Extracellular Vesicle‐Derived microRNAs as Fluid Biomarkers in Neurodegenerative Diseases: A Systematic Review

ABSTRACT

Given the absence of curative treatments for neurodegenerative diseases, early detection and therapeutic intervention are critical to slowing disease progression. Extracellular vesicles (EVs) have emerged as promising biomarkers for neurodegeneration, owing to their accessibility in bodily fluids and dynamic molecular cargo, including microRNAs (miRNAs). The last decade has seen accumulating evidence for miRNA dysregulation in circulating EVs from people with neurodegenerative diseases; however, assessing reproducibility between studies remains challenging, largely due to clinical and methodological heterogeneity. In this systematic review, we comprehensively searched the MEDLINE database for studies investigating miRNA expression in biofluids from people with neurodegenerative diseases. We extracted miRNA expression data from 185 peer‐reviewed publications, published until June of 2025, reporting altered miRNA levels in fluid‐derived EVs from people with neurodegenerative diseases. We consolidated results between studies to identify the most frequently dysregulated miRNAs across diseases, with a focus on Alzheimer's disease, Parkinson's disease, mild cognitive impairment, multiple sclerosis, amyotrophic lateral sclerosis, frontotemporal dementia, stroke, traumatic brain injury, and schizophrenia. Evaluating tissue specificity of frequently dysregulated miRNAs revealed enrichment of select miRNAs in the nervous system relative to blood and immune compartments. Summarizing miRNA regulation across biofluids emphasized consistencies between cerebrospinal fluid and plasma, but not serum. We highlight circulating miRNAs that may be reflective of neuropathology, including miR‐143‐3p, miR‐127‐3p, miR‐9‐5p, miR‐15a‐5p, and miR‐125b‐5p. Finally, we provide a repository of miRNA expression data from over 30 neurodegenerative conditions which can be exploited to further investigate miRNA regulation in diseases of interest.

Given the absence of curative treatments for neurodegenerative diseases, early detection and therapeutic intervention are critical to slowing disease progression. The last decade has seen accumulating evidence for microRNA dysregulation in circulating extracellular vesicles (EVs); however, assessing reproducibility between studies remains challenging. In this systematic review, we extract microRNA expression data from 185 peer‐reviewed publications reporting altered microRNA levels in fluid‐derived EVs from people with neurodegenerative diseases. Consolidating results between studies highlights microRNAs that may be reflective of neuropathology, including miR‐143‐3p, miR‐127‐3p, miR‐9‐5p, miR‐15a‐5p, and miR‐125b‐5p. To support future research, we provide a repository of microRNA expression data from over 30 neurodegenerative conditions.

Abbreviations

AD

Alzheimer's disease

AH

aqueous humor

ALS

amyotrophic lateral sclerosis

AMPK

AMP‐activated protein kinase

APP

amyloid precursor protein

ASIC1

acid‐sensing ion channel 1

ATP6V1A

ATPase H+ transporting V1 subunit A

Aβ

amyloid beta

BACE1

beta‐secretase 1

BDNF

brain derived neurotrophic factor

CISD1

CDGSH iron sulfur domain 1

DAPK1

death‐associated protein kinase 1

DLB

Lewy body dementia

DM

muscular dystrophy

DME

diabetic macular edema

DR

diabetic retinopathy

ERMP1

endoplasmic reticulum metallopeptidase 1

EV

extracellular vesicle

FASD

fetal alcohol spectrum disorder

FSTL1

follistatin‐related protein 1

FTD

frontotemporal dementia

GSKB

glycogen synthase kinase‐3 beta

HAND

HIV‐associated neurocognitive disorder

HD

Huntington's disease

HK2

hexokinase 2

IRBD

idiopathic REM sleep behavior disorder

LPS

lipopolysaccharide

MCI

mild cognitive impairment

MEDLINE

Medical Literature Analysis and Retrieve System Online

miRNA

microRNA

MS

multiple sclerosis

NGS

next‐generation sequencing

NLRP1

NLR family pyrin domain containing 1

NMO

neuromyelitis optica spectrum disorder

OPTN

optineurin

PCD

paraneoplastic cerebellar degeneration

PCR

polymerase chain reaction

PD

Parkinson's disease

PLS

primary lateral sclerosis

PM

pathological myopia

PME1

protein phosphatase methylesterase 1

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses

PSP

progressive supranuclear palsy

SAH

subarachnoid hemorrhage

SCA

spinocerebellar ataxia

SCD

subjective cognitive decline

SOX11

SRY‐box transcription factor 11

SZ

schizophrenia

TBI

traumatic brain injury

TD

tic disorder

THBS2

thrombospondin 2

VD

vascular dementia

VH

vitreous humor

VHL

von Hippel–Lindau

1. Introduction

Neurological disorders represent a major global disease burden, affecting over 40% of the world population (Steinmetz et al. 2024). At the forefront are neurodegenerative diseases, characterized by progressive degeneration of neurons in the central and peripheral nervous systems, ultimately resulting in debilitating cognitive and motor deficits (Bredesen et al. 2006). The process of aging, which increases the brain's susceptibility to neurodegenerative processes, is a major risk factor for most neurodegenerative diseases, including Alzheimer's disease (AD), Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS) (Hou et al. 2019). Despite an increasing prevalence in the aging population, there are no curative treatments for neurodegenerative diseases. Instead, current therapeutics are limited to slowing progression and managing symptomatic disease, highlighting the importance of early disease detection and intervention.

Extracellular vesicles (EVs) have gained significant interest as dynamic biomarkers in neurodegenerative diseases, promising insight into disease monitoring, prognostication, and treatment response. EVs are lipid bilayer‐bound nanovesicles released by nearly all cell types. The term “EV” is a hypernym for a variety of vesicular subtypes, defined by their size and biogenesis (Welsh et al. 2024). The most well‐studied are exosomes (50–150 nm; released via the endosomal pathway), microvesicles (100–1000 nm; released via direct budding from the plasma membrane), and apoptotic bodies (50–5000 nm; released by cells undergoing programmed cell death); however, increasing subtypes of vesicular and non‐vesicular particles are emerging as important contributors to biofluid composition (Jeppesen et al. 2023). EVs can cross the blood–brain barrier and are abundant in bodily fluids, underlining their potential as non‐invasive biomarkers in neurodegenerative diseases. Further, a growing body of literature has focused on isolating brain‐enriched EVs from peripheral biofluids by targeting specific proteins expressed on the EV surface; namely, L1 cell adhesion molecule (L1CAM) (Torres Iglesias et al. 2023; Bhargava et al. 2021; Banack et al. 2020, 2022; Katsu et al. 2019; Shi et al. 2014, 2016; Niu et al. 2020; Jiang et al. 2020; Zou et al. 2020; Fu et al. 2020; Chou et al. 2020; Agliardi et al. 2021, 2019; Dutta et al. 2021; Jiang, Hopfner, et al. 2021; Yan et al. 2024; Fiandaca et al. 2015; Kluge et al. 2022; Kapogiannis et al. 2015; Goetzl et al. 2015, 2018, 2019; Winston et al. 2016; Goetzl, Mustapic, et al. 2016; Goetzl, Kapogiannis, et al. 2016; Cha et al. 2019; Serpente et al. 2020, 2025; Gu et al. 2020; Li, Yao, et al. 2022; Durur et al. 2022; Visconte, Golia, et al. 2023; Kumar et al. 2023; Sbriscia et al. 2025; Leppik et al. 2025), neural cell adhesion molecule (NCAM) (Fiandaca et al. 2015; Jia et al. 2019, 2022; Li, Meng, et al. 2022; Li, Xia, et al. 2022) and sodium/potassium‐transporting ATPase subunit alpha‐3 (ATP1A3) (You, Zhang, et al. 2023) as markers of neuronal EVs, excitatory amino acid transporter 1 (GLAST) (Bhargava et al. 2021; Goetzl, Mustapic, et al. 2016; Goetzl et al. 2018, 2019; Kumar et al. 2023; Raineri et al. 2024) as a marker of astrocytic EVs, transmembrane protein 119 (TMEM119) (Visconte, Golia, et al. 2023; Kumar et al. 2023) as a marker of microglial EVs, and myelin oligodendrocyte glycoprotein (MOG) (Torres Iglesias et al. 2023, 2025; Dutta et al. 2021; Agliardi et al. 2023) as a marker of oligodendrocyte‐derived EVs. These subpopulations of EVs are thought to retain the pathological signature of their parent cell, underscoring their potential as real‐time readouts of neurodegenerative processes.

EVs contain diverse molecular cargo, including proteins, lipids, and nucleic acids. Among their most abundant cargo are microRNAs (miRNAs). MiRNAs are important post‐transcriptional regulators of gene expression, silencing target messenger RNAs through translational repression or degradation (O'Brien et al. 2018). Extensive research demonstrating miRNA dysregulation in neurodegenerative diseases underscore their critical involvement in nervous system function (Nelson et al. 2008). In circulation, miRNAs contained within EVs are remarkably stable and resistant to ribonuclease activity (Mitchell et al. 2008). Increasing evidence suggests that miRNA signatures within EVs outperform those in whole biofluids with respect to diagnostic capability and correlation with clinical measures (Endzeliņš et al. 2017; Min et al. 2019; Sangalli et al. 2020). Further, miRNAs are detectable in biofluids using standard techniques, including quantitative PCR, microarray, and next‐generation sequencing (NGS), offering improvements upon cost, accessibility, and scalability limitations associated with current neuroimaging‐based biomarkers.

Studies investigating EV‐miRNA regulation in neurodegenerative diseases have catapulted in recent years; however, consolidating results between studies remains challenging due to heterogeneity in patient cohorts, pre‐analytical variables, and EV isolation methods. Identifying repeated patterns of EV‐miRNA dysregulation across neurodegenerative diseases may reveal miRNAs that correlate with neurodegeneration and would be useful in monitoring disease progression or treatment response, while providing insight into molecular mechanisms underlying neurodegeneration. In this systematic review, we recorded miRNA expression from 185 studies identifying miRNA dysregulation in fluid‐derived EVs from people with neurodegenerative diseases. We highlight the most frequently dysregulated miRNAs across diseases, with a focus on AD, PD, mild cognitive impairment (MCI), multiple sclerosis (MS), ALS, frontotemporal dementia (FTD), traumatic brain injury (TBI), schizophrenia (SZ), and stroke. We further summarize trends in EV‐derived miRNA expression across biofluids, including plasma, serum, cerebrospinal fluid (CSF), tears, vitreous humor, urine, and saliva.

2. Methods

This systematic review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines (Page et al. 2021) (Data S1 and S2). The review protocol was not preregistered; however, it is provided in Data S3.

2.1. Study Design

The design of this study was guided by the PICO framework (Population, Intervention, Comparison, Outcome). Studies of interest involve patients with a neurodegenerative disease (Population), either subjected or not to a given therapeutic (Intervention), compared to a non‐neurodegenerative group (Comparison), reporting significant changes in miRNA expression within biofluid‐derived EVs (Outcome).

