA, and their average was calculated for use in the subsequent analyses.

miRNA Expression Profiling

We profiled the differentially expressed EV miRNA content in responders and non responders at 3 months after treatment using bulk miRNA sequencing for total circulating EVs. Thus, total RNA was extracted from the EVs using the miRNeasy Serum/Plasma Advanced Kit (Qiagen). On the basis of RNA quality, 10 ng of total RNA from each sample was used for library preparation using the SMARTer smRNA-Seq Kit (Illumina). Samples were sequenced individually on an Illumina platform (Illumina) following the manufacturer's recommendations. The libraries were validated by checking the size, purity, and concentration using an Agilent bioanalyzer (Agilent Technologies). The libraries were then pooled in equimolar amounts and sequenced using an Illumina HiSeq 2,500 instrument to generate 101 base reads. The quality of reads was assessed with FastQC (Babraham Bioinformatics, United Kingdom), and reads were trimmed and quality-filtered with Cutadapt v4.7. Reads passing quality control were aligned to miRNA sequences retrieved from miRbase v22 using mirDeep2. miRNA values are normalized, allowing for a better comparison between study groups. Differential expression analysis was performed with DESeq2 software, and miRNAs were considered differentially expressed with a |foldchange|≤0.5 or |foldchange|≥2 and p value ≤ 0.05.

Early Treatment Response Model

A score based on the CD3 + , CD20 + , L1CAM + , and MOG + levels and size, as well as miRNA levels at PRE-treatment and POST-treatment points, was calculated using backward stepwise logistic regression with the Wald method for an early therapeutic response in patients with MS.

Statistics

The statistical analysis was supervised by the Biostatistics Unit of the La Paz University Hospital using Statistical Package for the Social Sciences 23.0 (IBM). Categorical variables were described as percentages, and proportions between groups were compared using the χ 2 test; the Fisher exact test was used for dichotomous variables. Continuous variables were expressed as mean (SD) or median (interquartile range). The Student t test and analysis of vairance (ANOVA) with Bonferroni post hoc correction were used for multiple comparisons of normally distributed data. Kruskal-Wallis or Mann-Whitney U tests were used for the comparison of non-normally distributed data sets. A principal component analysis (PCA) was conducted using the experimental datasheet including the differentially expressed miRNAs found in patients with NEDA-3, NEDA-4, and NEDA-5. The correlations between EV levels and between miRNAs were analyzed using Pearson r correlations and linear regression models. To anticipate the therapeutic response, backward stepwise logistic regression using the Wald method, including all variables differentially expressed between NEDA-3 and EDA-3, NEDA-4 and EDA-4, and NEDA-5 and EDA-5, respectively, was conducted. For the development of the multivariate model, we first evaluated the univariate relationship between the variables of interest (EV levels and sizes, and miRNA levels) and NEDA/EDA. Subsequently, we selected those variables with statistically significant associations and applied the backward Wald method to obtain the final model. Using the final model obtained from this analysis, a receiver operating characteristic (ROC) curve was generated to determine the cutoff point with maximum sensitivity and specificity, thus allowing for the calculation of a therapeutic response score. p Values are represented as follows: *, p < 0.05; **, p < 0.01; ***, p < 0.001; these p values were considered significant.

Data Availability

Anonymized data not published within this article will be made available by request from any qualified investigator.

Results

The study enrolled 80 patients with relapsing-remitting MS starting a new DMT ( Figure 1A ). Of these, 49 patients were considered responders (NEDA-3) and 31 were non responders (EDA-3). Patient characteristics are summarized in the Table .

