Integrated microfluidic platforms for extracellular vesicles: Separation, detection, and clinical translation

Abstract

Extracellular vesicles (EVs), secreted by most living cells, encapsulate a diverse array of bioactive molecules from their parent cells, including proteins and nucleic acids. Recent studies underscore the potential of EVs as advanced biomarkers for the early diagnosis of a variety of clinical diseases. Nevertheless, traditional platforms for EVs separation and detection platforms working alone often involve multiple pieces of equipment and complex, multi-step protocols. This extends processing time and the likelihood of bioanalyte loss and cross-contamination, thereby impeding further EVs research. To date, few studies have effectively combined EVs separation, detection, and analysis functions into a single platform. Integrated microfluidic platforms present a compelling solution by enabling seamless progression from sample to result. These platforms can efficiently combine various separation and detection techniques, simplifying complex workflows and facilitating both efficient EVs separation and high-sensitivity detection. This review concentrates on integrated microfluidic platforms for EVs separation and detection, specifically examining whether the separation and detection units are fully integrated. Recent studies underscore the potential of EVs as promising biomarkers for early-stage diagnosis of diseases, including cancer and neurodegenerative disorders. Recent advances in EVs separation and analysis enable overcoming key translational barriers, accelerating their routine adoption in clinical diagnostics.

I. INTRODUCTION

A. EVs: Biogenesis and clinical potential

Extracellular vesicle (EVs; with diameters ranging from 30 to 150 nm) represent a specialized subtype of EVs, which are membrane-enclosed entities synthesized by living cells via the processes of endocytosis, fusion, and exocytosis.¹ EVs are categorized into apoptotic vesicles (>1 μm), microvesicles (100 nm to 1 μm), and EVs (30 to 150 nm).^2,3 EVs are instrumental in the stabilization and transfer of cellular cargo, encompassing specific nucleic acids, functional proteins, and lipids, thereby enhancing intercellular communication.⁴ The biogenesis of EVs is a multifaceted process: initially, the plasma membrane undergoes invagination to generate early sorting endosomes (ESEs). These ESEs subsequently mature into late sorting endosomes (LSEs), which further invaginate to form multivesicular bodies (MVBs).^5,6 Ultimately, MVBs are released into the extracellular milieu, transforming into EVs⁷ (Fig. 1). EVs are abundantly found across various biological fluids, including blood, urine, saliva, breast milk, lymphatic fluid, and cerebrospinal fluid.⁸ Recent investigations reveal that EVs are significantly more prevalent in the blood of cancer patients compared to healthy individuals. They are implicated in pivotal processes such as oncogenesis, disease progression, metastasis, and therapeutic monitoring.⁹ For instance, pancreatic cancer patients show elevated levels of glypican-1 EVs in blood, correlating with tumor burden.¹⁰ Moreover, EVs have emerged as biomarkers for evaluating therapeutic responses in cancer treatment and serve as indicators in liquid biopsies,¹¹ underscoring their potential as highly effective instruments for minimally invasive diagnostics and early screening in preventive healthcare.

FIG. 1.

Biogenesis process of EVs and schematic diagram of the molecular composition of EVs.

B. Limitations of conventional EVs analysis

A range of conventional techniques for the analysis of EVs, including separation, imaging/morphological characterization, molecular profiling, and clinical application, have been progressively developed. However, these methods still face notable limitations that hinder their broad clinical translation. Achieving high-purity and efficient EVs separation from complex biological samples remains a major challenge in the separation stage. Although ultracentrifugation is widely regarded as the gold standard,¹² it is time-consuming and labor-intensive, requires large sample volumes and expensive equipment, and may inadvertently activate platelets or immune cells,¹³ thereby compromising EVs purity and downstream analyses. For imaging and morphological characterization, high-resolution techniques such as transmission electron microscopy (TEM) and scanning electron microscopy (SEM) require elaborate sample preparation and are unsuitable for high-throughput applications.¹⁴ Alternative methods such as dynamic light scattering (DLS) and nanoparticle tracking analysis (NTA) are commonly used to measure particle size and concentration, but they lack sufficient resolution to distinguish heterogeneous EVs populations.¹⁵ In the molecular analysis stage, EVs surface markers (e.g., CD9, CD63, and CD81) are typically identified using Western blotting or enzyme-linked immunosorbent assay (ELISA).¹⁶ Although these methods provide specificity, their sensitivity is limited, and quantitative performance remains suboptimal. In summary, although conventional EVs separation and detection methods are valuable for basic research, their reliance on complex instrumentation and multi-step workflows limits their suitability for clinical diagnostics, which require high throughput, speed, and cost-effectiveness.

C. Microfluidics as an integrated solution

Microfluidic technology, as an integrated and multifunctional platform, spans and optimizes the entire workflow of EVs analysis and is increasingly emerging as a powerful tool to overcome the limitations of conventional methods.^17–19 These limitations underscore the need for integrated platforms that combine high purity, sensitivity, and scalability—criteria increasingly met by microfluidic technologies. In the separation stage, microfluidic devices can achieve high-throughput and highly specific EVs extraction through physical, chemical, or immunological mechanisms.^20,21 For imaging and morphological characterization, microfluidic chips are typically fabricated from transparent materials and designed to be compatible with various microscopy systems, enabling in situ visualization and analysis of EVs.²² At the molecular analysis level, microfluidic platforms can integrate multiple sensing modules for the rapid and highly sensitive detection of functional biomolecules such as EV-associated proteins and RNAs.^23,24 Finally, in clinical applications,²⁵ microfluidic systems offer advantages such as operational simplicity, low sample volume requirements, and cost-effectiveness, demonstrating strong potential for clinical translation. While microfluidics offers high throughput and integration, challenges remain in standardization and manufacturing scalability. Nevertheless, its ability to combine separation (acoustics, immunoaffinity) with on-chip detection (electrochemical sensors) makes it uniquely suited for clinical translation.²⁶ The ability of microfluidics to integrate separation, detection, and analysis on a single chip makes it a highly promising and practical platform for rapid EVs analysis, addressing the diverse needs of both fundamental research and clinical diagnostics. Some comprehensive reviews have outlined microfluidics-based technologies for EVs separation and detection.^27–29 In this review, we place particular emphasis on the integrated microfluidic platforms designed for EVs analysis. We discuss distinct objects of EVs analysis, with a focus on comprehensive EVs analysis, detailed analyses of EVs' proteins and RNA are also provided, using the frameworks of the “combined microfluidic platform” and the “shared microfluidic platform” as foundational models.

II. MICROFLUIDIC EVs SEPARATION AND DETECTION STRATEGIES

Microfluidics entails the precise manipulation of microscale fluid volumes within intricately designed micro- and nanostructured channels.³⁰ This advanced technology facilitates the seamless convergence of sample preparation, analytical detection, and diagnostic operations onto a singular microchip. Boasting key advantages such as minimal sample consumption, accelerated separation and purification processes, enhanced sensitivity, and superior separation efficiency,^7,31 microfluidics has rapidly established itself as a formidable tool in the realm of biomedical research.

A. Microfluidic EVs separation strategies

1. Microfluidic EVs separation strategies

Microfluidics has made significant strides in EVs applications, praised for its efficiency in separation and detection.³² Yet, these claims often overlook challenges such as low throughput, inconsistent performance across biological samples, and scalability issues. Despite advantages like miniaturization and automation, microfluidics faces ongoing problems with reproducibility, complex integration, and the need for specialized expertise. Having said that, these microfluidic technologies effectively leverage both the physical characteristics of EVs, such as size and density, and their chemical features, including nucleic acids and surface proteins, by using microchip platforms to facilitate efficient EVs separation and detection.^33,34 As a promising alternative to conventional methods, microfluidic-based approaches address current limitations through innovative integration strategies. The combination of techniques such as viscoelastic flow,³⁵ acoustofluidics,³⁶ dielectrophoresis,³⁷ and magnetic field-based microfluidic systems is anticipated to greatly enhance the efficiency of EVs separation,³⁸ enabling high-throughput, label-free separation of EVs directly from complex biological samples (Fig. 2). Table I summarizes various microfluidic-based strategies for exosome separation.