To enable comparisons in miRNA expression between different disease stages or timepoints, between different neurodegenerative diseases, or between neurodegenerative patients receiving a therapeutic and those receiving placebo or standard care, the following comparators were also included; however, they were not used for data synthesis: (1) comparisons to another neurodegenerative disease, (2) comparisons to a different disease stage or timepoint, (3) comparisons to neurodegenerative patients receiving a placebo or appropriate standard care. The expanded PICO framework is available in Data S3.

2.2. Eligibility Criteria

A detailed description of inclusion and exclusion criteria is provided in Data S3.

Articles adhering to the PICO framework and meeting the following inclusion criteria were included in this review: (1) articles that collect bodily fluid(s) from patients with a neurodegenerative disease; (2) articles that isolate EVs from a bodily fluid; (3) articles that evaluate miRNA expression in EVs using targeted or screening‐based approaches; and (4) articles that identify differentially expressed miRNAs in patients with a neurodegenerative disease relative to an eligible comparator. Only primary, peer‐reviewed articles that generate original miRNA expression data were included.

Animal studies, in vitro studies, reviews, meta‐analyses, conference proceedings, dissertations, retracted articles, and other non‐peer reviewed publications were excluded from this review.

To enable the identification of miRNAs that may have utility in monitoring disease progression or treatment response, including in prodromal stages of disease or following brain injury, the following categories of neurodegenerative conditions were considered: (1) primary neurodegenerative diseases (e.g., AD, PD); (2) conditions involving secondary neurodegeneration (e.g., stroke, TBI); (3) conditions involving cognitive changes more pronounced than with normal aging (e.g., MCI, HIV‐associated neurocognitive disorder); (4) conditions that may occur as comorbidities of neurodegenerative diseases (e.g., epilepsy, tic disorders). An expanded list of representative diseases for each of the above categories is available in Data S3.

2.3. Literature Search Strategy

With the assistance of a librarian (Alexandre Amar‐Zifkin), we comprehensively searched the Medical Literature Analysis and Retrieve System Online (MEDLINE) (via Ovid) database for publications evaluating miRNA expression in biofluids from people with neurodegenerative diseases. The search terms were selected to be inclusive of all neurodegenerative conditions, as outlined above, as well as all bodily fluids, including plasma, serum, CSF, saliva, urine, and ocular fluids. The search results include articles published until June of 2025 inclusively. Forward and backward citation searching of retrieved articles was also performed to identify relevant manuscripts. The detailed search strategy is provided in Data S4.

2.4. Article Selection

Results of the literature search were imported into an EndNote library. After removing duplicates, articles were assessed for eligibility following three rounds of evaluation (Juźwik et al. 2019) (1) screening of titles and abstracts; (2) screening of full texts; (3) full text review and data extraction (Figure 1).

FIGURE 1.

PRISMA flowchart outlining screening and selection of articles identifying miRNA dysregulation in fluid‐derived EVs from people with neurodegenerative diseases. A total of 1725 publications were identified through database and manual searching. After evaluating manuscripts using predefined inclusion and exclusion criteria (see Section 2), a total of 185 manuscripts were included in this review. References for all accepted manuscripts are provided in Data S5.

At the first evaluation stage, articles were assessed for relevance to the review topic and article type (primary research article, review, meta‐analysis, conference abstract). Articles that were unrelated to a neurodegenerative disease, miRNA, or did not generate original miRNA expression data were excluded. At the second evaluation stage, exclusion criteria were expanded to the following: (1) not a primary research article; (2) no reporting of miRNA regulation in human subjects; (3) no reporting of miRNA regulation in a bodily fluid; (4) no isolation of EVs from a bodily fluid; (5) only reporting cellular miRNA regulation in a bodily fluid; (6) a duplicate.

At the third evaluation stage, full manuscripts were reviewed for adherence to inclusion criteria, resulting in 185 publications eligible for data extraction. The complete reference list for all accepted manuscripts is provided in Data S5.

2.5. Data Extraction

The following information was extracted from each eligible publication and compiled into a predefined template: (1) article title, (2) publication year, (3) publication journal, (4) disease(s) studied, (5) biofluid(s) studied, (6) EV isolation method(s), (7) EV characterization method(s), (8) RNA analysis method(s), (9) miRNA name(s), (10) direction of regulation, and (11) geographical information. All miRNAs reported as significantly dysregulated were included in this review. Given that reproducibility is a major challenge across biomarker research, for studies performing next‐generation sequencing or microarray followed by PCR‐based validation of select miRNAs, only PCR‐validated miRNAs were included.

2.6. Selection of Frequently Dysregulated miRNAs Across Diseases

To establish a list of frequently dysregulated EV‐miRNAs across neurodegenerative diseases, we first selected diseases for which we identified > 5 independent publications reporting differential miRNA expression relative to healthy control subjects. This resulted in nine diseases (AD, ALS, FTD, MCI, MS, PD, stroke, SZ, and TBI) and 130 unique publications. To identify commonly dysregulated miRNAs across these diseases, we selected miRNAs that appeared in > 5% of the resulting publications and ≥four different diseases. This resulted in a list of 32 unique miRNAs, shown in Figures 5, 6, 7. Expression of these miRNAs in diseases that did not meet the above cutoffs is provided in Table 1.

FIGURE 5.

Bubble heatmap summarizing expression of frequently dysregulated EV miRNAs across neurodegenerative diseases. miRNA regulation in circulating EVs was summarized for neurodegenerative diseases with > 5 independent publications making a comparison to healthy subjects. The most frequently dysregulated miRNAs across diseases were selected (see Section 2 for selection strategy) and assigned a score to summarize directionality. A score of 1 (red) indicates that a miRNA was exclusively upregulated relative to healthy subjects. A score of −1 (blue) indicates that a miRNA was exclusively downregulated. A score of 0 (white) indicates mixed regulation. The sum of miRNA scores across diseases reveals a predominant upregulation or downregulation of that miRNA relative to healthy subjects. Bubble size corresponds to the number of unique observations of each miRNA as upregulated or downregulated. AD, Alzheimer's disease; ALS, amyotrophic lateral sclerosis; FTD, frontotemporal dementia; MCI, mild cognitive impairment; MS, multiple sclerosis; PD, Parkinson's disease; SZ, schizophrenia; TBI, traumatic brain injury.

FIGURE 6.

Tissue enrichment of frequently dysregulated EV‐derived miRNAs across neurodegenerative diseases. Heatmap of frequently dysregulated miRNAs in fluid‐derived EVs from people with neurodegenerative diseases. Color intensity corresponds to the scaled mean expression level of each miRNA across tissues of the nervous system, blood and immune system. miRNAs that are enriched in the nervous system relative to blood or immune cells are shown in bold. Human miRNA expression data was sourced from the miRNA Tissue Atlas 2025 (Rishik et al. 2024). See Data S6 for an expanded heatmap including all available bodily tissues.

FIGURE 7.

Bubble heatmap summarizing expression of frequently dysregulated EV‐derived miRNAs across biofluids. EV miRNA regulation was summarized across biofluids, irrespective of disease. The most frequently dysregulated miRNAs across diseases were selected (see Section 2 for selection strategy) and assigned a score to summarize directionality. A score of 1 (red) indicates that a miRNA was exclusively upregulated relative to healthy subjects. A score of −1 (blue) indicates that a miRNA was exclusively downregulated. A score of 0 (white) indicates mixed regulation. The sum of miRNA scores across biofluids reveals a predominant upregulation or downregulation of that miRNA relative to healthy subjects. Bubble size corresponds to the number of unique observations of each miRNA as upregulated or downregulated. CSF, cerebrospinal fluid; VH, vitreous humor.

TABLE 1.

Expression of frequently dysregulated EV‐derived miRNAs across neurodegenerative diseases.