Table Demographic and Clinical Data for the Study Participants Responders (n = 49) non responders (n = 31) Cohen D Demographics  Women, n (%) 25 (51) 18 (58.1) 0.22  Age, y 44.06 (9.03) 43.82 (11.23) 0.10 Clinical data  Time from diagnosis, mo 123.53 (120.48) 153.3 (137.01) −0,12  Baseline EDSS score 1.89 (2.07) 2 (2.29) −0,03  Baseline SDMT score 43.15 (13.94) 47 (12.51) −0,30  Brain volume (cc) 1,136.18 (145.72) 1,109.01 (147.39) 0.21 Treatments received  Natalizumab, n (%) 12 (24.5) 6 (19.4)  Teriflunomide, n (%) 2 (4.1) 3 (9.7)  Interferon, n (%) 6 (12.2) 0 (0)  Dimethyl fumarate, n (%) 7 (14.3) 6 (19.4)  Ocrelizumab, n (%) 7 (14.2) 3 (9.7)  Siponimod, n (%) 1 (2.0) 0 (0)  Cladribine, n (%) 14 (28.6) 11 (35.5) Abbreviations: EDSS = Expanded Disability Status Scale; N = number; SDMT = Symbol Digit Modalities Test. All values are mean (SD) unless otherwise noted. Mann–Whitney U test for continuous variables and Fisher's exact test for categorical variables were employed to determine statistically significant differences between groups.

Demographic and Clinical Data for the Study Participants

Abbreviations: EDSS = Expanded Disability Status Scale; N = number; SDMT = Symbol Digit Modalities Test.

All values are mean (SD) unless otherwise noted. Mann–Whitney U test for continuous variables and Fisher's exact test for categorical variables were employed to determine statistically significant differences between groups.

EV Characterization

EVs were characterized for size, morphology, and presence of specific EV marker profiles in their membrane. EV samples showed the typical morphology of lipid bilayer spheres with a size of approximately 200 nm by transmission electron microscope and less than 300 nm by NTA ( Figure 1B ). Positivity for the EV-specific tetraspanins CD9, CD63, and CD81, as well as Alix, allowed for a robust characterization of the EV samples by WB. The WB images and NTA histograms in Figure 1 , C and D , are representative of all EVs recovered from plasma samples. In the WB analysis, significant differences in the tetraspanin expression patterns (CD9, CD81, CD63) and Alix levels were obtained between different cell origins. These differences may arise because of the cellular origin of the EVs because various cell types are known to package EVs with distinct molecular compositions. Similarly, Alix expression, involved in EV biogenesis, may vary depending on the cellular origin and specific biogenesis pathways active in different cell types.

Levels and Size of EVs Reflect Treatment Response in Clinical Settings

To assess whether the pharmacologic washout period was sufficient to avoid affecting the levels and sizes of EVs, a baseline analysis was conducted comparing these parameters between naïve patients and those previously treated with DMTs. We did not find statistically significant differences in the baseline variables between both groups (eTable 1). However, this does not imply that they are entirely equivalent; rather, under the study conditions, no relevant differences were observed. 15 In fact, while significant differences can be observed, certain variables are at the threshold of significance ( p = 0.051 and 0.084).

No significant differences were found in PRE-treatment levels or size of EVs among patients with relapse compared with those without relapse or in those with EDSS progression. However, significantly higher levels and larger CD3 + EVs appeared in patients with new lesions on MRI (5.13 × 10 10 ± 1.05 × 10 11 particles/mL; 202.59 ± 163.32 nm) compared with those with no MRI activity (0.8 × 10 10 ± 2.42 × 10 10 particles/mL; 163.32 ± 37.64 nm; p = 0.023 and p = 0.031), thus leading to a decrease in the levels of these EVs in patients with NEDA-3 (0.8 × 10 10 ± 2.4 ×10 10 particles/mL) compared with those with evidence of disease activity (EDA-3; 2.39 × 10 10 ± 6.7 × 10 10 particles/mL; p = 0.017). Similarly, lower levels of L1CAM + EVs were found in patients with NEDA-3 (0.50 × 10 10 ± 9.9 × 10 9 particles/mL) compared with those with EDA-3 (0.84 × 10 10 ± 1.70 × 10 10 particles/mL; p = 0.05; Figure 2, A and B ).

PRE-Treatment Levels and Size of Immune and Nervous System EVs

(A) PRE-treatment EV levels in patients with relapse activity, new MRI findings, EDSS progression, and NEDA-3. (B) PRE-treatment EV size in patients with relapse activity, new MRI findings, EDSS progression, and NEDA-3. (C) PRE-treatment EV levels in patients with cognitive progression, NEDA-4, brain atrophy, and NEDA-5. (D) PRE-treatment EV size related to cognitive progression, NEDA-4, brain atrophy, and NEDA-5. The PRE-treatment analysis corresponds to the baseline study of EVs. ANOVA with Bonferroni post hoc correction was used for normally distributed data, and Mann-Whitney U tests were used for non-normally distributed data. EDSS = Expanded Disability Status Scale; EVs = extracellular vesicles; NEDA = No Evidence of Disease Activity.