FIG. 2.

A schematic illustration of strategies for the separation and detection of EVs. The separation techniques, such as inertial focusing, viscoelastic fluid, filtration, acoustofluidics, dielectrophoresis (DEP), and magnetic field, are employed to separate EVs from various sample types. For detection, methods like fluorescence detection, surface plasmon resonance (SPR) detection, colorimetric detection, surface-enhanced Raman scattering (SERS), mass spectrometry-based detection, and electrochemical detection assays are utilized to further identify and analyze the separated EVs.

TABLE I.

Summary of microfluidic EVs separation strategies.

Method	Throughput (μl/min)	Recovery (%)	Scalability	Clinical feasibility	References
Inertial focusing	1.67	52.8 ± 4.53	High throughput and robustness; membrane-free system; ease of operation	Efficient submicrometer particle separation; integration with clinical testing	39
Viscoelastic fluids	3.34	87	Suitable for large-scale production; progress steadily	High recovery and purity; low operational cost and ease of use; high repeatability	40
Filtration	2.4	81	Low sample volume; high sensitivity and efficiency; quantification on a single chip	Direct EVs separation from blood; quantification of disease markers; fingertip-based testing	41
Acoustofluidics	10	99	Automation and high reproducibility; continuous flow configuration; adjustable separation parameters	Speed and high efficiency; preservation of EVs integrity; minimal user intervention	36
Dielectrophoresis (DEP)	30–50	NA	Rapid separation and recovery process; no sample dilution required; use of whole blood	Minimally invasive technology; on-chip biomarker detection; faster results for diagnostics	42
Magnetic separation	100	NA	Enhanced separation efficiency; high sensitivity and low contamination; reversibility and expandability	Efficient separation of T-EVs; minimized contamination; integration with mass spectrometry	43

Inertial focusing, a specialized form of hydrodynamic focusing, leverages the interplay between Dean forces and inertial lift forces exerted on particles of varying sizes within microfluidic channels under the influence of Newtonian fluids.^44,45 The equilibrium of these forces generates distinct equilibrium positions along the curved channel, enabling size-dependent sorting of particles.^46,47 Tay et al. developed a spiral microchannel system for the rapid purification of submicrometer particles directly from whole blood, achieving a recovery rate of 52.8 ± 4.53% for particles smaller than 1 μm, with a reported resolution approaching 1 μm.³⁹ To further improve scalability, the authors later introduced an inertial microfluidic system based on previous work, enabling the direct separation of nanoscale EVs (exosomes, 50 to 200 nm) and medium-sized EVs (microvesicles, 200 nm to 1 μm) from whole blood.⁴⁸ That said, its sensitivity to nanoscale particles is limited due to inherently weak inertial forces at that scale. As a result, effective separation of small EVs remains challenging and often requires extended channel lengths to generate adequate Dean flow.⁴⁹ This trade-off between efficiency and system complexity raises concerns about scalability, particularly for clinical applications requiring rapid, high-throughput EVs processing.

Viscoelastic fluids enable particle separation by exploiting the elastic lift forces generated by the medium's viscoelastic properties,^50–53 typically composed of synthetic polymers such as polyvinylpyrrolidone (PVP), polyethylene oxide (PEO), and polyacrylamide (PAA).⁵⁴ Kim et al. first demonstrated the use of viscoelastic flow in straight microchannels to focus 500 and 200 nm particles, marking a key step toward EVs separation using this mechanism, despite the inability to focus 100 nm particles.⁵⁵ Building on this, Meng et al. developed a cascaded viscoelastic microfluidic device that enabled high-purity (>97%) and high-recovery (>87%) separation of small extracellular vesicles (sEVs, <200 nm) directly from whole blood.⁴⁰ The device achieved a 100 nm cutoff size and a throughput of 3000 μl/min, enabling rapid and nondestructive EVs extraction. While promising, viscoelastic microfluidics still faces challenges, including limited fluid stability, reduced reproducibility under high-throughput conditions, and poor scalability. Future efforts should aim to improve flow stability, control particle trajectories more precisely, and facilitate clinical translation through chip-level optimization.

Filtration leverages membranes with precisely defined pore sizes to selectively retain particles larger than the membrane pores, while permitting smaller particles, such as EVs, to pass through, thereby enabling size-based separation.^56–58 This method can be classified into three primary types: membrane filtration, column filtration, and ultrafiltration (UF).^59–61 Chen et al. developed a microfluidic platform for EVs separation from whole blood using dual membrane filtration, incorporating a 0.2 μm polycarbonate membrane to enhance performance.⁴¹ The system achieved over 99% sEV separation efficiency and an 81.1% recovery rate from just 2 μl of blood. Despite its high separation efficiency, the device faced limitations such as membrane clogging and potential EVs damage. To overcome these issues, Li et al. introduced a centrifugal microfluidic system combining dual nanofilters with pneumatic pulse filtration. This pulsatile flow reduced clogging and particle deposition, enabling EVs separation within 30 min and yielding 91% diagnostic accuracy for early breast cancer.⁶² Filtration-based methods offer simplicity and high purity but typically require external forces such as centrifugation, pressure, or vacuum to drive fluids through membrane pores. As samples concentrate, membrane fouling becomes more likely, reducing efficiency. Additionally, dependence on external actuation hinders integration into clinical workflows.

Acoustofluidics integrates external acoustic fields with microfluidic frameworks to actively sort particles by size,^63–65 offering significant advantages in both precision and biocompatibility.^66–68 Wu et al. developed an acoustofluidic device employing tilted surface acoustic waves (TSAWs) for continuous-flow exosome separation, achieving a recovery rate of approximately 90%.⁶⁹ They later introduced a rapid, label-free two-module system, where the first module removed blood cells while maintaining an EVs recovery rate above 99%, and the second module purified EVs with a reported purity of 98.4%.³⁶ An additional acoustic module was subsequently integrated to further improve performance. These results underscore the potential of acoustic fields for efficient, high-purity EVs separation. However, challenges remain, including frequent air bubble formation within the acoustic field and high system costs, which may hinder clinical translation. Improving device stability, reducing system complexity, and validating clinical applicability in large-scale studies are essential next steps.

Dielectrophoresis (DEP) refers to the migration of particles in a non-uniform electric field, where differential forces induce particle or cellular polarization, generating a net dipole moment on the particle's surface.⁷⁰ This polarization allows the particles to move either along or counter to the electric field, contingent upon factors such as excitation frequency, particle size, and electrode configuration.³⁷ Lewis et al. employed an alternating current electrokinetic (ACE) microarray chip to separate EVs via DEP. In this system, EVs were concentrated in high-field regions around circular microelectrodes, while larger cells were excluded into low-field zones. The device enabled direct EVs enrichment from 30 to 50 μl of whole blood without prior sample processing.⁴² It was later optimized for in situ immunofluorescent labeling and surface protein analysis on-chip.¹⁰ Although the method showed high enrichment efficiency, the recovery rate was not reported, raising concerns about overall yield and sample loss. DEP is appealing for its ability to selectively manipulate EVs using tunable electrical parameters, such as frequency and voltage. Nonetheless, it still faces challenges, including low throughput and the requirement for low-conductivity buffers, which may limit compatibility with biological fluids. Improving scalability and operational stability will be essential for translating DEP-based EVs systems into clinical use.

Magnetic separation involves the capture of EVs through an immunoaffinity reaction, wherein magnetic beads (MB) bind to antibodies. This process utilizes magnetic forces generated by a magnetic field to manipulate the particles effectively.^71,72 Niu et al. developed the develop a fluid multivalent magnetic interface (FluidmagFace) in a microfluidic chip, a system that employs affinity magnetic beads (MB) for efficient EVs separation and proteomic analysis. Magnetic separation allows precise manipulation of MB-bound targets, improving separation efficiency by 13.9% compared to non-bivalent binding methods.⁴³ This technique effectively and selectively captures EVs from plasma samples with high sensitivity. While magnetic separation offers excellent affinity and selectivity, its effectiveness can be affected by external interference, limiting reliability in complex biological samples. Although the technique demonstrates promising sensitivity and specificity, overcoming the impact of environmental factors remains a critical challenge for its broader clinical implementation.