miRNA	Biofluid	Directionality (count relative to control) a	Comparison to healthy control b	Other comparison(s)
let‐7e‐5p	Plasma EVs	Up (5)	AD (Durur et al. 2022) FTD (Sproviero et al. 2021) PD (Nie et al. 2020; Sproviero et al. 2021) TBI (Devoto et al. 2020, 2022	PD vs AD (Nie et al. 2020)
		Down (5)	AD (Chai et al. 2024; Nie et al. 2020) HAND (O'Meara et al. 2019) Stroke (Qin et al. 2024) TD (Wang, Xu, et al. 2022)	HIV (lower neuropsychological performance) vs HIV (higher neuropsychological performance) (O'Meara et al. 2019)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)	PM (G2 degree of myopic maculopathy) vs PM (G3/4 degree of myopic maculopathy) (You, Wu, et al. 2023)
let‐7i‐5p	Plasma EVs	Up (5)	AD (Chai et al. 2024) MS (Kimura et al. 2018; Matar et al. 2022) PD (Nie et al. 2020) SZ (Du et al. 2023)
		Down (1)	AD (Gamez‐Valero et al. 2019)
	Serum EVs	Down (3)	MCI (Wang, Yuan, et al. 2022) PCD (Tveit Solheim et al. 2024) PD (Miyamoto et al. 2022)
miR‐9‐5p	Plasma EVs	Up (4)	AD (Jia et al. 2022; Kumar et al. 2023) PD (Li, Cao, et al. 2024) ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021)
		Down (3)	AD (Kumar et al. 2023; Sun et al. 2023) MCI (Kumar et al. 2023; Sun et al. 2023) HAND (O'Meara et al. 2019)	MCI vs AD (Sun et al. 2023)
	Serum EVs	Up (1)	TBI (Ko et al. 2018)
		Down (0)		TBI (biperiden‐treated) vs TBI (placebo) (Villena‐Rueda et al. 2024)
	Tear EVs	Up (1)	DR (Hu et al. 2022)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)
		Down (0)		DR (proliferative) vs Macular hole (Kot and Kaczmarek 2022)
miR‐15a‐5p	Plasma EVs	Up (2)	AD (Jia et al. 2022) ALS (Saucier et al. 2019)	PD‐IRBD vs IRBD (Li, Cao, et al. 2024) Stroke (haemorrhagic) vs Stroke (ischaemic) (Kalani et al. 2020) Stroke (intraparenchymal haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020) Stroke (5 days after subarachnoid haemorrhage) vs Stroke (1 day after subarachnoid haemorrhage) (Kikkawa et al. 2017) Stroke (1, 5 and 7 days after subarachnoid haemorrhage) vs Hydrocephalus (Kikkawa et al. 2017)
		Down (0)		Stroke (3 days after subarachnoid haemorrhage) vs Stroke (1 day after subarachnoid haemorrhage) (Kikkawa et al. 2017)
	Serum EVs	Up (5)	AD (Cheng et al. 2015, 2020; Li, Xie, et al. 2020; Song et al. 2022) VD (Li, Xie, et al. 2020) PCD (Tveit Solheim et al. 2024)
		Down (1)	Stroke (Otero‐Ortega et al. 2021)	Stroke (cortical‐subcortical ischaemic) vs Stroke (subcortical ischaemic) (Otero‐Ortega et al. 2021)
	CSF EVs	Up (1)	MS (Mohammadinasr et al. 2023)	Stroke (1, 3 and 5 days after subarachnoid haemorrhage) vs Hydrocephalus (Kikkawa et al. 2017) Stroke (3 and 5 days after subarachnoid haemorrhage) vs Stroke (1 day after subarachnoid haemorrhage) (Kikkawa et al. 2017)
	VH EVs	Up (0)		DR (proliferative) vs Macular hole (Kot and Kaczmarek 2022)
		Down (1)	PM (You, Wu, et al. 2023)
miR‐16‐5p	Plasma EVs	Up (4)	AD (Visconte, Fenoglio, et al. 2023) ALS (Liu, Ding, et al. 2023; Sproviero et al. 2021) Stroke (Pir et al. 2024)	ALS (bulbar‐onset) vs ALS (spinal‐onset) (Liu, Ding, et al. 2023)
		Down (2)	ALS (Liu, Ding, et al. 2023) Glaucoma (An et al. 2025)
	Serum EVs	Up (2)	MS (Cuomo‐Haymour et al. 2022) PCD (Tveit Solheim et al. 2024)	MS (relapsing‐remitting) vs Clinically isolated syndrome (Cuomo‐Haymour et al. 2022)
	CSF EVs	Up (1)	AD (Sandau et al. 2022)
		Down (1)	AD (McKeever et al. 2018)
	Tear EVs	Down (1)	Glaucoma (Tamkovich et al. 2019)
	VH EVs	Up (0)		DR (proliferative) vs Macular hole (Kot and Kaczmarek 2022)
		Down (1)	PM (You, Wu, et al. 2023)
miR‐18a‐5p	Plasma EVs	Up (3)	ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021) TBI (Devoto et al. 2020, 2022)	Stroke (post‐mesenchymal stem cell therapy) vs Stroke (standard care) (Bang et al. 2022)
	Serum EVs	Up (1)	MS (Mohammadinasr et al. 2023)
		Down (1)	PD (Miyamoto et al. 2022)
	CSF EVs	Up (1)	MS (Mohammadinasr et al. 2023)
miR‐20a‐5p	Plasma EVs	Up (3)	AD (Jia et al. 2022) ALS (Sproviero et al. 2021) PD (Li, Cao, et al. 2024)	Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage + ischaemic stroke) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
		Down (3)	AD (Nie et al. 2020; Wang, Zhen, et al. 2022) Glaucoma (An et al. 2025)
	Serum EVs	Up (3)	AD (Cheng et al. 2015, 2020; Li, Xie, et al. 2020) MCI (Li, Xie, et al. 2020) VD (Li, Xie, et al. 2020)
		Down (2)	PD (Miyamoto et al. 2022) MCI (Wang, Yuan, et al. 2022)
	CSF EVs	Down (2)	AD (Sandau et al. 2022) Encephalitis (Li, Gu, et al. 2022)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)	PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐21‐5p	Plasma EVs	Up (3)	AD (Nie et al. 2020) Glaucoma (Li, Zhang, et al. 2024) Stroke (Wang, Li, et al. 2018)	Stroke (subacute ischaemic) vs Stroke (acute ischaemic) (Wang, Li, et al. 2018) Stroke (recovery phase ischaemic) vs Stroke (acute ischaemic) (Wang, Li, et al. 2018)
		Down (3)	AD (Fitz et al. 2021) HD (Herrero‐Lorenzo et al. 2024) TBI (Ghai et al. 2020)	AD vs DLB (Gamez‐Valero et al. 2019) MS (relapsing‐remitting treatment responder) vs MS (relapsing‐remitting treatment non‐responder) (Torres‐Iglesias et al. 2025) HD (premanifest 1.5 year follow‐up) vs HD (premanifest baseline) (Herrero‐Lorenzo et al. 2024)
	Serum EVs	Up (1)	MS (Cuomo‐Haymour et al. 2022)	TBI (fracture) vs Fracture (Lin et al. 2023)
		Down (2)	AD (Batabyal et al. 2023) PCD (Tveit Solheim et al. 2024)	PD (impaired balance) vs PD (normal balance) (Vaitkienė et al. 2024) PD (motor fluctuations) vs PD (no motor fluctuations) (Vaitkienė et al. 2024) DME (+ type 2 diabetes) vs Type 2 diabetes alone (Jiang, Cao, et al. 2021)
	CSF EVs	Up (3)	Encephalitis (Li, Gu, et al. 2022) MS (Mohammadinasr et al. 2024) Stroke (Scheiber et al. 2024)
	VH EVs	Up (0)		DR (proliferative) vs Macular hole (Kot and Kaczmarek 2022)
		Down (1)	PM (You, Wu, et al. 2023)
	Urine EVs	Down (1)	DM (Catapano et al. 2018)
miR‐22‐3p	Plasma EVs	Up (5)	AD (Chai et al. 2024; Nie et al. 2020) ALS (Liu, Ding, et al. 2023) Stroke (Qin et al. 2024) SZ (Du et al. 2023)	AD (with medial temporal atrophy) vs AD (without medial temporal atrophy) (Chai et al. 2024) Stroke (intraparenchymal haemorrhage) vs Stroke (subarachnoid haemorrhage) (Kalani et al. 2020) Stroke (large artery atherosclerosis) vs Stroke (small vessel occlusion) (Qin et al. 2024)
		Down (1)	TBI (Munoz et al. 2021)
	Serum EVs	Up (3)	AD (Dong et al. 2021) PCD (Tveit Solheim et al. 2024) PD (Manna et al. 2021)	MS (relapsing‐remitting IFNβ–treated) vs MS (relapsing‐remitting treatment‐naive) (Manna et al. 2018)
	CSF EVs	Down (1)	AD (Batabyal et al. 2023)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)
miR‐27a‐3p	Plasma EVs	Up (5)	AD (Nie et al. 2020; Wang, Zhen, et al. 2022) Dementia (Jang et al. 2024) IRBD (Li, Cao, et al. 2024) MS (Jang et al. 2024) PD (Jang et al. 2024) TBI (Puffer et al. 2021)	AD (with depression) vs AD (without depression) (Wang, Zhen, et al. 2022) TBI (altered consciousness) vs TBI (normal consciousness) (Puffer et al. 2021) TBI (blast‐related) vs TBI (blunt trauma) (Devoto et al. 2022)
		Down (1)	HD (Herrero‐Lorenzo et al. 2024)	Stroke (subarachnoid haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Down (1)	ALS (Xu et al. 2018)
	CSF EVs	Up (1)	AD (Jain et al. 2019)
miR‐92a‐3p	Plasma EVs	Up (3)	AD (Visconte, Fenoglio, et al. 2023) FTD (Manzini et al. 2025) MS (Kimura et al. 2018)	FTD vs AD (Manzini et al. 2025)
		Down (2)	AD (Manzini et al. 2025) FTD (Manzini et al. 2025) Glaucoma (An et al. 2025)	MS (relapsing‐remitting treatment responder) vs MS (relapsing‐remitting treatment non‐responder) (Torres‐Iglesias et al. 2025)
	Serum EVs	Up (2)	MS (Cuomo‐Haymour et al. 2022) PCD (Tveit Solheim et al. 2024)	MS (relapsing‐remitting) vs Clinically isolated syndrome (Cuomo‐Haymour et al. 2022)
		Down (1)	SZ (Barnett et al. 2023)
	CSF EVs	Up (1)	Encephalitis (Zhang et al. 2025)
	VH EVs	Up (0)		DR (proliferative) vs Macular hole (Kot and Kaczmarek 2022)
miR‐93‐5p	Plasma EVs	Up (2)	AD (Jia et al. 2022) ALS (Liu, Ding, et al. 2023)	Stroke (haemorrhagic) vs Stroke (ischaemic) (Kalani et al. 2020) Stroke (intraparenchymal haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
		Down (3)	Glaucoma (An et al. 2025) IRBD (Li, Cao, et al. 2024) PD (Nie et al. 2020)
	Serum EVs	Up (6)	AD (Batabyal et al. 2023; Cheng et al. 2015, 2020; Li, Xie, et al. 2020) Stroke (Zhou, Xu, et al. 2022) PCD (Tveit Solheim et al. 2024)	AD vs MCI (Li, Xie, et al. 2020) AD vs VD (Li, Xie, et al. 2020)
		Down (1)	SZ (Barnett et al. 2023)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)	PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐125b‐5p	Plasma EVs	Up (5)	AD (Kumar et al. 2023) MCI (Kumar et al. 2023) ALS (Sproviero et al. 2021) Stroke (Wu et al. 2024) Epilepsy (Wang, Wang, et al. 2022) TD (Wang, Xu, et al. 2022)	HIV (lower neuropsychological performance) vs HIV (higher neuropsychological performance) (O'Meara et al. 2019) Stroke (intracerebral haemorrhage with poor outcomes) vs Stroke (intracerebral haemorrhage with good outcomes) (Wu et al. 2024)
		Down (5)	AD (Jia et al. 2022; Kumar et al. 2023; Lugli et al. 2015; Sun et al. 2023) MCI (Kumar et al. 2023; Sun et al. 2023) PD (Li, Cao, et al. 2024)	MS (relapsing‐remitting treatment non‐responder) vs MS (relapsing‐remitting treatment responder) (Torres‐Iglesias et al. 2025)
	Serum EVs	Up (3)	AD (Song et al. 2023; Song et al. 2022) MCI (Song et al. 2023; Wang, Yuan, et al. 2022)
	CSF EVs	Up (1)	AD (McKeever et al. 2018)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)
		Down (0)		DR (proliferative) vs Macular hole (Kot and Kaczmarek 2022)
miR‐127‐3p	Plasma EVs	Up (3)	AD (Jia et al. 2022) PD (Li, Cao, et al. 2024) Stroke (Wu et al. 2024)
		Down (2)	ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021) HAND (O'Meara et al. 2019)	Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage + ischaemic stroke) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage) (Kalani et al. 2020)
	Serum EVs	Up (2)	ALS (Lo et al. 2021) MS (Ebrahimkhani et al. 2017)	MS (active relapsing‐remitting MS after 6 months fingolimod) vs MS (quiescent relapsing‐remitting MS after 6 months fingolimod) (Ebrahimkhani et al. 2019)
		Down (1)	PCD (Tveit Solheim et al. 2024)
	CSF EVs	Up (1)	PD (Tong et al. 2022)
		Down (1)	HD (Sanchez et al. 2021)
	Tear EVs	Up (1)	DR (Hu et al. 2022)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)
miR‐132‐5p	Plasma EVs	Up (2)	AD (Kumar et al. 2023) FTD (Sproviero et al. 2021) MCI (Kumar et al. 2023)
		Down (1)	AD (Kumar et al. 2023) MCI (Kumar et al. 2023)
	Serum EVs	Down (1)	MS (Mohammadinasr et al. 2023)
	CSF EVs	Up (2)	AD (Gui et al. 2015) Encephalitis (Li, Gu, et al. 2022)
		Down (1)	MS (Mohammadinasr et al. 2023)
	VH EVs	Down (0)		PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐143‐3p	Plasma EVs	Up (6)	AD (Jia et al. 2022) ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021) Stroke (Qin et al. 2024; Tiedt et al. 2017; Wu et al. 2024) TBI (Puffer et al. 2021)	Stroke (intracerebral haemorrhage with poor neurological outcomes) vs Stroke (intracerebral haemorrhage with good neurological outcomes) (Wu et al. 2024) TBI (poor neurological outcomes) vs TBI (good neurological outcomes) (Wu et al. 2024) TBI (altered consciousness) vs TBI (normal consciousness) (Puffer et al. 2021)
	Serum EVs	Up (4)	AD (Cheng et al. 2015; Li, Xie, et al. 2020) MCI (Li, Xie, et al. 2020) PD (Miyamoto et al. 2022) Stroke (Zhou, Xu, et al. 2022) VD (Li, Xie, et al. 2020)
	Tear EVs	Up (1)	DR (Hu et al. 2022)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)