Higher PRE-treatment CD3 + EV levels were found in patients with cognitive progression (0.69 × 10 10 ± 1.40 × 10 10 particles/mL) compared with stable patients (0.23 × 10 10 ± 0.63 ×10 10 particles/mL; p = 0.035), thus leading to lower levels and smaller EVs in patients with NEDA-4 (0.45 × 10 10 particles/mL, 164 ± 31.63 nm) compared with those with EDA-4 (1.15 × 10 10 particles/mL, 191.20 ± 53.78 nm; p = 0.001 and p = 0.035). When analyzing NEDA-4, we also found lower levels of L1CAM + EVs in patients with NEDA-4 (0.10 × 10 10 ± 0.22 × 10 10 particles/mL) compared with patients with EDA-4 (0.74 × 10 10 ± 1.52 × 10 10 particles/mL; Figure 2C ). These levels were also lower in patients with NEDA-5 (0.017 × 10 10 ± 0.010 × 10 10 particles/mL) than in those with EDA-5 (0.63 × 10 10 ± 1.10 × 10 10 particles/mL; p = 0.05; Figure 2C ).

Three months after treatment onset (POST-treatment), lower MOG + EV levels were observed in patients with EDSS progression (0.22 × 10 10 ± 0.12 × 10 10 particles/mL) compared with stable patients (0.62 × 10 10 ± 0.97 × 10 10 particles/mL; p = 0.039; Figure 3A ) and larger L1CAM + EVs were found in patients with new MRI lesions (209.96 ± 39.35 nm) compared with those with no lesions (187.54 ± 45.47 nm; Figure 3B ). Smaller CD20 + EVs were found in patients with cognitive progression (168.87 ± 28.18 nm) than in stable patients (199.30 ± 29.76 nm; p = 0.014; Figure 3D ). No significant differences were observed in any other variables of treatment response.

POST-Treatment Levels and Size of Immune and Nervous System EVs

(A) POST-treatment EV levels in patients with relapse activity, new MRI findings, EDSS progression, and NEDA-3. (B) POST-treatment EV size in patients with relapse activity, new MRI findings, EDSS progression, and NEDA-3. (C) POST-treatment EV levels in patients with cognitive progression, NEDA-4, brain atrophy, and NEDA-5. (D) POST-treatment EV size related to cognitive progression, NEDA-4, brain atrophy, and NEDA-5. POST-treatment analysis corresponds to the study of EVs conducted at 3 months. ANOVA with Bonferroni post hoc correction was used for normally distributed data, and Mann-Whitney U tests were used for non-normally distributed data. EDSS = Expanded Disability Status Scale; EVs = extracellular vesicles; NEDA = No Evidence of Disease Activity.

EV-Derived miRNAs Are Expressed Differentially Among Responders and Non responders

To verify that the effect of previous treatments on miRNAs had disappeared with the drug washout period, we compared miRNA expression between naive and non-naive patients in the PRE-treatment phase. We did not find statistically significant differences in the baseline variables between both groups. However, this does not imply that they are entirely equivalent; rather, under the study conditions, no relevant differences were observed 15 (eTable 2). Moreover, to verify that the miRNA expression of this study is not modified by age, a correlation analysis between miRNA expression and age of each individual patient was performed (eTable 3), and no significant correlation was observed for the selected miRNAs.