Building on previous research, the proposed platform demonstrates considerable promise for clinical applications, particularly in areas such as noninvasive liquid biopsy and personalized medicine. By enabling the detection and analysis of tumor-derived EVs (T-EVs) in blood, the platform facilitates early cancer diagnosis and continuous monitoring of treatment responses.⁴³ Moreover, it offers the potential to identify biomarkers associated with drug resistance, thereby aiding in the customization of cancer treatments. Beyond oncology, the system's ability to separate and analyze circulating microparticles (MPs) provides a valuable diagnostic tool for assessing vascular health, offering insights into conditions such as atherosclerosis and diabetes-related cardiovascular risks.^73,74

From a clinical adoption perspective, while acoustofluidics requires a higher initial investment due to the specialized equipment involved, it offers substantial long-term cost savings.⁷⁵ Its automated, label-free operation reduces the need for consumables and manual labor, making it more cost-efficient over time compared to techniques like ultrafiltration, which are burdened by ongoing costs for membrane replacements and maintenance. As the system scales for high-throughput applications, the per-test cost of acoustofluidics becomes more favorable, further enhancing its potential for widespread clinical use. This cost comparison underscores the economic advantages of acoustofluidics, positioning it as a promising solution for clinical adoption in both oncology and vascular health diagnostics.⁷⁶ Current comparisons favor older benchmarks; emerging alternatives require direct validation. Multicenter reproducibility studies remain imperative before cost advantages can be deemed generalizable.^77,78

2. Purity metrics in EVs separation strategies

A significant challenge in microfluidic-based EVs separation is the lack of standardized metrics for assessing EVs purity, which hinders the ability to directly compare the performance of different platforms. Purity is typically quantified by evaluating the presence of contaminants, such as lipoproteins and soluble proteins, which may co-isolate with EVs during the separation process.⁷⁹ However, the methods used to assess purity vary widely across studies, affecting the reported purity values and their interpretability. For instance, acoustofluidic separation techniques have reported a purity of 98.4%, typically by excluding lipoproteins from the analysis.³⁶ This step is crucial for ensuring the specificity of the isolated EVs population. However, some studies fail to fully detail the specific methods employed to exclude lipoproteins and address the potential impact of other co-isolated components.^80,81 A more thorough explanation of the purity quantification process, including the methods used to remove lipoproteins and non-EVs contaminants, is essential for ensuring accurate purity assessments.

In contrast, the viscoelastic flow-based method reports a purity of greater than 97% yet does not account for the potential co-isolation of soluble proteins.⁴⁰ While soluble proteins are not directly associated with EVs, their presence can influence the overall purity of the isolated EVs.⁸² This issue complicates the interpretation of purity measurements and highlights the need to consider all potential contaminants when evaluating EVs separation techniques.

Microfluidic technologies have demonstrated significant potential in efficiently separating EVs from complex biological fluids.⁸³ Nonetheless, a major challenge remains in achieving high purity of the separated EVs, particularly due to the co-separation of contaminants such as lipoproteins and soluble proteins. These contaminants can compromise the integrity and specificity of the EVs population, thus undermining the reliability of subsequent analyses. Therefore, addressing these sources of contamination is essential for enhancing the performance of microfluidic EVs separation techniques. Several microfluidic approaches, including acoustofluidics and viscoelastic flow-based methods, have been developed to improve EVs separation purity.^26,84 Acoustofluidics can effectively minimize lipoprotein contamination by adjusting the acoustic field parameters to preferentially capture EVs while reducing the co-separation of larger lipid particles.⁸⁵ However, despite achieving high purity by excluding lipoproteins, acoustofluidic systems may still face difficulties in separating EVs from soluble proteins, which share similar size and density characteristics.⁸⁶ To further mitigate this issue, incorporating additional steps such as affinity-based capture or size-exclusion filtration may be necessary to reduce soluble protein contamination.

B. Microfluidic EVs detection strategies

We reviewed microfluidic strategies for EVs separation, emphasizing that accurate downstream detection relies on EVs being separated with high purity.⁸⁷ Various EVs detection techniques have been used in disease prognosis and therapeutic monitoring. Microfluidic methods such as fluorescence assays, SPR, SERS, mass spectrometry, and electrochemical sensing offer benefits in miniaturization and integration (Fig. 2). Nevertheless, they often require expensive equipment and high reagent volumes and may lack sensitivity for detecting low-abundance biomarkers.^88–92 Nonetheless, rapid advances in microfluidic technology are driving its use in EV-based diagnostics. These platforms show strong potential for high-throughput, sensitive detection from small sample volumes. Future work should improve robustness, sensitivity, and clinical adaptability while addressing standardization and reproducibility to enable broader clinical use.

Fluorescence detection, known for its high precision and sensitivity, offers a rapid and efficient method for EVs detection, particularly when combined with microfluidic platforms.⁸⁷ EVs are typically captured on microfluidic chips and labeled with fluorescent dyes or quantum dots.^93–95 The resulting signals are analyzed via fluorescence microscopy or flow cytometry, with their presence and intensity serving as diagnostic indicators.⁹⁶ Chinnappan et al. developed an aptamer-based magnetic biosensor using magnetic nanobeads functionalized with anti-CD63 aptamers. By integrating immunomagnetic separation (IMS), the system enhanced fluorescence detection of EVs, with CD63 serving as the target marker. Carbon-coated magnetic beads adsorbed FAM (6-Carboxyfluorescein)-labeled ssDNA aptamers, resulting in a strong fluorescence signal and a detection limit of 1457 particles/ml.⁹⁷ Fluorescence-based methods enable rapid and sensitive EVs detection and hold promise for in situ analysis of clinical samples. However, their clinical translation is limited by nonspecific binding, high cost, and complex procedures. Additionally, the small size and heterogeneity of EVs challenge labeling consistency and quantification. Quantum dots (QDs) offer excellent photostability but pose toxicity risks due to heavy metals like cadmium and lead, limiting their clinical use.^98,99 Biodegradable alternatives, such as fluorescent nanoparticles from organic dyes or biocompatible polymers, provide similar optical properties with enhanced safety profiles. These materials can also be engineered to degrade in specific biological environments, improving biocompatibility.¹⁰⁰ Exploring these options could lead to safer, more reliable biosensing platforms for applications like EVs analysis and biomarker detection.

Surface plasmon resonance (SPR) is an advanced optical biosensing technology that exploits the phenomenon of total internal reflection of light at the interface between a prism and a metal film.¹⁰¹ This interaction generates a diminishing wave within a photophobic medium, alongside a plasmonic wave present in the medium.¹⁰² Luo et al. developed a portable SPR-based EVs sensing platform that integrates microfluidics with a high-performance trilayer structure: a gold mirror, SiO₂ spacer, and nanoporous gold sensing layer. The SiO₂ layer served as an optical cavity, and the sensor surface was functionalized with CD63 or EpCAM aptamers for EVs detection. The average bulk sensitivity values for the Si/Au hole, Au/Au hole, and CIMH sensors were 316, 391, and 431 nm/RIU, with corresponding FOM values of 1.93, 12.1, and 29.2, respectively.¹⁰³ This platform achieved a limit of detection of 6 × 10⁵ particles/ml, demonstrating its potential for ultrasensitive EVs analysis. While SPR offers advantages in label-free, real-time detection for clinical diagnostics, its widespread use is hindered by the complexity, cost, and size of current systems. To unlock its full clinical potential, miniaturization, system integration, and simplified operation are critical.