miR‐146a‐5p	Plasma EVs	Up (7)	AD (Nie et al. 2020) ALS (Banack et al. 2020, 2022, 2024) PD (Li, Cao, et al. 2024) IRBD (Li, Cao, et al. 2024) Stroke (Qin et al. 2024) SZ (Du et al. 2023)	PD vs ALS (Banack et al. 2024) ALS vs PLS (Banack et al. 2024) MS (relapsing‐remitting treatment non‐responder) vs MS (relapsing‐remitting treatment responder) (Torres‐Iglesias et al. 2025)
		Down (3)	AD (Aharon et al. 2020) PD (Sproviero et al. 2021) TBI (Ghai et al. 2020)	Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage + ischaemic stroke) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Down (1)	PCD (Tveit Solheim et al. 2024)	MS (relapsing‐remitting IFNβ–treated) vs MS (relapsing‐remitting treatment‐naive) (Manna et al. 2018) SZ (treatment‐resistant) vs SZ (non‐treatment resistant) (Barnett et al. 2023)
	CSF EVs	Up (2)	MS (Mohammadinasr et al. 2024) Stroke (Scheiber et al. 2024)
	Tear EVs	Down (1)	DR (Hu et al. 2022)
miR‐150‐5p	Plasma EVs	Up (1)	TBI (Xia et al. 2021)	MS (cognitively impaired) vs MS (cognitively preserved) (Scaroni et al. 2022)
		Down (5)	AD (Jia et al. 2022; Lugli et al. 2015) DR (Mazzeo et al. 2018) PD (Li, Cao, et al. 2024) TBI (Devoto et al. 2020)	DR (type 1 diabetes with proliferative DR) vs Type 1 diabetes alone (Mazzeo et al. 2018) Stroke (large artery atherosclerosis) vs Stroke (small vessel occlusion) (Qin et al. 2024)
	Serum EVs	Up (2)	MS (Mohammadinasr et al. 2023) PCD (Tveit Solheim et al. 2024)
		Down (0)		MS (positive fingolimod responder) vs MS (baseline) (Ebrahimkhani et al. 2019)
	CSF EVs	Up (1)	MS (Mohammadinasr et al. 2023)	AD (female) vs AD (male) (Sandau et al. 2022)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)
miR‐151a‐3p	Plasma EVs	Up (4)	ALS (Banack et al. 2020, 2024) PD (Li, Cao, et al. 2024) IRBD (Li, Cao, et al. 2024) Stroke (Qin et al. 2024)	ALS vs PLS (Banack et al. 2024)
		Down (2)	AD (Gamez‐Valero et al. 2019) ALS (Sproviero et al. 2021)	Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage + ischaemic stroke) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Up (1)	PD (Miyamoto et al. 2022)
		Down (1)	PCD (Tveit Solheim et al. 2024)	AD vs MCI (Wang, Yuan, et al. 2022)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)	PM (G2 degree of myopic maculopathy) vs PM (G3/4 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐151a‐5p	Plasma EVs	Up (5)	AD (Nie et al. 2020) ALS (Banack et al. 2020, 2022, 2024) Stroke (Qin et al. 2024)	ALS vs PD (Banack et al. 2024) ALS vs PLS (Banack et al. 2024)
		Down (1)	ALS (Sproviero et al. 2021) PD (Sproviero et al. 2021) FTD (Sproviero et al. 2021)	Stroke (intraparenchymal haemorrhage) vs Stroke (subarachnoid haemorrhage) (Kalani et al. 2020) Stroke (intraparenchymal haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Up (2)	PD (Miyamoto et al. 2022; Tong et al. 2022)
		Down (2)	PD (He et al. 2021) PCD (Tveit Solheim et al. 2024)	PD (stage 4) vs PD (stage 2) (He et al. 2021) PD (stage 4) vs PD (stage 3) (He et al. 2021)
	CSF EVs	Up (1)	PD (Tong et al. 2022)
miR‐185‐5p	Plasma EVs	Up (5)	AD (Chai et al. 2024) ALS (Saucier et al. 2019) HAND (Arizono et al. 2025) SZ (Du et al. 2023) TBI (Ko et al. 2020)
		Down (2)	AD (Lugli et al. 2015; Wang, Zhen, et al. 2022)	AD (with depression) vs AD (without depression) (Wang, Zhen, et al. 2022) Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage + ischaemic stroke) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Up (2)	AD (Batabyal et al. 2023) PCD (Tveit Solheim et al. 2024)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)	PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐199a‐3p	Plasma EVs	Up (6)	AD (Nie et al. 2020) ALS (Banack et al. 2020, 2024; Cheng et al. 2023) IRBD (Li, Cao, et al. 2024) Stroke (Qin et al. 2024)	ALS vs PLS (Banack et al. 2024) Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage) (Kalani et al. 2020)
	Serum EVs	Up (1)	Stroke (Otero‐Ortega et al. 2020)
		Down (3)	PD (He et al. 2021) Stroke (Otero‐Ortega et al. 2021) PCD (Tveit Solheim et al. 2024)	PD (stage 2) vs PD (stage 3) (He et al. 2021) PD (stage 2) vs PD (stage 4) (He et al. 2021)
	CSF EVs	Up (1)	AD (Batabyal et al. 2023)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)	PM (G2 degree of myopic maculopathy) vs PM (G3/4 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐199a‐5p	Plasma EVs	Up (5)	ALS (Banack et al. 2020, 2024) PD (Li, Cao, et al. 2024) IRBD (Li, Cao, et al. 2024) DM (Catapano et al. 2020) Stroke (Qin et al. 2024)	ALS vs PLS (Banack et al. 2024)
		Down (2)	AD (Chai et al. 2024; Sproviero et al. 2021) ALS (Sproviero et al. 2021) PD (Sproviero et al. 2021)
	Serum EVs	Up (1)	PD (Miyamoto et al. 2022)
		Down (1)	TBI (Huang et al. 2023)	PD vs PSP (Manna et al. 2021) SZ (treatment‐resistant) vs SZ (non‐treatment resistant) (Barnett et al. 2023)
	Tear EVs	Up (1)	DR (Hu et al. 2022)
	VH EVs	Down (1)	PM (You, Wu, et al. 2023)	PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐210‐3p	Plasma EVs	Up (2)	AD (Kumar et al. 2023) ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021) MCI (Kumar et al. 2023)	MCI (amyloid‐positive) vs MCI (amyloid‐negative) (Mankhong et al. 2022) AD (amyloid positive) vs AD (amyloid negative) (Mankhong et al. 2022)
		Down (2)	AD (Kumar et al. 2023) MCI (Kumar et al. 2023) Stroke (Mainali et al. 2025)	Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage + ischaemic stroke) (Kalani et al. 2020) Stroke (subarachnoid haemorrhage) vs Stroke (intraparenchymal haemorrhage) (Kalani et al. 2020) Stroke (ischemic) vs Stroke (intraparenchymal haemorrhage) (Kalani et al. 2020)
	Serum EVs	Up (1)	PCD (Tveit Solheim et al. 2024)	DR (non‐proliferative DR with type 2 diabetes) vs Type 2 diabetes alone (Yang et al. 2023)
	VH EVs	Up (1)	PM (You, Wu, et al. 2023)
miR‐342‐3p	Plasma EVs	Down (3)	AD (Fitz et al. 2021; Jia et al. 2022; Lugli et al. 2015)	Stroke (haemorrhagic) vs Stroke (ischaemic) (Kalani et al. 2020) Stroke (intraparenchymal haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Up (3)	ALS (Lo et al. 2021) MS (Ebrahimkhani et al. 2017; Mohammadinasr et al. 2023)	MS (relapsing‐remitting) vs MS (progressive) (Ebrahimkhani et al. 2017)
		Down (3)	AD (Cheng et al. 2015, 2020; Li, Xie, et al. 2020) MCI (Li, Xie, et al. 2020) VD (Li, Xie, et al. 2020)
	CSF EVs	Up (1)	MS (Mohammadinasr et al. 2023)	AD (female) vs AD (male) (Sandau et al. 2022)
	Tear EVs	Up (1)	DME (Torimura et al. 2024)
miR‐409‐3p	Plasma EVs	Down (1)	AD (Sproviero et al. 2021) ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021)	Transient tic disorder vs Chronic tic disorder (Wang, Xu, et al. 2022) Transient tic disorder vs Tourette syndrome (Wang, Xu, et al. 2022)
	Serum EVs	Up (2)	MS (Ebrahimkhani et al. 2017) PD (Miyamoto et al. 2022)
		Down (2)	AD (Wang, Yuan, et al. 2022) PCD (Tveit Solheim et al. 2024)	AD vs MCI (Wang, Yuan, et al. 2022)
	CSF EVs	Up (3)	AD (Batabyal et al. 2023; Sandau et al. 2022) PD (Gui et al. 2015)	PD vs AD (Gui et al. 2015)
		Down (1)	PD (Tong et al. 2022)
	VH EVs	Down (0)		PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐432‐5p	Plasma EVs	Up (1)	PD (Li, Cao, et al. 2024) IRBD (Li, Cao, et al. 2024)
		Down (2)	AD (Sproviero et al. 2021) ALS (Sproviero et al. 2021) Encephalitis (Xie et al. 2025) PD (Sproviero et al. 2021)	Transient tic disorder vs Chronic tic disorder (Wang, Xu, et al. 2022) Transient tic disorder vs Tourette syndrome (Wang, Xu, et al. 2022) Encephalitis (acute‐phase anti‐NMDAR encephalitis) vs Encephalitis (stable‐phase anti‐NMDAR encephalitis) (Xie et al. 2025)
	Serum EVs	Up (1)	MS (Ebrahimkhani et al. 2017)	MS (progressive) vs MS (relapsing‐remitting) (Ebrahimkhani et al. 2017) MS (active relapsing‐remitting disease after 6 months. fingolimod) vs MS (quiescent relapsing‐remitting disease after 6 months. fingolimod) (Ebrahimkhani et al. 2019) MS (stable fingolimod responder) vs MS (baseline) (Ebrahimkhani et al. 2019)
		Down (2)	AD (Wang, Yuan, et al. 2022) MCI (Wang, Yuan, et al. 2022) PCD (Tveit Solheim et al. 2024)
	CSF EVs	Up (1)	MS (Mohammadinasr et al. 2023)
miR‐451a	Plasma EVs	Up (2)	AD (Visconte, Fenoglio, et al. 2023) ALS (Sproviero et al. 2021)	Stroke (intraparenchymal haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
		Down (3)	AD (Wang, Zhen, et al. 2022) Glaucoma (An et al. 2025) HD (Herrero‐Lorenzo et al. 2024)	AD vs DLB (Gamez‐Valero et al. 2019) AD (with depression) vs AD (without depression) (Wang, Zhen, et al. 2022)
	Serum EVs	Up (5)	AD (Duan et al. 2024; Wang, Yuan, et al. 2022) MS (Cuomo‐Haymour et al. 2022; Ebrahimkhani et al. 2017) PCD (Tveit Solheim et al. 2024)	AD vs MCI (Wang, Yuan, et al. 2022) MS (relapsing‐remitting) vs Clinically isolated syndrome (Cuomo‐Haymour et al. 2022)
		Down (2)	AD (Reho et al. 2025) SZ (Barnett et al. 2023)	MS (relapsing‐remitting IFNβ–treated) vs MS (relapsing‐remitting treatment‐naive) (Manna et al. 2018) SZ (cognitive deficits) vs SZ (cognitively spared) (Barnett et al. 2023)
	CSF EVs	Down (1)	AD (McKeever et al. 2018)
miR‐483‐5p	Plasma EVs	Up (3)	AD (Liu, Chen, et al. 2023) MCI (Liu et al. 2024) Stroke (Wu et al. 2024)	HIV (lower neuropsychological performance) vs HIV (higher neuropsychological performance) (O'Meara et al. 2019)
		Down (2)	PD (Li, Cao, et al. 2024) TBI (Ghai et al. 2020)
	Serum EVs	Up (3)	AD (Wang, Yuan, et al. 2022) MS (Cuomo‐Haymour et al. 2022) PCD (Tveit Solheim et al. 2024)	Encephalitis (HHV‐6‐associated acute encephalopathy) vs Febrile seizure (Torii et al. 2022)
	CSF EVs	Up (1)	AD (Batabyal et al. 2023)
miR‐485‐3p	Plasma EVs	Up (1)	IRBD (Li, Cao, et al. 2024) PD (Li, Cao, et al. 2024
		Down (1)	AD (Sproviero et al. 2021) ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021) PD (Sproviero et al. 2021)	Transient tic disorder vs Chronic tic disorder (Wang, Xu, et al. 2022) Transient tic disorder vs Tourette syndrome (Wang, Xu, et al. 2022)
	Serum EVs	Up (1)	PD (Miyamoto et al. 2022)	MS (progressive) vs MS (relapsing‐remitting) (Ebrahimkhani et al. 2017)
		Down (1)	AD (Wang, Yuan, et al. 2022)
	CSF EVs	Up (0)		PD vs Other neurodegenerative diseases (AD/MS/HD) (Tong et al. 2022)
		Down (1)	PD (Tong et al. 2022)
	Salivary EVs	Up (1)	AD (Ryu et al. 2023)	AD (Aβ PET positive) vs AD (Aβ PET negative) (Ryu et al. 2023)
miR‐625‐3p	Plasma EVs	Up (2)	PD (Li, Cao, et al. 2024) Stroke (Qin et al. 2024)
		Down (1)	AD (Sproviero et al. 2021) ALS (Sproviero et al. 2021) FTD (Sproviero et al. 2021) PD (Sproviero et al. 2021)	Stroke (intraparenchymal haemorrhage) vs Stroke (subarachnoid haemorrhage) (Kalani et al. 2020) Stroke (intraparenchymal haemorrhage) vs Stroke (ischaemic) (Kalani et al. 2020)
	Serum EVs	Down (1)	PCD (Tveit Solheim et al. 2024)
	CSF EVs	Up (1)	AD (Batabyal et al. 2023)
	VH EVs	Down (0)		PM (G3/4 degree of myopic maculopathy) vs PM (G2 degree of myopic maculopathy) (You, Wu, et al. 2023)
miR‐4433b‐5p	Plasma EVs	Up (1)	IRBD (Li, Cao, et al. 2024) PD (Li, Cao, et al. 2024)
		Down (2)	AD (Sproviero et al. 2021) ALS (Sproviero et al. 2021) Encephalitis (Xie et al. 2025) FTD (Sproviero et al. 2021) PD (Sproviero et al. 2021)	Encephalitis (acute‐phase anti‐NMDAR encephalitis) vs Encephalitis (stable‐phase anti‐NMDAR encephalitis) (Xie et al. 2025) Stroke (intraparenchymal haemorrhage) vs Stroke (subarachnoid haemorrhage) (Kalani et al. 2020)
	Serum EVs	Up (1)	PD (Miyamoto et al. 2022)
		Down (3)	AD (Wang, Yuan, et al. 2022) PCD (Tveit Solheim et al. 2024) TBI (Huang et al. 2023)	AD vs MCI (Wang, Yuan, et al. 2022)