To understand the role of EV miRNA expression in treatment response, we performed bulk miRNA sequencing in which 180 miRNAs were identified among patients in the study. Unsupervised analysis resulted in 26 dysregulated miRNAs that reflect clinical parameters and reveal treatment response. The differential expression of miRNA levels in responders and non responders is displayed in eFigure 1. Downregulation of miR-184 and let-7c-5p was observed in those patients who suffered a relapse during the follow-up year, while miR-28-3p, miR-326, miR-4433b-3p, miR-328-3p, miR-484, miR-101-3p, and miR-92b-3p were downregulated in nonrelapsing MS patients. Patients with no abnormal MRI showed lower levels of miR-877-5p, miR-12136, and miR-4433b-3p while miR-98-3p, miR-98-5p, and miR-1-3 were less expressed in patients with new MRI findings. Decreased levels of miR-28-3p and miR-326 were found in patients with EDSS stability during follow-up while let-7c-5p and miR-184 were downregulated in patients who showed disease progression. Some of the miRNAs differentially expressed in one of these variables remain in patients with NEDA-3, including miR-28-3p, and those associated with EDA-3, including MiR-1-3, and miR-98-5p (eFigure 1A). Progression in cognitive impairment was associated with a repression of miR-144-5p and miR-660-5p while miR-98-3p was found to be lower in stable patients. Among these miRNAs, miR-660-5p, miR-222-3p and miR-144-5p were downregulated in patients with EDA-4 (eFigure 1B). Brain atrophy was associated with a downregulation of miR-125b-5p, miR-574-5p, and miR-98-5p, whereas downregulation of miR-24-3p, miR-23a-3p, miR-146a-5p, and miR-1260b was not related to brain volume atrophy. Within this miRNA, miR-23a-3p remained associated with NEDA-5 and miR-574 with EDA-5 (eFigure 1C).

In addition, miR-126-5p, miR-326, miR-4433b, and miR-28-3p were the most downregulated miRNAs in patients with NEDA-3, NEDA-4, and NEDA-5, whereas miR-660, miR98-5p, miR-574, and miR-144-5p were downregulated in patients with EDA-3 as well as those with EDA-4 and EDA-5 ( Figure 4A ).

Differentially Expressed miRNAs Between Responders and Non responders

(A) Heatmap analysis representing the differential expression of the most abundant miRNAs from EVs in patients showing NEDA-3, NEDA-4, and NEDA-5. In the color bar, blue represents high expression and green represents low expression. The colors of the heatmap represent the average expression level of the corresponding miRNA for each patient group. The miRNAs included in the heatmap were selected based on those that were differentially expressed shown in the volcano plots. Principal component analysis of the miRNAs differentially expressed in patients with (B) NEDA-3, (C) NEDA-4, and (D) NEDA-5. Patients with NEDA-3 (n = 49), EDA-3 (n = 31), NEDA-4 (n = 65), EDA-4 (n = 15); NEDA-5 (n = 73), and EDA-5 (n = 7). mRNA = microRNA; NEDA = No Evidence of Disease Activity.

PCA revealed distinctly different profiles of miR-28-3p, miR-660-5p, let-1-3p, and miR-98-5p in patients with NEDA-3 and EDA-3. PCA components, which included miR-28-3p, miR-660-5p, miR-222-3p, and miR-144-5p, were able to differentiate patients with NEDA-4 from those with EDA-4. Finally, PCA showed distinctly different profiles of miR-92a-3p, miR-21-5p, let-7a-5p, miR-23a-3p, miR-486-5p, miR-19b-3p, and miR-574-5p in patients with NEDA-5 and EDA-5 ( Figure 4, B and D , eTable 4).

miR-186-5p Expression Was Associated With Brain Structure Atrophy

miR-186-5p expression in circulating EVs correlated negatively with brain structure atrophy, including brain parenchymal fraction ( R = −0.563, p = 0.003), total brain volume ( R = −0.753, p = 0.003), white matter ( R = −0.778, p = 0.001), gray matter ( R = −0.739, p = 0.001), cortical gray matter ( R = −0.699, p = 0.001), cerebellum ( R = −0.751, p = 0.001), hippocampus ( R = −0.696, p = 0.001), thalamus ( R = −0.650, p = 0.001), caudate nucleus ( R = −0.684, p = 0.001), globus pallidus ( R = −0.794, p = 0.001), or lesion volume ( R = 0.427, p = 0.037) (eFigure 1D).