Colorimetric detection is a method for identifying exosomes through a color change in solution, leveraging the simplicity of operation and signal visualization. This approach is integrated with a microfluidic platform to enhance its efficiency and practicality.^104,105 Chen et al. developed a ZnO nanowire-coated scaffold chip for colorimetric detection of EVs using TMB, achieving a detection limit of 2.2 × 10⁷ particles/ml.¹⁰⁵ While simple, colorimetric detection often requires complementary methods to enhance sensitivity. Vaidyanathan et al. introduced a tunable AC hydrodynamic technique that utilizes nanoshear forces to improve exosome capture and detection, with a fivefold increase in efficiency, detecting over 2.76 × 10⁶ particles/ml.¹⁰⁶ That said, both methods face challenges, including reaction conditions, matrix interference, and nonspecific binding. Improvements in scalability, robustness, and reproducibility are crucial for clinical application, with future efforts focusing on high-throughput, clinical-grade platforms.

Surface-enhanced Raman scattering (SERS) is a sophisticated vibrational spectroscopy technique that amplifies Raman signals through plasmon excitation on irregular metal surfaces, predominantly gold (Au) or silver (Ag).¹⁰⁷ SERS offers highly sensitive detection of low-concentration analytes,^108–110 and can be seamlessly integrated into microfluidic systems for high-throughput and precise analysis.¹¹¹ To achieve highly sensitive detection of EVs, Lee et al. developed an SERS substrate with gold nanopillars that generate hotspots, enabling precise miRNA detection of EVs at detection limits over 100 times lower than other methods; the detection sensitivity range is 1 aM to 100 nM.¹¹² This was accomplished using a locked nucleic acid probe. While SERS offers exceptional sensitivity and label-free detection, challenges persist, including unstable signal enhancement, background noise, and high equipment costs. Future efforts should focus on improving signal stability, reducing noise, optimizing cost and scalability, and developing more robust substrates with better system integration for clinical applications.

As an integral component of proteomics, mass spectrometry offers several advantages in bioassays, including high throughput, exceptional specificity, remarkable sensitivity, and cost-effectiveness.^113,114 Consequently, it has been extensively employed in the analysis of EVs' proteins and amino acids. Shan et al. developed a matrix comprising gold nanoparticles (AuNPs) and cellulose nanocrystals (CNCs) for the direct analysis of full proteins in serum-derived EVs, effectively addressing ion suppression caused by protein aggregation.¹¹⁵ This method reduced the detection limit to 0.01 mg/ml, enhancing sensitivity, automation, and cost-efficiency, and enabling the detection of a wide range of proteins with excellent reproducibility. The specificity of this detection method reaches 83.2%. However, the high cost of AuNPs, the complexity of sample preparation, and the challenges associated with data analysis hinder their broader clinical adoption. Future advancements should focus on streamlining sample preparation, lowering costs, and simplifying data analysis to facilitate more widespread clinical application.

Electrochemical detection relies on electrochemical reactions that convert sample concentration and structure into measurable electrochemical potentials.^116,117 This method is straightforward, efficient, and highly selective, allowing for the quantification of EVs through alterations in electrochemical signals. Jiang et al. developed a poly(dopamine) (PDA)-assisted aptamer-based DNA microelectrode sensor for electrochemical EVs detection. The PDA deposition enhances the surface complexity, significantly improving both sensitivity (81.9%) and precision. Utilizing electrochemical impedance spectroscopy (EIS), the sensor achieves a detection limit of 1.39 × 10² particles/ml (approximately 14 EVs) within 180 min, enabling single-particle detection.¹¹⁸ While this method shows great promise for early cancer diagnosis, challenges remain, particularly with electrode lifespan and the potential economic burden of frequent replacements. Future research should focus on extending the electrode lifespan, enhancing system integration, and addressing reproducibility issues to facilitate broader clinical adoption.

Meanwhile, emerging techniques, such as digital ELISA and nanopore-based EVs counting, are transforming the field of EVs detection by providing exceptional sensitivity and resolution. Digital ELISA, with its attomolar sensitivity, enables the detection of low-abundance biomarkers within complex biological samples, making it particularly advantageous for early disease diagnosis and continuous monitoring.^95,119–122 In contrast, nanopore-based EVs counting offers single-particle resolution, allowing for precise enumeration of individual EVs and the detailed characterization of their heterogeneity.¹²³ The integration of these advanced technologies with microfluidic platforms has the potential to overcome current limitations in both sensitivity and scalability. By combining the high-throughput capabilities of microfluidics with the superior sensitivity of digital ELISA and the single-particle resolution of nanopore-based counting, it becomes possible to significantly enhance the efficiency and accuracy of EVs analysis.^124,125 This integration could thus facilitate more reliable diagnostic tools and enable more effective personalized medicine applications.

III. INTEGRATED MICROFLUIDIC PLATFORM FOR EVs ANALYSIS

Building upon the previous systematic review of strategies for the separation and detection of EVs, significant advancements have been made in the methodologies employed for EVs separation and analysis. The microfluidic separation platform has evolved to enhance EVs extraction, achieving both high yield and exceptional purity, while detection modalities continue to diversify. Nevertheless, many studies tend to apply separation and detection techniques independently.^126,127 This fragmented approach complicates experimental protocols and significantly increases both time and costs, ultimately hindering research efficiency and limiting practical applicability.¹²⁸ Recently, many researchers have shifted their focus toward integrating separation and detection systems, leading to significant advancements in both sample preparation and direct analytical evaluation. As summarized in this review, current integration strategies can be primarily categorized into two paradigms: the “combined microfluidic platform” and the “shared microfluidic platform” (Fig. 3). The combined microfluidic platform uses a serial configuration of separation, enrichment, and detection units interconnected via microchannels or pipelines, facilitating the continuous operation of both enrichment and detection processes. In contrast, the shared microfluidic platform eliminates the need for an additional detection system, enabling enrichment and detection functionalities to operate on a single platform, supporting in situ and real-time EVs analysis. Furthermore, depending on the specific requirements for subsequent EVs analysis, detection methods across various integrated platforms can be further classified into comprehensive content analysis and bioactive substance profiling, which include the assessment of proteins and nucleic acids. Table II provides a comparison of combined microfluidic platforms and shared microfluidic platforms for EVs. Evaluating long-term operational costs of the combined and shared microfluidic platforms involves factors like consumables, maintenance, and scalability. The combined platform, with its modular design, incurs higher consumable and maintenance costs due to additional reagents and frequent calibration. This can limit its feasibility in resource-limited settings. In contrast, the shared platform has lower initial consumable costs but requires careful optimization to prevent process interference. Maintaining high performance over time may lead to higher maintenance costs.

FIG. 3.

A schematic representation of combined and shared microfluidic platforms built on integrated microfluidic systems. In the combined microfluidic platform, the separation and detection regions are mechanically linked via microchannels or tubing, enabling continuous operation. EVs are first separated in the separation region and then directed downstream to the detection region. In contrast, the shared microfluidic platform lacks a distinct detection area; here, EVs are both separated and captured within the separation region, where detectors are directly applied for in situ analysis of the EVs.

TABLE II.

Comparison of combined microfluidic platforms and shared microfluidic platforms for EVs.

Type	Merits	Demerits	Clinical suitability
Combined microfluidic platforms	High flexibility; multi-step optimization; precise analysis	Increased complexity; higher cost; slower throughput	Suitable for detailed; high-precision analyses; less ideal for rapid, point-of-care diagnostics
Shared microfluidic platforms	Streamlined; cost-effective; real-time analysis	Lower flexibility; limited advanced detection options	Ideal for rapid; point-of-care diagnostics and real-time monitoring

A. Combined microfluidic platforms for EVs analysis

Following the separation process of tumor-derived EVs, the integrated system typically evaluates these EVs based on three key biophysical and biochemical characteristics: quantification (EVs count) and the analysis of bioactive molecules. A prevalent method for counting EVs involves correlating their accumulation on the chip with the quantification of fluorescence intensity.^95,129 Furthermore, individual EVs can be encapsulated and separated for subsequent counting analyses. The application of microfluidics technology in the detection of individual EVs holds significant promise, particularly when integrated with advanced analytical techniques such as digital droplet technology [including digital PCR (dPCR) and digital ELISA] and nanoarray technology.^95,119–122 Over the past few years, numerous scholars have dedicated themselves to the development of combined microfluidic platforms.^130–132 Table III provides an overview of various combined microfluidic platforms.