Abbreviations: AD, Alzheimer's disease; ALS, amyotrophic lateral sclerosis; DLB, Lewy body dementia; DM, muscular dystrophy; DME, diabetic macular edema; DR, diabetic retinopathy; FTD, frontotemporal dementia; HAND, HIV‐associated neurocognitive disorder; HD, Huntington's disease; HHV‐6, human herpesvirus 6; HIV, human immunodeficiency virus; IRBD, idiopathic REM sleep behaviour disorder; MCI, mild cognitive impairment; MS, multiple sclerosis; NMDAR, N‐methyl‐D‐aspartate receptor; NMO, neuromyelitis optica spectrum disorder; PCD, paraneoplastic cerebellar degeneration; PD, Parkinson's disease; PM, pathological myopia; PSP, progressive supranuclear palsy; SAH, subarachnoid hemorrhage; SZ, schizophrenia; TBI, traumatic brain injury; VD, vascular dementia.

^a Count refers to the number of unique articles making a comparison between disease and healthy control.

^b A miRNA may be listed as up‐ and down‐regulated in the same study when conflicting directionality was observed in different EV subpopulations.

2.7. Calculation of miRNA Ratios

To evaluate whether a miRNA shows consistent directionality (upregulation or downregulation) across neurodegenerative diseases or biofluids, we visualized miRNA expression using heatmaps. A directionality ratio was calculated for each miRNA according to the following equation: (count of unique observations of upregulated miRNA minus count of unique observations of downregulated miRNA) divided by the total number of unique observations of the miRNA. Only comparisons between disease and healthy control were included. When a single study reported a miRNA as dysregulated in multiple biofluids or multiple EV subpopulations, these were counted as additional observations. When the miRNA strand (3p or 5p) was not reported in the source publication and could not otherwise be determined, that miRNA was excluded from the analysis.

2.8. Nomenclature

The included manuscripts use a variety of terms to describe vesicles, including “exosomes” and “microvesicles”. The term “extracellular vesicle” (EV) is used throughout this review, regardless of terminology used in source publications. Similarly, the included articles use several acronyms to describe PCR‐based RNA detection methods, including “qPCR”, “RT‐PCR”, “RT‐qPCR” and “ddPCR”. The term “PCR” is used throughout this review for consistency. The term “neurodegenerative disease” is used to encompass all neurodegenerative categories outlined in Section 2.2.

3. Results

3.1. Study Characteristics

After removing duplicates, our initial search strategy resulted in 1705 publications. Of these, 511 publications were excluded based on title and abstract review, resulting in 1194 publications evaluated at the manuscript level. After applying predefined inclusion and exclusion criteria (see Section 2), a total of 185 publications were included in this review (Figure 1).

The included articles were published between the years of 2014–2025, with a growth in publication frequency observed from 2020 onwards (Figure 2a). A total of 122 contributing journals were identified, with the top six journals accounting for 20% (37/185) of publications (Figure 2b). A total of 31 disease categories involving primary or secondary neurodegeneration were identified (Figure 2c). Of these, 12 diseases were studied in ≥five independent publications: AD, stroke, PD, TBI, MCI, MS, ALS, diabetic retinopathy (DR), SZ, vascular dementia (VD), encephalitis, and FTD. The remaining 19 diseases were each studied in ≤four publications. Analyzing the geographical distribution of study sites revealed 25 contributing countries, with major contributions arising in China (42%), the United States (15%), and Italy (9%) (Figure 2d).

FIGURE 2.

Characteristics of manuscripts included in review. (a, b) Bar plots showing number of accepted articles published per year (a) and top contributing journals (b). (c) Bar plot showing number of articles published per disease category. Manuscripts that are shared between multiple diseases are included in the article count for each disease studied. (d) World map showing geographical distribution of study sites. Circle size corresponds to the number of publications from a given site. (e) Donut plot showing the proportion of manuscripts using each biofluid as a source of EVs. Plasma was the most frequently studied biofluid across manuscripts. Manuscripts that are shared between multiple biofluids are included in the article count for each biofluid studied. (f) Donut plot showing the proportion of manuscripts isolating bulk EVs or an EV subpopulation from source biofluid(s). Manuscripts that isolated both bulk EVs and specific subpopulation(s) were counted twice. (g) Alluvial plot showing the specific EV subpopulations investigated across studies, as well as their source biofluid. The width of the alluvia corresponds to the number of manuscripts studying each subpopulation. AD, Alzheimer's disease; ALS, amyotrophic lateral sclerosis; AMD, age‐related macular degeneration; DLB, Lewy body dementia; DM, muscular dystrophy; DME, diabetic macular edema; DR, diabetic retinopathy; FASD, fetal alcohol spectrum disorder; FTD, frontotemporal dementia; HAND, HIV‐associated neurocognitive disorder; HD, Huntington's disease; IRBD, idiopathic REM sleep behavior disorder; MCI, mild cognitive impairment; MS, multiple sclerosis; NMO, neuromyelitis optica spectrum disorder; PCD, paraneoplastic cerebellar degeneration; PD, Parkinson's disease; PM, pathological myopia; PSP, progressive supranuclear palsy; SAH, subarachnoid hemorrhage; SCA, spinocerebellar ataxia; SCD, subjective cognitive decline; SZ, schizophrenia; TBI, traumatic brain injury; VD, vascular dementia.

The most frequently studied biofluid across publications was plasma (52%), followed by serum (32%), CSF (11%), tears (1.4%), aqueous humor (0.9%), urine (0.9%), vitreous humor (0.9%) and saliva (0.5%) (Figure 2e). While most studies (84%) analyzed bulk EVs from a given biofluid, 16% of studies used a targeted approach to enrich for a subpopulation of EVs deriving from a specific cell type (Figure 2f). The putative cellular origins and proteins targeted for immunocapture were neurons (L1CAM, NCAM, AMPH1; amphiphysin 1, MAP1B; microtubule associated protein 1B; GAP43; growth associated protein 43, NLGN3; neuroligin‐3, GluR1/GluR2; glutamate receptor 1/2, TAG1; transient axonal glycoprotein), astrocytes (GLAST), microglia (TMEM119), oligodendrocytes (PDGFRA; platelet‐derived growth factor receptor alpha), pericytes (PDGFRB; platelet‐derived growth factor receptor beta), endothelial cells (CD31; platelet endothelial cell adhesion molecule‐1, CD105/CD144; endoglin and cadherin‐5), myeloid cells (IB4; isolectin B4), adipocytes (FABP4; fatty acid binding protein 4) and unspecified (ABCA1; ATP‐binding cassette subfamily A member 1) (Figure 2g).