Combination of L1CAM + , CD3 + , and MOG + EV and miRNA Levels Shows High Robustness for the Early Therapeutic Response Model

After establishing CD3 + EV levels, L1CAM + EV levels, miR-28-3p, miR-660-5p, let-1-3p and miR-98-5p as NEDA-3 biomarkers in patients with MS; CD20 + EV levels, CD3 + EV levels, L1CAM + EV levels, miR-28-3p, mir-222-3p, miR-144-5p, and miR-660-5p as NEDA-4 biomarkers; and CD3 + EV levels, L1CAM + EV levels, miR-92a-3p, miR-21-5p, miR126-5p, miR-7a-5p, miR-23a-3p, and miR-19b-3p as NEDA-5 biomarkers, we moved on to determine whether these levels were correlated and whether the combination thereof could provide a more robust therapeutic response model. CD3 + , MOG + , and L1CAM + EV levels have a significant positive correlation (eFigure 2A). Moreover, miR-222-3p has a significant positive correlation with miR-98-5p, miR-144-5p, and miR-92-3p (eFigure 2B).

The scores obtained for these combinations following a binary logistic regression model were significantly lower in patients with NEDA-3, NEDA-4, and NEDA-5 ( Figure 5A ). A ROC curve analysis was then performed to determine a cutoff score for the combination of change in the PRE-treatment and POST-treatment CD3 + EV levels (eFigure 3A), L1CAM + EV levels, and miR-98-5p levels ( Figure 5B ; area under the curve [AUC] = 0.86, sensitivity = 78%, and specificity = 100%) associated with NEDA-3; for the combination of MOG + EV levels, L1CAM + EV levels, miR-144-5p, and miR-98-5p ( Figure 5C ; AUC = 1.0, sensitivity = 100%, and specificity = 100%) associated with NEDA-4; and for the combination of L1CAM + EV levels, miR-92a-3p, and miR-126-5p ( Figure 5D ; AUC = 0.93, sensitivity = 94.7%, and specificity = 75%) associated with NEDA-5. This analysis confirmed that the combination of EV levels and miRNAs produced a more potent therapeutic response model compared with both on their own. More details about the model are available in eTable 5.

Combination of EV Levels and miRNAs Generates an Early Therapeutic Response Model

(A) The differences in scores of the logistic regression model in responders and non responders. (B) The receiver operating characteristic (ROC) curve describing the performance value of L1CAM + and CD3 + EV levels and miR98-5p for NEDA-3. (C) ROC curve describing the performance value of L1CAM + and MOG + EV levels, miR-222-3p, and miR-144-5p for NEDA-4. (D) ROC curve describing the performance value of L1CAM + EV levels, miR-92-3p, and miR-126-5p for NEDA-5. The black line shows the area under the curve (AUC), and the red point shows optimal Youden cutoff; specificity and sensitivity values are shown. The ROC curve analysis was performed using the Wilson-Brown test, with 95% CI. A logistic regression model was applied for therapeutic response, using a receiver operating characteristic (ROC) analysis. EV = extracellular vesicle; NEDA = No Evidence of Disease Activity.

Discussion

This study has shown that EV levels and sizes are indicative of treatment response in patients with MS. Elevated EV levels and larger CD3 + EVs were observed in patients with new MRI lesions, which correlated with reduced levels in patients with NEDA-3 and NEDA-4 compared with those with EDA. Levels of CD20 + EVs were higher in patients experiencing cognitive decline. Larger L1CAM + EVs were detected in patients with new MRI lesions. Certain miRNAs, including miR-28-3p, miR-326, miR-98-5p, miR-144-5p, miR-98-3p, miR-23a-3p, miR-146a-5p, and miR-186-5p, were found to be downregulated in patients with relapse, EDSS progression, cognitive impairment, and brain atrophy. Moreover, a therapeutic response model integrating L1CAM + , CD3 + , and MOG + EVs with miRNAs showed strong potential in forecasting therapeutic response.

Treatment response has traditionally been assessed based on relapse rate, new radiologic activity, and disability progression, terms that led to the emergence of the NEDA-3 composite. 16 While this measure is highly sensitive to inflammatory activity, it has been proposed that NEDA should also take into consideration other important measures of the neurologic condition of patients, such as their cognitive status and the volumetric analysis of the brain, to make NEDA a more effective tool for treatment evaluation. 17 Recent studies have made the development of new biomarkers for disease control, including EVs, possible. Such advances may have potential as early markers of the current stage of inflammation and neurodegeneration of patients in contrast to a blind and prolonged follow-up process characterized by waiting for clinical relapse, disease progression, or the presence of new abnormal MRI findings, all of which increase disability in patients. In this context, brain cells secrete EVs containing information about the brain damage and repair component of MS, and immune system cells release EVs that could reveal the immune processes of the disease. 18 Bearing this in mind, this article explores the potential utility of EVs as a liquid biopsy for a clinical decision-making tool that incorporates pathologic components into the concept of NEDA.