TABLE III.

Comparison of combined microfluidic platforms for EVs.

Platform	Separation technique	Detection technique	Sample	Limit of detection (LOD)	Work time	References
ExoPCD chip	Y-shaped microcolumn array	Electrochemical detection	Serum	4.39 × 103 particles/ml	3.5 h	133
Apta-magnetic biosensor	Immunomagnetic bead separation	Fluorescence detection	Cell culture supernatant	1457 exosomes/ml	NA	97
ExoELISA	Droplet-encapsulated single exosomes	Fluorescence counting	Cell culture supernatant	10 exosomes/μl	NA	95
EXID system	Serpentine microchannel	Microcolumn immunocapture assay	Blood	10.76 exosomes/μl	2 h	134
EVLET	Vibrating membrane filtration (VMF)	Thermophoretic amplification	Plasma	4.1 × 105 EVs	100 min	135
ExoSD chip	Immunomagnetic nanoparticles (IMNPs) separation	Fluorescence detection	Serum	NA	NA	71
ExoDEP chip	Microsphere-mediated dielectrophoretic separation	Fluorescence detection	Cell line supernatant	193 exosomes/ml	NA	130
PS-ED chip	Inertial focusing	Antibody capture	Whole blood	95 particles/μl	NA	132
FEMC	Filtration	Electrochemical detection	Serum	1 × 104 particles/ml	1 h	136
HiMEc	Inertial focusing	Electrochemical detection	Plasma	1 × 104 to 1 × 108 EVs/ml	NA	137
Nanomixing-microchip and SERS barcoding system	Nanomixing-microchip	SERS detection	Serum	NA	NA	138
Microfluidic-SERS method	Filtration	SERS detection	Plasma	2 EVs/μl	5 h	139
Nanochannel array membrane and immunocapture chip	Filtration	Immunocapture detection	Serum	NA	1.5 h	140
Indium-tin-oxide (ITO) sensor	Multi-orifice flow-fractionation (MOFF) channel	Electrochemical detection	Plasma	15 EVs/μl	NA	141
Double-filtration unit and photonic crystal (PC)	Filtration	Fluorescence detection	Serum	8.9 × 103 EVs/ml	NA	142
PCR-free integrated microfluidics platform	Acoustofluidics	Concentration/sensing detection	Plasma	1 pM	30 min	24
Sample treatment and miRNA quantification module chip	Micromixer separation	dPCR-based miRNA quantification	Blood	11 copies/ml	4.3 h	143

1. Combined platforms of overall EVs levels analysis

The comprehensive analysis of overall EVs levels is a crucial step in assessing the distribution and functionality of EVs in vivo. This analysis encompasses not only the absolute quantification of EVs but also their relative abundance and trends of variation across different biological samples.^144–146 Commonly employed techniques include NTA, DLS, and FACS, all of which effectively provide insights into the size distribution and concentration of EVs. Microfluidic technologies have recently gained attention as complementary tools for EVs quantification, offering advantages such as low sample volume requirements, integration of multiple functions, and potential for automation.¹⁴⁷ However, their sensitivity often varies with device design and surface chemistry, leading to inconsistent results. Furthermore, complex fabrication and operational requirements hinder standardization and reproducibility.

The ExoPCD chip developed by Xu et al. offers a promising strategy for integrated EVs separation and electrochemical detection, achieving a low sample requirement (30 μl) and a detection limit of 4.39 × 10³ particles/ml for CD63-positive EVs. Its two-stage design, which combines Y-shaped microcolumns for enrichment with a downstream electrochemical sensing region, enables in situ analysis and improves workflow efficiency.¹³³ Tim4-PS-based magnetic enrichment further enhances specificity by reducing nonspecific capture. Yet, the platform's reliance on antibody-functionalized magnetic beads and a custom chip structure may pose challenges in terms of fabrication, cost, and reproducibility. Its focus on single-marker detection also limits adaptability compared to multiplexed or label-free platforms. Future improvements should aim to reduce processing time, expand marker coverage, and simplify device architecture. Incorporating machine learning-based signal analysis could further enhance robustness and clinical applicability. Chinnappan et al. introduced an “apta magnetic biosensor” platform featuring anti-CD63 aptamers immobilized on the surface of magnetic nanobeads [Fig. 4(a)]. The described method combines flow-through magnetic separation with immunomagnetic separation (IMS) for EVs separation, using a rotating magnet assembly system (rMAS) and aptamer-based fluorescence detection.⁹⁷ While the system achieves a detection limit of 1457 EVs/ml and its sensitivity is promising, there are concerns about scalability and reproducibility. Additionally, while the rMAS and microfluidic channels improve automation, the complexity and cost of the system hinder its clinical adoption. Future developments should focus on improving scalability, robustness, and specificity, while simplifying the system for broader clinical use.^124,125 Liu et al. developed a droplet-based microfluidic platform, termed droplet digital ExoELISA, for ultrasensitive immunoassays and single-EV counting [Fig. 4(b)]. By encapsulating CD63 antibody–conjugated magnetic beads within droplets and labeling breast cancer-derived EVs with GPC-1, the system achieves a detection limit as low as 1 × 10⁴ EVs/ml (∼10⁻¹⁷ M), demonstrating strong potential for early cancer diagnostics.⁹⁵ Despite its high sensitivity, the platform faces notable challenges. The droplet generation process and magnetic bead manipulation introduce operational complexity and may affect reproducibility across laboratories. Additionally, reliance on antibody specificity can lead to cross-reactivity or inconsistent binding in heterogeneous samples. Compared to label-free or continuous-flow systems, droplet-based methods may also struggle with scalability due to precise flow control requirements. Future efforts should aim to streamline droplet handling, validate performance across diverse clinical samples, and incorporate multiplexing to broaden diagnostic applications. Lu et al. developed the EXID system, an integrated microfluidic platform combining serpentine channels and a microcolumn array for efficient EVs separation and detection [Fig. 4(c)]. Using anti-PD-L1-labeled immunomagnetic beads and quantum dot (QD)-based single-bead fluorescence analysis, the system enables multiplexed EVs profiling with a detection limit of 10.76 EVs/μl and completes analysis within 2 h.¹³⁴ This approach shows strong potential for tumor subtype classification and guiding immunotherapy. Nonetheless, the device's structural complexity may hinder large-scale manufacturing and workflow integration. Additionally, QD-based detection, while enhancing signal resolution, requires careful calibration and poses potential concerns regarding photostability and biocompatibility. The system's reliance on antibody specificity may also limit its reproducibility across heterogeneous EVs populations. To sum up, without addressing these technical and translational challenges, the promising analytical capabilities of current microfluidic platforms risk remaining confined to research settings.

FIG. 4.

The combined platform was used for overall EVs content analysis. (a) An “apta magnetic biosensor” platform that utilizes anti-CD63 aptamers immobilized on the surfaces of magnetic nanobeads. Reprinted with permission from Chinnappan et al., Biosens. Bioelectron. 220, 114856 (2023). Copyright 2023 Elsevier.⁹⁷ (b) The droplet digital ExoELISA microfluidic platform for single EVs counting and immunoassays. Reprinted with permission from Liu et al., Nano Lett. 18(7), 4226–4232 (2018). Copyright 2018 Authors, licensed under a Creative Commons Attribution License.⁹⁵ (c) Immunomagnetic beads labeled with anti-PD-L1 fluorescent probes capture EVs, which are then separated within a serpentine channel before entering the microcolumn analysis area for single-bead analysis. Reprinted with permission from Lu et al., Biosens. Bioelectron. 204, 113879 (2022). Copyright 2022 Elsevier.¹³⁴

2. Combined platforms of EVs bioactive substances analysis

The analysis of bioactive substances within EVs is a critical step in elucidating their roles in cellular communication, disease progression, and therapeutic interventions.^148,149 The biomolecules transported by EVs, including proteins, nucleic acids (such as mRNA and miRNA), and glycans, participate in a myriad of physiological and pathological processes.^150,151 Consequently, both quantitative and qualitative analyses of these molecules have become fundamental to understanding the functional dynamics of EVs. Currently, some research groups have utilized thermophoresis to target EVs glycans as detection objects, thereby breaking away from conventional methods of EVs detection and analysis [Fig. 5(a)].¹³⁵

FIG. 5.