Methodological considerations, including choice of EV isolation technique, are known to influence downstream miRNA detection (Van Deun et al. 2014). The most frequently used EV isolation method across studies was polymer‐based precipitation (45%), followed by ultra/differential centrifugation (20%), membrane affinity columns (13%), size exclusion chromatography (9%), density gradient centrifugation (3%), custom method (3%), or immunocapture (1%). Thirteen studies (6%) used commercially available EV isolation kits for which the mechanisms of action are proprietary (labeled as “other”) (Figure 3a,b). In terms of experimental approaches used for miRNA profiling, over half of the included articles (59%) were discovery‐based, whereas 41% were targeted for specific miRNAs (Figure 3c,d). Of the studies performing NGS or microarray, 53% and 63% of studies, respectively, validated results using a secondary PCR‐based technique (Figure 3c).

FIGURE 3.

Summary of EV isolation and miRNA detection methods used across studies. (a) Donut plot showing the various EV isolation methods used across studies, as well as the proportion of manuscripts using each isolation method. (b) Stacked bar chart showing the distribution of EV isolation methods used per year. (c) Donut plot showing the various miRNA detection methods used across studies, as well as the proportion of manuscripts using each detection method. For screening methods (NGS and microarray), the proportion of studies validating screening results using a secondary PCR‐based method is also shown. (d) Stacked bar chart showing the distribution of miRNA detection methods used per year. NGS; next generation sequencing, PCR; polymerase chain reaction.

A detailed summary of methodological details, including EV isolation and characterization methods, is provided for all studies included in this review (Figure 4).

FIGURE 4.

Detailed methodological characteristics of accepted manuscripts. Methodological information was extracted from each manuscript and visualized in a heterogeneity plot, as previously employed by Barnes et al. (2023). Studies are listed in chronological order, by publication year. Colors correspond to different methodological categories: Inclusion of healthy controls, cohort size*, miRNA detection method(s), EV source biofluid(s), primary EV isolation method(s)** and EV characterization method(s)**. *Cohort size was calculated using the sum of all patients included in screening and, where applicable, validation experiments. When multiple diseases were studied, patients from all neurodegenerative diseases were included in this calculation. Healthy controls were excluded from the calculation of cohort size. **When EVs from multiple sample types were analyzed (e.g., in vitro assays or animal models), only EV isolation and characterization methods performed on relevant samples were included. NGS, next generation sequencing; TRPS, tunable resistive pulse sensing.

3.2. Trends in EV‐Derived miRNA Regulation Across Neurodegenerative Diseases

We first aimed to determine whether trends in EV‐miRNA dysregulation can be identified across neurodegenerative diseases. To this end, we analyzed miRNA expression data from each of the 185 publications, extracting miRNAs that were significantly upregulated or downregulated and their specific comparison(s). To visualize miRNA regulation across diseases, we focused on diseases with sufficient data (> 5 publications) reporting differential miRNA expression relative to healthy control subjects. This resulted in nine diseases: AD, ALS, FTD, MCI, MS, PD, stroke, SZ, and TBI. We then selected the most frequently dysregulated miRNAs across diseases by calculating the number of times each miRNA was reported as up‐ or downregulated (see Section 2), resulting in a list of 32 miRNAs, whose expression is summarized in Figure 5. A detailed summary of miRNA expression, including in diseases that did not meet the above cutoffs, is provided in Table 1.

While miRNA expression within individual diseases is distinct, patterns of miRNA regulation can be observed across diseases (Figure 5). When considering directional regulation of miRNAs within each disease, directionality was generally mixed, with the exceptions of ALS, MS, and stroke, in which miRNA expression was predominantly upregulated. When comparing miRNA regulation between diseases, 22 miRNAs (69%) showed a predominant upregulation across diseases, while seven miRNAs (22%) were generally downregulated. Three miRNAs (9%), miR‐92a‐3p, miR‐132‐3p, and miR‐342‐3p, showed no consistency in directionality. Of the miRNAs showing a general upregulation across diseases, 10 miRNAs were upregulated in at least four of the predominant diseases: miR‐9‐5p, miR‐16‐5p, miR‐18a‐5p, miR‐22‐3p, miR‐27a‐3p, miR‐127‐3p, miR‐143‐3p, miR‐146a‐5p, miR‐151a‐5p, and miR‐483‐5p. Of these, two miRNAs, miR‐143‐3p and miR‐16‐5p, were upregulated in every disease in which they were identified. More specifically, miR‐143‐3p was upregulated in AD, ALS, FTD, MCI, PD, stroke, and TBI, while miR‐16‐5p was upregulated in AD, ALS, MS, and stroke. Similarly, miR‐22‐3p was among the most consistently regulated miRNAs across diseases, being upregulated in AD, ALS, PD, stroke, and SZ, although downregulated in TBI. Of the miRNAs showing a general downregulation across predominant diseases, only one was downregulated in at least four diseases: miR‐4433b‐5p.

3.3. Tissue Specificity of Frequently Dysregulated miRNAs Across Neurodegenerative Diseases

Most studies included in this review (84%) were performed in blood; therefore, expression of frequently dysregulated miRNAs may be driven by the predominant EV populations in plasma and serum, including platelet‐ and peripheral immune cell‐derived EVs (Auber and Svenningsen 2022). To identify miRNAs that may be reflective of nervous system pathology, we next evaluated their expression levels in the nervous system, blood and immune system using data obtained from the miRNA Tissue Atlas (Rishik et al. 2024) (Figure 6).

Notably, of the 32 frequently dysregulated miRNAs identified in Figure 5, 11 miRNAs (34%) were more highly enriched in the nervous system than in blood or immune cells. These miRNAs include let‐7e‐5p, miR‐9‐5p, miR‐15a‐5p, miR‐125b‐5p, miR‐127‐3p, miR‐143‐3p, miR‐151a‐5p, miR‐199‐3p, miR‐432‐5p, miR‐483‐5p, and miR‐485‐3p. Four of these miRNAs, let‐7e‐5p, miR‐9‐5p, miR‐125b‐5p, and miR‐127‐5p, showed clear enrichment across nervous system tissues with little to no enrichment in blood or immune compartments (Figure 6).

In contrast, other miRNAs were more strongly enriched in blood or immune cells than in nervous system tissues. Most notably, miR‐92a‐3p, miR‐93‐5p, miR‐146a‐5p, miR‐150‐5p, and miR‐625‐3p showed little to no enrichment in nervous system tissues, with strong expression in blood and/or immune compartments. The miRNAs showing the highest enrichment in immune cells, particularly in B lymphocytes, were miR‐18a‐5p, miR‐21‐5p, miR‐22‐3p, miR‐93‐5p, miR‐146a‐5p, miR‐150‐5p, miR‐342‐3p, and miR‐625‐3p. Their expression pattern suggests that in circulation, these miRNAs may be more reflective of immune regulation than nervous system pathology.

3.4. Trends in EV‐Derived miRNA Regulation Across Biofluids

Consistency in miRNA regulation between biofluids, particularly between CSF and blood, is an important consideration when evaluating peripheral biomarkers for central nervous system diseases. To determine whether miRNA regulation is consistent or conflicting across fluid compartments, we summarized miRNA expression across biofluids, irrespective of disease (Figure 7).

Although limited data is available for saliva, urine, and ocular fluids, which were studied in ≤ 3 publications each, trends in miRNA regulation were identifiable within major biofluids. More specifically, miRNA expression was predominantly upregulated in CSF and plasma, whereas serum showed mixed directionality. Among the frequently dysregulated miRNAs across neurodegenerative diseases (Figure 5), 19 miRNAs (59%) showed a predominant upregulation across biofluids, while 10 miRNAs (31%) were predominantly downregulated. Three miRNAs (9%) showed no consistency in directionality. When comparing miRNA regulation between CSF and blood derivatives, the correlation between CSF and plasma was stronger than between CSF and serum. Notably, of the 18 upregulated or downregulated miRNAs identified in CSF, eight miRNAs (44%) showed consistent directionality in plasma (miR‐15a‐5p, miR‐18a‐5p, miR‐27a‐3p, miR‐146a‐5p, miR‐151a‐5p, miR‐199a‐3p, miR‐451a, and miR‐483‐5p), while only four miRNAs (22%) showed consistent directionality in serum (miR‐15a‐5p, miR‐92a‐3p, miR‐125b‐5p, and miR‐483‐5p). Further, EV‐miRNA expression in plasma and serum was often anticorrelated. Despite 32 miRNAs (97%) being identified in both fluids, only nine miRNAs (29%) showed matching directions of regulation: miR‐15a‐5p, miR‐16‐5p, miR‐21‐5p, miR‐22‐3p, miR‐143‐3p, miR‐185‐5p, miR‐483‐5p, miR‐485‐3p, and miR‐4433b‐5p. Interestingly, two miRNAs, miR‐15a‐5p and miR‐483‐5p, were consistently upregulated across CSF, plasma, and serum.

4. Discussion

Given the paucity of therapeutic options for neurodegenerative diseases, there is an urgent need for biomarkers to monitor disease progression and inform treatment decisions. In this respect, EV‐derived miRNAs have garnered significant attention due to their stability in peripheral biofluids and ability to reflect underlying disease pathology. However, methodological heterogeneity has complicated the interpretability and reproducibility of findings, making it challenging to consolidate results between studies. In this review, we extracted miRNA expression data from 185 publications identifying miRNA dysregulation in circulating EVs from people with neurodegenerative diseases. We summarized methodological approaches for each study and identified trends in miRNA dysregulation across neurodegenerative diseases and biofluids.

4.1. Methodological Considerations and Their Influence on Circulating miRNA Expression

The studies included in this review demonstrate wide methodological variability, particularly with respect to biofluid processing and choice of EV isolation technique (Figure 4). Pre‐analytical variables, including collection and preparation of plasma and serum, are the leading source of testing errors and sample rejection in clinical laboratories (Lippi et al. 2008). Importantly, miR‐16‐5p and miR‐451, which were among the most frequently dysregulated miRNAs across diseases and biofluids (Figures 5 and 7), are known to be associated with hemolysis in plasma and serum (Kirschner et al. 2013, 2011). While this does not necessarily preclude their use as biomarkers, caution should be taken when interpreting their dysregulation in circulation (Tiberio et al. 2015). Additionally, the most frequently used EV isolation method across studies was polymer‐based precipitation, which is known to co‐isolate protein contaminants, particularly from blood (Brennan et al. 2020). In circulation, miRNAs are not only contained within EVs but exist in vesicle‐free form through binding to high‐density lipoproteins or RNA binding proteins, including Argonaute2 (Arroyo et al. 2011; Turchinovich et al. 2011; Vickers et al. 2011). Consequently, the possibility of such non‐vesicular miRNAs cannot be excluded in our analysis of these studies.