Since the concept of EVs first emerged in 1971, numerous studies have focused on identifying their functional roles in disease and maximizing their clinical benefits. 6,19,20 In this study, the levels and size of CD3 + EVs were higher in patients with new MRI lesions, which is consistent with previously published findings. 6,19 Traditionally, T cells have been key players in the pathogenesis of MS, orchestrating the immune-mediated and inflammatory response components of the disease. 21 Indeed, T cells display a proinflammatory profile and become autoreactive during clinical relapses or new CNS lesions. 22 The role of EVs in T-cell activation, differentiation, and effector mechanisms in MS has been demonstrated previously. Therefore, CD3 + EVs may play a role as markers of T-lymphocyte reactivity during the active phase of the disease, as reported in the literature, 6,20 and, therefore, act as a biomarker for clinical relapse.

Our study also found that patients with new CNS lesions had higher levels of L1CAM + EVs than those in remission, thus leading to an increase in the levels of these EVs in patients with EDA-3, EDA-4, and EDA-5. Similarly, changes in levels of MOG + EVs were observed in patients with MS with EDSS progression. Neurons and oligodendrocytes are fundamental cells comprising the CNS, and they are the primary cells affected during MS activity. Thus, myelin damage and the consequent axonal degeneration are the main histopathologic hallmarks of the disease. 22 Along these lines, recent studies have reported that L1CAM + EVs may act as a marker of the molecular events underlying neuronal damage in MS plaques. 23 In addition, as we have reported elsewhere, changes in MOG + EVs are correlated with motor dysfunction. 6 This information leads us to propose a new potential role for these EVs as markers of neurologic impairment within the context of disease activity.

Differences in CD20 + EV levels were found in patients with progression in cognitive impairment compared with stable patients. The role of B lymphocytes in MS has emerged in recent years, thus demonstrating that they participate in numerous pathophysiologic processes of the disease. 24-26 One of the most important recent findings involving B cells has been the presence of tertiary meningeal lymphoid structures, also known as B-cell follicles, associated with progressive forms of MS. 27 These structures have also been correlated with cognitive impairment. 28 Not unexpectedly, CD20 + EV release was dysregulated in patients with cognitive impairment, thus supporting the idea that B-cell follicles are implicated in this form of disease progression. 29 The dysregulation of CD20 + EV release in patients with cognitive impairment suggests their potential as biomarkers for this condition, and targeting CD20 + EVs may, therefore, provide valuable insights into the mechanisms underlying cognitive impairment in MS.

Recent breakthroughs in deep sequencing technology have had a profound impact on our understanding of miRNA and represent an innovative tool for the description of the transcriptome profile of patients. 30 This study analyzes the role of dysregulated miRNA expression as a biomarker for clinical parameters associated with treatment response. To understand the role of dysregulated miRNA expression as a biomarker for clinical parameters associated with treatment response, in this study, we quantified the expression of miRNAs in circulating total EVs among responders and non responders and correlated the differential expression with clinical assessments used to monitor response. The downregulation of miR-28-3p and miR-326 is associated with a lack of relapse or progression in EDSS during follow-up. Our findings align with previous studies showing that miR-326 is downregulated in nonrelapsing MS patients. These studies show that low levels of miR-326 suppress the differentiation and activation of autoreactive Th17 cells, which are crucial in triggering relapse. 31 These findings are further supported by similar results observed in the levels of EVs released by T lymphocytes. Furthermore, additional studies have indicated that reduced levels of miR-28-3p increase the production of anti-inflammatory IL-10. 32 IL-10 plays a critical role in suppressing the inflammatory cascade and promoting immune tolerance in patients with MS. This aligns with our observation that miR-28-3p is also downregulated in patients with NEDA-3, thereby reinforcing the notion that miR-28-3p downregulation in EVs helps control inflammation, maintain homeostasis, and support long-term disease stability. Furthermore, the downregulation of miR-98 and miR-1-3p was found in patients with new MRI lesions, and that of miR-98-5p in patients with EDA-3 (non responders). In this context, previous studies have indicated that downregulation of miR-98-5p induces a post-transcriptional suppression of IL-10, 33 which may suggest that the downregulation of miR-98-5p promotes a prolonged state of inflammation in non responders. This IL-10–related hotspot is important in our work because it could be the most relevant miRNA-derived differential biological function between responders and non responders.