The combined platform was used for EVs bioactive substances analysis. (a) The lectin-based thermophoretic method (EVLET) streamlines the processes of vibrational membrane filtration (VMF) and thermophoretic amplification on a microfluidic platform for the analysis and detection of EVs glycans in cancer diagnostics. Reprinted with permission from Li et al., Nat. Commun. 15(1), 2292 (2024). Copyright 2024 Nature.¹³⁵ (b) The dual-filter electrochemical microfluidic chip (FEMC) platform facilitates on-chip separation of EVs and continuous in situ electrochemical analysis of surface proteins. Reprinted with permission from Wang et al., Biosens. Bioelectron. 239, 115590 (2023). Copyright 2023 Elsevier.¹³⁶ (c) An integrated platform for EVs separation and detection combining helical microfluidic channels with electrochemical devices (HiMEc). Reprinted with permission from Kwon et al., Biosens. Bioelectron. 267, 116792 (2025). Copyright 2025 Elsevier.¹³⁷ (d) A microfluidic platform with integrated size filtration for SERS analysis of plasma from osteosarcoma patients. Reprinted with permission from Han et al., Biosens. Bioelectron. 217, 114709 (2022). Copyright 2022 Elsevier.¹³⁹ (e) A microfluidic platform for EVs detection using integrated SAW lysis without PCR and chemical reagents. Reprinted with permission from Ramshani et al., Commun. Biol. 2(1), 1–9 (2019). Copyright 2019 Nature.²⁴

a. EVs' protein analysis.

The in-depth investigation of EVs proteomics lays a foundational framework for the discovery of potential EVs assays, which can be employed for early disease diagnosis and therapeutic monitoring.¹⁵² By comparing the EVs' proteome of healthy individuals with those suffering from various diseases, researchers have identified characteristic proteins associated with disease progression, thereby informing the development of personalized treatment strategies.¹⁵³

Wang et al. developed a dual-filter electrochemical microfluidic chip (FEMC) that integrates EVs separation and in situ detection into a streamlined workflow [Fig. 5(b)]. By combining dual filtration with multiplexed screen-printed electrodes (SPEs) functionalized with MB@UiO-66 and specific antibodies, the platform enables rapid detection of multiple tumor markers (PMSA, EGFR, CD81, and CEA) from blood samples.¹³⁶ It achieves a detection limit of 1 × 10⁴ particles/ml within 1 h, offering a 100-fold sensitivity improvement over traditional ELISA. Despite its advantages, the platform faces practical challenges. The use of multiple SPEs and pre-functionalized materials increases fabrication complexity and cost, potentially limiting scalability. Moreover, while MB@UiO-66 enhances signal output, it may introduce background noise if not properly controlled across diverse biological fluids. Although effective for protein markers, the platform's applicability to other biomolecules like RNA or lipids remains unverified, which could restrict its diagnostic versatility. Kwon et al. developed the HiMEc platform, integrating helical microfluidic channels with electrochemical sensors for efficient EVs separation and detection [Fig. 5(c)]. By combining size-based sorting with antibody-mediated lipoprotein capture, the system removes over 83% of lipoproteins and achieves an EVs recovery rate above 87%. Quantification of PD-L1 and PD-1 on EVs showed signal intensities 2.3× and 1.2× higher than those obtained via ultracentrifugation and SEC (size exclusion chromatography), respectively, with a detection range from 1 × 10⁴ to 1 × 10⁸ EVs/ml.¹³⁷ Despite its strengths, HiMEc's structural complexity may limit scalability and standardization. Furthermore, its current design focuses on protein biomarkers, with limited validation for nucleic acid detection. While it offers enhanced purity and sensitivity compared to conventional methods, further optimization is needed to improve usability, broaden analytical targets, and ensure reproducibility across clinical samples. To enhance detection sensitivity and efficiency, integrated microfluidic platforms utilizing SERS offer notable advantages,^108–110 enabling high-throughput and sensitive detection of low-abundance analytes.¹¹¹ Wang et al. developed a nano-hybrid microchip integrated with SERS to detect four melanoma-related protein biomarkers directly from 5 μl of serum, without the need for extensive EVs pre-enrichment.¹³⁸ The system employs a sandwich immunoassay format and achieves a signal amplification of 3.7- to 4.2-fold, enhancing both sensitivity and specificity while streamlining the workflow. Despite these advantages, SERS-based detection remains sensitive to substrate variability and hotspot reproducibility, potentially affecting consistency. Additionally, the platform's current application is limited to protein biomarkers, with unproven adaptability to other EVs contents such as miRNAs or lipids, which may constrain its broader diagnostic utility. Han et al. developed a SERS-integrated microfluidic platform with side filtration to quantify EVs biomarkers, CD63, vimentin (VIM), and EpCAM, in plasma from osteosarcoma patients [Fig. 5(d)]. The system achieved a detection limit of 2 EVs/μl using just 50 μl of plasma and completed analysis within 5 h, demonstrating strong diagnostic performance (100% sensitivity, 90% specificity, and 95% accuracy).¹³⁹ These results highlight the platform's potential for accurate osteosarcoma classification and the utility of SERS in multiplexed EVs detection. Having said, that challenges remain for clinical translation. The reproducibility of Raman signal enhancement is limited by nanostructure uniformity and hotspot variability. Additionally, the high diagnostic accuracy was based on a small sample size and requires validation in broader patient populations. The use of multiple biomarkers also adds assay complexity, which may hinder standardization and scalability in clinical workflows. Gurudatt et al. developed an electrochemical microfluidic biochip that integrates a porous multi-orifice flow-fractionation (MOFF) channel with an indium-tin-oxide (ITO) sensor for rapid and sensitive detection of epithelial and mesenchymal markers on EVs, enabling evaluation of the epithelial–mesenchymal transition (EMT) index in pancreatic cancer.¹⁴¹ With a detection limit of 15 EVs/μl and a clinical quantification threshold of 50 EVs/μl, the platform delivers results within 5 min and effectively distinguishes IPMN (intraductal papillary mucinous neoplasm) from SCA samples, highlighting its diagnostic and commercialization potential. Yet, several limitations remain. EMT-based assessment, while biologically meaningful, can be affected by tumor heterogeneity and may not generalize across cancer types. The integration of MOFF and ITO components may complicate fabrication and hinder scalability. Moreover, the system's performance in complex clinical samples, particularly those with interfering biomolecules, requires further validation. These platforms often incorporate structurally complex components, such as dual filters or nanostructured substrates, which hinder scalability and increase production costs. Advancing clinical translation will require simplifying designs, improving manufacturability, and validating performance across diverse sample types.

b. EVs' RNA analysis.

In addition to the widespread utilization of proteins in the clinical evaluation of EVs, nucleic acids also play an indispensable role. The nucleic acids found within EVs, such as mRNA and miRNA, closely resemble the expression profiles of their originating cells, thereby providing valuable insights into the biological states and functions of those cells.^131,154,155 Consequently, the transfer of nucleic acids between disparate cells via EVs' carriers is pivotal not only for intercellular communication but also significantly influences the onset and progression of various diseases.