While methodological standardization would improve reproducibility across studies, it is not always feasible. Many large cohort studies rely on biological material available through biobanks, for which biofluid processing protocols are fixed and can vary between different centers. When sample availability is limited, the types of downstream EV isolation methods may also become restricted. Although methodological heterogeneity is often inevitable, trends in miRNA expression that persist regardless of biofluid or EV preparation methods may facilitate biomarker discovery.

4.2. EV‐Derived miRNAs That May be Reflective of Neuropathology

Despite significant methodological heterogeneity between studies, we identified several miRNAs that were consistently dysregulated across neurodegenerative diseases and showed enriched expression in the nervous system relative to blood or immune cells (Figure 8). Information regarding their utility as biomarkers and functional implications in experimental models of neurodegenerative diseases is provided below.

FIGURE 8.

Chord diagram showing overlap between EV‐derived miRNAs that are frequently dysregulated across neurodegenerative diseases and enriched in the nervous system relative to blood or immune cells. miR‐15a‐5p was enriched in the nervous system and consistently upregulated across CSF, plasma, and serum relative to healthy controls. miR‐125b‐5p was enriched in the nervous system and identified in the largest number of unique studies as dysregulated relative to healthy controls. miR‐9‐5p was enriched in brain tissue and identified as dysregulated across four biofluids and nine different diseases relative to healthy controls. miR‐143‐3p was enriched in the nervous system and identified as the most consistently upregulated miRNA across neurodegenerative diseases relative to healthy controls. miR‐127‐3p was enriched in the nervous system and identified as dysregulated across five biofluids and 11 different diseases relative to healthy controls.

4.2.1. miR‐143‐3p (Upregulated)

The miRNA showing the most consistent directionality across diseases was miR‐143‐3p, which was upregulated in nine different diseases (Table 1), enriched in the nervous system relative to blood and immune cells (Figure 6), and showed consistent directionality in plasma and serum (Figure 7).

When evaluating the practicality of EV‐derived miRNAs as biomarkers, an important consideration is whether their expression in circulating EVs is reflected in whole biofluids, which may represent simpler mediums for biomarker discovery (Couch 2023). We detected miR‐143‐3p as exclusively upregulated in EVs from AD, ALS, MCI, PD, FTD, VD, DR, TBI and stroke (Figure 5, Table 1). Interestingly, in most of these diseases, previous studies have identified upregulation of miR‐143‐3p in whole biofluids without the need for upstream EV isolation. Indeed, miR‐143‐3p was upregulated in serum (Waller, Goodall, et al. 2017; Waller, Wyles, et al. 2017) and CSF (Waller, Wyles, et al. 2017) from sporadic ALS patients, as well as in plasma from three independent ALS cohorts (Soliman et al. 2021; Kmetzsch et al. 2022; Ruffo et al. 2023). In MCI, plasma miR‐143‐3p was upregulated in an Australian cohort of 38 individuals, as well as in 21 cognitively normal individuals with amyloid‐beta (Aβ) positivity (Guévremont et al. 2022). In another study of 48 individuals with PD, miR‐143‐3p was upregulated in whole blood (Yadav et al. 2023). Further, in serum, miR‐143‐3p was upregulated in 48 behavioral‐variant FTD patients (Denk et al. 2018), and in another study of plasma from nine FTD patients (Kmetzsch et al. 2022). Finally, in VD, serum miR‐143‐3p was upregulated in a cohort of 106 individuals (Yang et al. 2022). The exception to the above examples is AD, where in the absence of EV isolation, circulating miR‐143‐3p is consistently downregulated (Lusardi et al. 2017; Čarna et al. 2023; Jia et al. 2021; Dong et al. 2015; Tan et al. 2023). These findings demonstrate that in many contexts, regulation of miR‐143‐3p is consistent between circulating EVs and whole biofluids, suggesting that upstream EV isolation may unnecessarily complicate its use as a biomarker. Of note, miR‐143‐3p has previously been suggested as brain‐enriched (Zhou, Xu, et al. 2022), and its upregulation is not limited to the above‐mentioned neurodegenerative diseases. Indeed, it has also been detected as upregulated in muscular dystrophy (serum) (Koutsoulidou et al. 2022; Kakouri et al. 2022), glaucoma (aqueous humor) (Hubens et al. 2021), age‐related macular degeneration (serum) (Aggio‐Bruce et al. 2023), Charcot–Marie–Tooth disease (plasma) (Wang et al. 2021), and head injury (serum) (Sandmo et al. 2022).

Functionally, miR‐143‐3p has been described as both neuroprotective and pathogenic in experimental models of neurodegenerative diseases. In a cell model of PD, miR‐143‐3p was found to directly target and downregulate expression of von Hippel–Lindau (VHL), whose genetic ablation has previously been shown to rescue dopaminergic neurodegeneration in C. elegans (Chen et al. 2019). Accordingly, overexpression of miR‐143‐3p ameliorated mitochondrial dysfunction in a cell model of PD via the AMPK/PGC1A axis, a master regulator of mitochondrial biogenesis (Liang et al. 2023). In a cell model of AD, overexpression of miR‐143‐3p was found to inhibit secretion of Aβ, reduce phosphorylation of tau and APP, and promote neurite outgrowth (Wang, Shui, et al. 2022). These neuroprotective effects were attributed to targeting of DAPK1, encoding a serine–threonine kinase that is upregulated in the hippocampus of patients with AD (Kim et al. 2016, 2014) and promotes tau stability and phosphorylation (Kim et al. 2014), APP processing, secretion of Aβ (Kim et al. 2016), and neuronal death (You et al. 2017).

Interestingly, miR‐143‐3p shows an opposing role in experimental models of acute brain injury, exerting primarily pathogenic effects. In a mouse model of intracerebral hemorrhage, EV‐encapsulated miR‐143‐3p was predominantly secreted by activated astrocytes and trafficked into brain microvascular endothelial cells, where it induced upregulated expression of cell adhesion molecules. Here, miR‐143‐3p was found to target Atp6v1a, blocking autophagic degradation of cell adhesion molecules and promoting excessive transendothelial migration of neutrophils into the brain, exacerbating brain injury (Wu et al. 2024). Inhibition of miR‐143‐3p has also been shown to reduce neuronal apoptosis following ischemia/reperfusion injury, both in vitro and in vivo, through targeting of the proinflammatory regulator Fstl1 (Wang and Liu 2021; Chaly et al. 2014). In neurons, overexpression of miR‐143 suppresses glucose uptake through targeting of hexokinase 2 (Hk2), while its inhibition reduces neuronal death following oxygen–glucose deprivation (Zeng et al. 2017).

4.2.2. miR‐15a‐5p (Upregulated)

We identified miR‐15a‐5p as exclusively upregulated in EVs from AD, ALS, MS, VD, and PCD (Table 1). Moreover, it was one of only two miRNAs that were consistently upregulated across CSF, plasma, and serum (Figure 7), and its expression was relatively enriched in the nervous system compared to blood or immune cells (Figure 6).

In contrast to miR‐143‐3p, few corroborating results exist in whole biofluids for miR‐15a‐5p. In AD, miR‐15a‐5p was upregulated in one study using plasma (Jia et al. 2021), but downregulated in another two studies using plasma, serum, and CSF (Denk et al. 2018; Čarna et al. 2023). Similarly, in ALS, miR‐15a‐5p was downregulated in two studies of plasma and whole blood (Kmetzsch et al. 2022; Liguori et al. 2018), conflicting with its expression in EVs. In these contexts, EV‐associated miR‐15a‐5p may provide information that cannot be captured at the level of whole biofluids, reinforcing the importance of upstream EV isolation. Accordingly, in a study of diabetic retinopathy, EV‐associated miR‐15a in plasma was significantly correlated with neuronal damage in the retina, while whole circulating miR‐15a showed no correlation (Sangalli et al. 2020).

The available literature suggests a predominantly pathogenic role for this miRNA in experimental models of neurodegenerative diseases. In hippocampal neurons, overexpression of miR‐15a suppresses dendritic morphogenesis, while its inhibition rescues dendritic developmental defects related to MeCP2, the primary protein responsible for Rett syndrome (Gao et al. 2015). Here, miR‐15a was found to target brain‐derived neurotrophic factor (Bdnf), whose protein levels are frequently decreased across neurodegenerative diseases (Bathina and Das 2015; Mariga et al. 2017). In a rodent model of TBI, genetic deletion of the miR‐15a/16–1 cluster alleviated gray and white matter damage, glial activation, and infiltration of peripheral immune cells into the brain, while improving sensorimotor and cognitive functions (Zhou et al. 2024). Similar results have been observed in experimental stroke, with deletion of miR‐15a/16–1 improving blood–brain barrier integrity and neurological recovery, partially attributed to reduced targeting of the pro‐angiogenic factors Vegfa and Fgf2 (Sun et al. 2020; Ma et al. 2020; Yang et al. 2017). In a rodent model of vascular cognitive impairment and dementia, deletion of miR‐15a/16‐1 alleviated myelin and neuronal loss via reduced targeting of the anti‐inflammatory mediators Akt3 and Il10ra (Zhou, Sun, et al. 2022). Further, miR‐15a‐5p has been implicated in schizophrenia, with its inhibition improving cognitive performance and reducing hyperactivity in a rodent model of the disease (Xu et al. 2024).

Contrasting with the above results, in a cell model of temporal lobe epilepsy, overexpression of miR‐15a‐5p led to increased viability and reduced apoptosis in hippocampal neurons (Li, Pan, et al. 2020).

4.2.3. miR‐9‐5p (Mixed Directionality)

Mir‐9‐5p is known to be enriched in brain tissue and is among the most commonly dysregulated miRNAs across neurodegenerative diseases and their animal models (Juźwik et al. 2019). We identified miR‐9‐5p as dysregulated in circulating EVs from AD, ALS, DR, FTD, HAND, MCI, PD, PM, and TBI, corroborating previous findings.

In experimental models of AD, miR‐9‐5p seems to play complex and often conflicting roles. In a cell model of AD, overexpression of miR‐9‐5p protected against cell apoptosis, mitochondrial damage and oxidative stress via suppression of Gsk3b (Liu et al. 2020), encoding a protein implicated in AD pathogenesis (Hanger et al. 1992; Lovestone et al. 2015; Lauretti et al. 2020). miR‐9‐5p has also been associated with synaptic maintenance, with its overexpression rescuing Aβ‐induced synaptotoxicity and tau phosphorylation via the CAMKK2‐AMPK signaling pathway (Chang et al. 2014). Further, miR‐9‐5p has been shown to target beta‐secretase 1 (BACE1), exerting neuroprotective effects in cell models of AD (Ding et al. 2022). Conversely, in late stages of AD mice, inhibition of miR‐9‐5p contributed to improved Aβ clearance, mobility, and cognition. These effects were attributed to reduced targeting of the autophagic regulator optineurin (Optn) and a resulting increase in autophagic activity (Chen et al. 2021). Inhibition of miR‐9‐5p in a rodent model of chronic cerebral hypoperfusion, closely linked with AD and VD, reduced neuronal loss and attenuated synaptic, learning and memory impairments (Wei et al. 2017). Further, miR‐9‐5p was recently shown to directly target Sirt1 and regulate mitochondrial dysfunction and mitophagy in cellular and animal models of AD (Wang, Sun, et al. 2025).