We further identified an association between miR-144-5p and miR-98-3p with cognitive progression. This finding coincides with that of a previous study that profiled miR-144-5p and miR-98-3p as dysregulated miRNAs in patients with this condition. 34 In our study, brain structure atrophy was also associated with miR-23a-3p and miR-146a-5p dysregulation, which is consistent with recent data suggesting that miR-23a-3p correlates with brain atrophy in patients with Alzheimer disease, another neurologic disease. 34 Other authors have reported that miR-146a-5p suppresses neurogenesis in the brain and, consequently, induces greater atrophy in the brain structures. 35 Finally, we found strong negative correlations, as reflected in the R values, between miR-186-5p expression and the atrophy of all cerebral structures, thus suggesting that this miRNA could act as a biomarker for loss of brain matter. This study may reveal a new function of miR-186-5p, although further research will be needed to confirm these findings.

In addition, we identified a positive correlation between the levels of miR-222-3p and miR-98-5p, miR-144-5p, and miR-92-3p, thus providing stronger evidence that these microRNAs are involved in response to treatment. These microRNAs are implicated in the regulation of innate immunity induced by lipopolysaccharide, which is also associated with the development of various autoimmune diseases, including MS. 36 It has been demonstrated that this potent antigenic stimulus dysregulates microRNAs to modulate the inflammatory response. This microRNA network likely controls excessive inflammatory responses in the CNS.

We also constructed an early response model for the different NEDA statuses based on the levels of EVs and miRNAs. We found a robust score based on levels of L1CAM + and CD3 + EV and miR-98-5p. In agreement with these results, other authors have also highlighted miR-98-5p in peripheral blood samples as a potential biomarker of response to treatment in MS. 37,38 The downregulation of miR-98-5p during relapse and its involvement in crucial regulatory pathways of inflammation underscore its significance in disease progression and response to therapy. Integrating these biomarkers into clinical practice could enhance the precision of treatment strategies and improve long-term outcomes for patients with MS.

A limitation of our study is that miRNA analysis was performed on total circulating EVs because of insufficient yields from isolating EVs of different origins (11.39% of CD20 + , 14.65% of CD3 + , 4.88% of L1CAM + , and 6.29% of MOG + vesicles from total circulating EVs). Advances in EV isolation technologies may allow for more specific analyses in the future, thereby providing a clearer understanding of miRNA roles and origins. Another limitation is the use of L1CAM. While L1CAM is a gold standard for identifying neuronal-derived EVs, its specificity has been questioned because it is also expressed in other CNS cells, peripheral nervous system cells, and non-neural tissues. 39,40 To improve precision, we also used MOG as a surface biomarker for CNS-derived EVs. Our next step is to validate these findings in an independent cohort.

In conclusion, this study highlights the significant potential of EV miRNA expression profiling as a biomarker for clinical parameters related to treatment response in patients, with a particular focus on early assessment. By analyzing miRNA expression as early as 3 months after initiating treatment, our findings demonstrate that specific miRNAs, such as miR-28-3p, miR-326, miR-98-5p, miR-144-5p, miR-98-3p, miR-23a-3p, miR-146a-5p, and miR-186-5p, exhibit differential expression patterns that correlate with key clinical outcomes, including relapse incidence, new lesions in the CNS, cognitive progression, and brain structure atrophy. These early biomarkers not only support the utility of miRNA expression levels in monitoring disease activity and treatment efficacy but also provide an early measure for anticipating irreversible lesions in the nervous system. This early detection aligns with current therapeutic response evaluation methods and offers the potential for timely intervention to mitigate long-term damage, thus representing a step toward personalized medicine for this immune-mediated, demyelinating, and inflammatory disease.