Ramshani et al. developed a PCR- and chemical-free microfluidic platform integrating surface acoustic wave (SAW)-based lysis with an enrichment film sensor, enabling absolute quantification of EV-derived miRNAs (e.g., miR-21) in plasma or serum [Fig. 5(e)]. The system achieves a 1 pM detection limit from just 20 μl of sample within 30 min, with <10% uncertainty, offering high analytical precision and streamlined operation.²⁴ Even though this approach represents a major advance in EVs RNA analysis due to its integration and speed, several limitations persist. The efficacy of SAW-based lysis may vary with sample composition, affecting reproducibility, and its utility across broader miRNA panels or more complex fluids remains to be fully assessed. Compared to fluorescence or thermophoretic systems, this method reduces procedural complexity and cost but may require further calibration to ensure consistent lysis across diverse EVs subtypes. Subsequently, Sung et al. developed an integrated microfluidic platform that automates EVs separation, miRNA-21 extraction, reverse transcription (RT), and digital PCR (dPCR) for ovarian cancer (OvCa) diagnostics.¹⁴³ The system operates without an external pump, streamlining workflow and enhancing suitability for point-of-care use. It achieves a detection limit of 11 copies/ml with quantification inaccuracy under 12%, outperforming conventional qPCR in sensitivity and precision. Despite its analytical strength, the platform has limitations. Its reliance on dPCR demands specialized, costly equipment, which may limit clinical scalability. Additionally, the integration of multiple processing steps could introduce variability, affecting reproducibility across diverse samples. Compared to PCR-free systems like SAW-based platforms, this method offers greater accuracy but with increased complexity and resource demands. Future research should focus on miniaturizing dPCR components, reducing processing time, and validating performance across larger, heterogeneous cohorts.

As EVs become increasingly important in clinical diagnostics, particularly for biomarker detection like miRNA and mRNA, efficient RNA analysis methods are essential.^156,157 Microfluidic platforms offer significant advantages over traditional plate-based assays, including improved sensitivity, scalability, and operational efficiency.¹⁵⁸ Unlike plate-based methods, which are limited by sample volume and long processing times, microfluidics enables rapid, high-throughput analysis with minimal sample waste, crucial in clinical settings. By integrating processes such as EVs separation, RNA extraction, and detection into a single, closed system, microfluidic platforms reduce contamination risks and human error.¹⁵⁹ Their precise control over reaction conditions enhances RNA quantification accuracy, particularly for low-abundance biomarkers. Additionally, microfluidic devices are ideal for point-of-care applications, offering faster and more practical solutions than conventional laboratory assays. Overall, microfluidics provides superior sensitivity, automation, and efficiency, making it an excellent choice for scalable EVs RNA analysis in clinical diagnostics.

B. Shared microfluidic platforms for EVs analysis

Distinct from the previously summarized integrated platforms for separation and detection, the platforms in question merge the enrichment and detection units into a singular chip structure, eschewing the need for additional detection areas.^130,160 These platforms primarily incorporate finely designed microchannels for the capture of EVs, typically employing immunoaffinity capture methods on the inner surfaces of these channels, thus enhancing both the accuracy and efficiency of early clinical detection. To simultaneously achieve enrichment and detection without an independent detection area, these integrated platforms leverage the unique characteristics of various material interfaces, introducing diverse measurement technologies that facilitate separation and enrichment while enabling timely in situ detection to address varying experimental requirements.^161,162 Table IV presents a summary of various shared microfluidic platforms.

TABLE IV.

Comparison of shared microfluidic platforms for EVs.

Platform	Separation technique	Detection technique	Sample	Limit of detection	Work time	References
MoSERS microchip	MoS2 nanocavities	SERS	Blood	1.23%	30 min	163
TIRF	Anti-CD63 capture	TIRF microscope fluorescence detection	Plasma	NA	NA	164
PMMA and a nanoporous gold (Au) nanocluster	AuNC membrane capture	SPR	Urine	1000 particles/ml	NA	165
nPLEX	Nanopore capture	SPR	Ascites	3000 exosomes (670 aM)	NA	166
Exodisc	Filtration	ELISA	Urine	NA	1 h	104
ACE microarray chip	Microelectrode chip capture	Fluorescence detection	Blood	NA	90 min	10
DNA cage-based thermophoretic assay	DNA cage selective recognition	Thermoelectrophoresis	Serum	2.05 fM	NA	167
DTTA	RET-based DNA tetrahedron (FDT) label EVs	Thermoelectrophoresis	Serum	14 aM	NA	168
DNA three-way junction biosensor	Multicolor DNA biosensor label EVs	Fluorescence detection	Serum	0.116 g/ml	NA	169
Immunobiochip	Immunocapture	Fluorescence detection	Serum	NA	4 h	170

1. Shared platforms of overall EVs levels analysis

To facilitate the comprehensive counting of EVs and real-time monitoring of total content on the same platform, researchers are dedicated to developing integrated analytical systems aimed at enhancing detection efficiency and ensuring data consistency.¹⁷¹ Such systems not only provide absolute counts of EVs but also allow for the real-time monitoring of overall EVs' content within samples, thereby offering more precise and comprehensive information to support clinical diagnostics and research. Current overall EVs analysis predominantly emphasizes the total count of patient-specific EVs, alongside the evaluation of therapeutic efficacy and strategies for disease prevention related to these individual EVs.¹⁷² However, this approach may overlook the intricate complexities of EVs' composition and their diverse functional roles in various biological contexts.¹⁷³ A more refined strategy that incorporates both qualitative and quantitative assessments of EVs could yield deeper insights into their involvement in disease mechanisms and therapeutic responses, ultimately advancing personalized treatment approaches. Single EVs analysis relies on detection methods focused on individual EVs.¹⁷⁴ Microfluidic technology is particularly well suited for this purpose due to its high efficiency, accuracy, and minimal sample requirements.¹²³

To enhance the sensitivity and efficiency of single-EV detection, Jalali et al. developed a multi-channel microfluidic device embedded with MoSERS nanocavity microchips [Fig. 6(a)]. These chips, composed of silver, zinc oxide, and molybdenum disulfide (MoS₂), enable precise EVs separation and SERS-based detection in small sample volumes (<10 μl).¹⁶³ The system achieved 87% diagnostic accuracy and identified glioblastoma (GBM) mutations in blood samples from 12 patients, with sample preparation completed in just 30 min. Though the platform demonstrates strong potential for clinical diagnostics, especially in oncology, its complexity and reliance on nanomaterial fabrication may challenge scalability and routine clinical adoption. Although the MoSERS platform shows strong potential for single-EV diagnostics, particularly in GBM detection, challenges remain in reproducibility, fabrication complexity, and clinical integration. Future work should aim to improve material consistency, streamline device manufacturing, and validate performance across varied patient groups and EVs subtypes. He et al. developed a single-vesicle imaging platform using total internal reflection fluorescence (TIRF) microscopy for direct, quantitative analysis of tumor-derived EVs from as little as 1 μl of plasma [Fig. 6(b)].¹⁶⁴ The system captures EVs via antibodies bound to activated aptamer probes, triggering a hybridization chain reaction (HCR) targeting PTK7-exon and amplifying fluorescence signals for single-EV detection. While the platform offers high spatial resolution and sensitivity, its clinical applicability is limited by the need for specialized TIRF equipment and reliance on specific gene targets, which may reduce its generalizability across cancer types. Compared to other single-EV approaches such as MoSERS or electrochemical systems, this method excels in imaging precision but falls short in throughput and scalability. Future improvements should focus on simplifying system integration, broadening biomarker compatibility, and enhancing usability in clinical settings. Yang et al. developed an integrated microfluidic device combining polymethyl methacrylate (PMMA) with nanoporous gold (Au) nanocluster films for EVs separation and in situ detection from patient urine samples [Fig. 6(c)]. EVs are first captured on the AuNC membrane via antibodies and then hybridized with gold nanorods (AuRs), forming a sandwich complex that amplifies plasmonic scattering signals. This system achieved a detection limit below 1000 particles/ml, demonstrating strong potential for noninvasive cancer diagnostics.¹⁶⁵ Despite its sensitivity and compact integration, the device faces challenges related to the fabrication of nanostructured gold surfaces and the complexity of dual-antibody binding, which may affect reproducibility in heterogeneous fluids. Compared to TIRF-based or electrochemical methods, this plasmonic approach offers enhanced signal strength but may sacrifice scalability and operational simplicity. Future efforts should aim to simplify fabrication, improve reproducibility, and ensure compatibility with point-of-care applications, potentially through automation or AI-assisted analysis.

FIG. 6.