In experimental models of PD, miR‐9‐5p has been reported as both neuroprotective and pathogenic. Overexpression of miR‐9‐5p inhibited MPTP‐induced apoptosis in dopaminergic neurons and improved motor function in PD mice. Here, miR‐9‐5p was found to target Scrib, indirectly modulating beta‐catenin signaling (Xiao et al. 2022). In contrast, miR‐9‐5p knockdown in cell models of PD reduced apoptosis, inflammation, and oxidative stress (Chen et al. 2023; Wang et al. 2019). As reported in experimental models of AD, these effects were partially mediated through regulation of SIRT1.

In experimental models of stroke, miR‐9‐5p plays a primarily neuroprotective role (Shen et al. 2022; Wei et al. 2016; Gai et al. 2023; Zhao et al. 2024, 2025; Wang, Yang, et al. 2018; Cao et al. 2020; Chi et al. 2019; Nampoothiri and Rajanikant 2019). In hypoxic–ischemic mice, overexpression of miR‐9‐5p reduced cell apoptosis, brain atrophy and neuroinflammation, while ameliorating neurobehavioral outcomes. Here, miR‐9‐5p was found to target Ddit4, a regulator of mTOR‐dependent autophagy (Gai et al. 2023). Similar neuroprotective effects were reported through miR‐9‐5p‐mediated suppression of Cxcl11, a chemoattractant for activated T cells (Zhao et al. 2024), and Zbtb20, an upstream regulator of NRF2‐mediated antioxidant expression (Zhao et al. 2025). In ischemic rats, overexpression of miR‐9a‐5p alleviated infarct volume and neurological deficits through targeting of Atg5, an essential contributor to autophagosome assembly (Wang, Yang, et al. 2018). In the same rodent model, miR‐9a‐5p was found to directly target Nlrp1 and reduce NLRP1 inflammasome‐mediated injury both in vitro and in vivo (Cao et al. 2020). In cell models of stroke, overexpression of miR‐9‐5p protects neurons from hypoxic–ischemic injury, partially through targeting of the endoplasmic reticulum stress mediator ERMP1 (Chi et al. 2019; Nampoothiri and Rajanikant 2019).

In rodent models of TBI, the effects of miR‐9‐5p seem to be stage dependent. In acute TBI, overexpression of miR‐9‐5p alleviated neuroinflammation and blood–brain barrier damage via suppression of Ptch1, a key component of the Hedgehog signaling pathway (Wu et al. 2020). Another study found similar results in the subacute phase of TBI, with overexpression of miR‐9a‐5p reducing TBI‐induced tissue damage via direct suppression of Elavl1 and a resulting downregulation of VEGF (Feng et al. 2024), whose overexpression has been shown to exacerbate TBI (Wu et al. 2022; Yüksel et al. 2013; Zhou et al. 2020). Conversely, in the chronic phase after TBI, inhibition of miR‐9‐5p improved neurological outcomes by promoting astrocyte proliferation and release of astrocyte‐derived neurotrophic factors. Here, miR‐9‐5p inhibition in astrocytes was proposed to promote neuronal synaptic remodeling via regulation of the extracellular matrix protein THBS2 (Wu et al. 2021).

Evidently, miR‐9‐5p plays multifaceted roles in neurodegenerative diseases through complex modulation of autophagic pathways, neuroinflammation, mitochondrial function, and synaptogenesis.

4.2.4. miR‐125b‐5p (Mixed Directionality)

We identified miR‐125b‐5p as dysregulated in 17 unique studies and eight different diseases, albeit with mixed directionality (Table 1). Further, its expression was enriched in the nervous system relative to blood or immune cells (Figure 6).

In experimental models, miR‐125b‐5p has been reported as neuroprotective, with its overexpression rescuing neurons from oxygen and glucose deprivation‐induced death and improving neuronal integrity in a rodent model of stroke (Dong et al. 2024). These neuroprotective effects were attributed to targeting of the acid‐sensing ion channel 1 (Asic1), an important mediator of acidosis‐induced neuronal injury. Interestingly, EV‐encapsulated miR‐125b‐5p was able to reproduce these neuroprotective effects in cellular and rodent models of ischemic injury (Dong et al. 2024). Similarly, in a rat model of AD, administration of a miR‐125‐5p mimic suppressed neuronal apoptosis, reduced levels of pro‐inflammatory proteins, and improved performance in behavioral measures of learning and memory (Ren et al. 2024). Accordingly, overexpression of miR‐125b‐5p in cellular models of AD suppresses apoptosis, alleviates Aβ‐induced oxidative stress, and reduces expression of pro‐inflammatory factors. These in vitro effects were partially attributed to targeting of BACE1, encoding an enzyme required for generation of toxic Aβ (Ren et al. 2024; Li, Xu, et al. 2020). MiR‐125‐5p has also shown neuroprotective effects in a rodent model of neuropathic pain, suppressing inflammation via regulation of the transcription factor Sox11 (Wang et al. 2024).

Conversely, conflicting evidence suggests a dichotomous role for this miRNA in different experimental contexts. In the mouse prefrontal cortex, overexpression of miR‐125b increased tau phosphorylation and reduced dendritic spine density, which was attributed to targeting of Ncam1 and downstream regulation of GSK3B. Interestingly, the same study identified miR‐125b‐5p as significantly upregulated in the prefrontal cortex of mice with dementia (Zhang et al. 2019). Conflicting results have been observed in mouse neuroblastoma cells, with inhibition of miR‐125b‐5p inducing tau hyperphosphorylation by reducing translational repression of protein phosphatase methylesterase‐1 (Pme1), an enzyme indirectly regulating tau phosphorylation (Zhao et al. 2021). In cultured microglia, overexpression of miR‐125b‐5p following lipopolysaccharide (LPS) treatment was found to promote M1 polarization and expression of pro‐inflammatory cytokines. Neurons treated with conditioned media from miR‐125b‐5p‐overexpressing microglia exhibited increased apoptosis (Li, Fan, et al. 2024). Opposing results have also been reported, with overexpression of miR‐125b‐5p suppressing M1 polarization and promoting M2 polarization in LPS‐treated microglial cells (Wang et al. 2024). Further investigation of its contribution to neurodegenerative mechanisms may clarify the utility of miR‐125b‐5p as a functional biomarker.

4.2.5. miR‐127‐3p (Mixed Directionality)

We identified miR‐127‐3p as dysregulated across 11 different diseases (Table 1), with enriched expression in the nervous system relative to blood or immune cells (Figure 6).

While we identified its directionality in circulating EVs as mixed, miR‐127‐3p is frequently downregulated in tissues of the CNS in the context of neurodegenerative diseases. Indeed, downregulation of miR‐127‐3p has been reported in post‐mortem brain tissue from human stroke (Loppi et al. 2021), PD (Hoss et al. 2016) and multiple system atrophy (Wakabayashi et al. 2016), well as in rodent models of TBI (Puhakka et al. 2017) and MS (Juźwik et al. 2018).

Functionally, overexpression of miR‐127‐3p promotes neuronal loss and axonal degeneration in the injured spinal cord (He et al. 2016), while its inhibition reduces neuronal autophagy in the hypoxic–ischemic cortex, promotes neurite outgrowth in cultured spinal neurons, and alleviates hypoxia‐induced injury in cortical neurons (He et al. 2016; Zhang et al. 2021). These effects were mediated by miR‐127‐3p targeting of Cisd1, encoding a mitochondrial membrane protein with implications in neuroinflammation, stroke, and PD (Geldenhuys et al. 2025, 2017; Li et al. 2025; Saralkar et al. 2021; Martinez et al. 2024; Bitar et al. 2024; Wang, Li, et al. 2025). Interestingly, in a rodent model of encephalopathy, upregulation of EV‐associated miR‐127‐3p was conserved at the level of the brain, CSF, and plasma, supporting its potential as a circulating biomarker for neuropathology (Xiao et al. 2024).

4.3. Consistency in miRNA Regulation Across CSF, Plasma and Serum

We also summarized expression of the most frequently dysregulated miRNAs across different biofluids (Figure 7). CSF is in direct contact with the brain and spinal cord, and its molecular composition most accurately reflects brain pathophysiology. Blood‐based biomarkers, however, have rapidly progressed as proxies for CSF, bypassing limitations associated with accessibility, invasiveness, and repeated sampling (Duits et al. 2016; Hazan et al. 2024). When comparing miRNA expression between CSF and blood derivatives, we found that CSF and plasma were often correlated, whereas CSF and serum were predominantly anticorrelated. Further, when comparing plasma and serum, most miRNAs showed conflicting directionality. Blood coagulation, involved in serum preparation, is known to induce exogenous EV release from activated platelets, contaminating the serum EV profile (Zhang et al. 2022). Platelet‐derived EVs, and consequently miRNAs, may therefore be at the root of inconsistencies between serum, plasma, and CSF. Indeed, concentrations of platelet‐associated EVs and miRNAs are significantly higher in serum than in plasma, emphasizing the importance in choice of biofluid (Wang et al. 2012).

5. Conclusion

In preparing this systematic review, we have made available a repository of miRNA expression data from the available literature on fluid‐derived EVs in neurodegenerative diseases (Table S1). This data can be used as a resource for a variety of applications, including comparing miRNA expression between neurodegenerative diseases of interest based on existing data in the literature, comparing in‐house miRNA expression data to patient cohorts available in the literature, or evaluating miRNA regulation in specific contexts (e.g., treated versus untreated patients or early versus advanced disease). Together, our analysis reinforces that despite interstudy heterogeneity, trends in EV‐derived miRNA regulation are detectable across a range of neurological disorders, supporting their potential as non‐invasive biomarkers in neurodegenerative diseases.

Author Contributions

Aliyah Zaman: writing – original draft, writing – review and editing, conceptualization, methodology, visualization. Sienna S. Drake: conceptualization, visualization, methodology. Alyson E. Fournier: conceptualization, writing – review and editing, supervision, resources, funding acquisition.

Funding

This work was supported by MS Canada, Canadian Institutes of Health Research, Fonds de recherche du Québec‐Santé and Myelin Repair Foundation.

Conflicts of Interest

During the review process for the manuscript, Alyson Fournier was Deputy Editor‐in‐Chief at Journal of Neurochemistry. The authors declare no conflicts of interest

Supporting information

Data S1: Supporting Information.

Table S1: Extracted microRNA expression data from publications included in review.

Zaman, A. , Drake S. S., and Fournier A. E.. 2026. “Extracellular Vesicle‐Derived microRNAs as Fluid Biomarkers in Neurodegenerative Diseases: A Systematic Review.” Journal of Neurochemistry 170, no. 1: e70323. 10.1111/jnc.70323.

Data Availability Statement

All data analyzed in this review was extracted from previously published manuscripts (references provided in Data S5). All extracted data that supports the findings of this review are provided in Table S1.