The shared platform was used for overall EVs content analysis. (a) The multi-channel fluid device combines embedded array nanocavity microchips (MoSERS microchips) with an SERS detection platform for analysis. Reprinted with permission from Jalali et al., ACS Nano 17(13), 12052–12071 (2023). Copyright 2023 Authors, licensed under a Creative Commons Attribution License.¹⁶³ (b) A single-vesicle detection imaging platform based on total internal reflection fluorescence (TIRF) microscopy for the direct observation of tumor-derived EVs. Reprinted with permission from He et al., Anal. Chem. 91(4), 2768–2775 (2019). Copyright 2019 Authors, licensed under a Creative Commons Attribution License.¹⁶⁴ (c) An integrated microfluidic apparatus for separation and in situ detection of polymethyl methacrylate (PMMA) and nanoporous gold (Au) nanocluster films modified with trapping antibodies was constructed. Reprinted with permission from Yang et al., Biosens. Bioelectron. 163, 112290 (2020). Copyright 2020 Elsevier.¹⁶⁵

2. Shared platforms of overall EVs bioactive substances analysis

Recently, the advent of integrated analytical systems leveraging microfluidic platforms and nanotechnology has significantly enhanced the efficiency and accuracy of EVs' protein detection, making these systems particularly well suited for high-throughput screening of small-volume samples.¹⁷⁵ Microfluidic EVs protein detection predominantly employs immunocapture methodologies. This approach involves the selective capture of target proteins via antibody-functionalized microchannels or microbeads, facilitating a streamlined “one-stop” operation encompassing EVs capture, separation, and protein detection.¹⁷⁶

a. EVs' protein analysis.

To bolster the accuracy and repeatability of data derived from EVs protein analysis, Lee et al. introduced the nPLEX sensor, utilizing transmissive-based surface plasmon resonance (TSI) to enable sensitive, label-free EVs protein analysis through nanopore arrays. This approach improves accuracy and repeatability by tuning the cyclicality of nanopores and electromagnetic fields to detect distinct EVs proteins. While highly sensitive, its reliance on nanopore fine-tuning may limit scalability and application in complex biological matrices.¹⁶⁶ In a departure from traditional assays like ELISA, Woo et al. presented the Exodisc microfluidic platform, integrating two nanofilter sets for high-sensitivity EVs protein analysis in bladder cancer urine samples [Fig. 7(a)]. With an efficient 30-min separation process, it demonstrates impressive 95% separation efficiency and a 100-fold concentration of target proteins like GAPDH and CD9.¹⁰⁴ This platform stands out for its automation and potential in liquid biopsy applications, but its clinical adoption may face challenges related to sample variability and reproducibility in different patient populations. To separate and enrich EVs from whole blood for subsequent detection, Lewis et al. developed an AC electromotor (ACE) microarray chip device for separating EVs from small blood samples in just 20 min [Fig. 7(b)].¹⁰ This system achieved high sensitivity (99%) and decent specificity (82%) for biomarkers like glypican-1 and CD63. Nonetheless, it struggles with distinguishing EVs from other biomarkers, a limitation that may impact its clinical precision. The device exhibits limitations in its inability to effectively distinguish EVs from other high-content biomarkers, although it still facilitates a rapid workflow from a single drop of whole blood to the final detection result. Overall, these advancements show promise for liquid biopsy technologies, yet improvements in scalability, reproducibility, and specificity will be critical for broader clinical application.

FIG. 7.

The shared platform was used for EVs bioactive substances analysis. (a) Exodisc microfluidic platform for urinary EVs enrichment and separation in patients with bladder cancer. Reprinted with permission from Woo et al., ACS Nano 11(2), 1360–1370 (2017). Copyright 2017 Authors, licensed under a Creative Commons Attribution license.¹⁰⁴ (b) The device based on an alternating current electric (ACE) microarray chip directly separates and detects EVs in the whole blood of patients with pancreatic and colon cancer. Reprinted with permission from Lewis et al., ACS Nano 12(4), 3311–3320 (2018). Copyright 2018 American Chemical Society.¹⁰ (c) A DNA cage-based thermophoretic method enables highly sensitive in situ detection of microRNA within EVs. Reprinted with permission from Zhao et al., Angew. Chem., Int. Ed. 62(24), e202303121 (2023). Copyright 2023 Wiley VCH GmbH.¹⁶⁷ (d) The DNA Tetrahedron-Based Thermostrophy Assay platform enables in situ detection of EVs mRNA, facilitating the distinction between prostate cancer (PCa) patients and those with benign prostatic hyperplasia (BPH). Reprinted with permission from Han et al., Nano Today 38, 101203 (2021). Copyright 2021 Elsevier.¹⁶⁸ (e) An immunobiochemical chip that selectively captures tumor-associated proteins (such as EGFR and PD-L1) using antibodies and visualizes the fluorescence intensity of magnetic beads (MB) with total internal reflection fluorescence (TIRF) microscopy. Reprinted with permission from Yang et al., ACS Appl. Mater. Interfaces 10(50), 43375–43386 (2018). Copyright 2018 Authors, licensed under a Creative Commons Attribution License.¹⁷⁰

b. EVs' RNA analysis.

The shared microfluidic platform optimizes the operational workflow by effectively integrating the separation of EVs with the extraction and analysis of RNA, thereby significantly enhancing both the efficiency and accuracy of the analytical process. In contrast to traditional methodologies, the design of this shared integrated platform not only mitigates sample loss at each procedural step but also enhances the overall repeatability of the experiments through automated processing.

The identification of EVs' RNA mainly focuses on miRNAs and mRNA. Zhao et al. used molecular probes to detect miRNAs in EVs, advancing their role in cancer diagnostics [Fig. 7(c)].¹⁶⁷ The integrated platform for mRNA analysis minimizes losses and reduces processing time. While this approach shows promise for early cancer diagnostics, its effectiveness depends on the specificity and sensitivity of the probes, which can vary across different miRNA targets and biological samples. Han et al. developed a DNA tetrahedron-based thermophoretic assay (DTTA) for in situ mRNA detection within EVs, achieving a detection limit of 14 aM [Fig. 7(d)].¹⁶⁸ The platform combines FDT with thermal accumulation to differentiate between prostate cancer (PCa) and benign prostatic hyperplasia (BPH) based on PSA (prostate-specific antigen) mRNA expression in EVs. While it efficiently integrates EVs separation, FDT internalization, and signal assessment in a single microfluidic device, the reliance on FRET (förster resonance energy transfer) and thermophoretic techniques may pose challenges in scalability and reproducibility, limiting its broader clinical adoption. Wang et al. developed an integrated microdevice using a DNA three-way junction for in situ detection of multiple miRNAs in breast cancer patients. The device detects miR-21, miR-27a, and miR-375 with detection limits of 0.116, 0.125, and 0.287 μg/ml, respectively.¹⁶⁹ By enabling simultaneous quantification of miRNAs without the need for EVs cleavage or complex RNA processing, the device offers a streamlined approach. However, its sensitivity may be limited by hybridization specificity and interference from abundant non-target RNAs in biological samples. Yang et al. developed an immunobiochip for in situ detection of EVs' miRNAs using total internal reflection fluorescence (TIRF) microscopy.¹⁷⁰ The platform captures tumor-associated proteins (e.g., EGFR and PD-L1) and detects RNA via electrostatic interactions between a cationic lipid complex and EVs. This interaction activates molecular beacons (MB), allowing TIRF to measure RNA levels. While TIRF offers high sensitivity, its reliance on specialized equipment and challenges with complex biological matrices limit its practicality. While these platforms mark significant progress in EV-based RNA detection, their clinical application is limited by scalability, sensitivity, and methodological complexity. Future improvements should focus on simplifying the platforms, enhancing reproducibility, and optimizing performance in complex biological fluids. Additionally, integrating emerging technologies, like AI-driven data analysis, could address current analytical bottlenecks.

C. Challenges for clinical adoption

Despite the significant promise of combined and shared microfluidic platforms for EVs analysis, several challenges must be overcome to facilitate their clinical adoption. The combined microfluidic platform presents complexities in design due to its reliance on multiple interconnected units for separation, enrichment, and detection. This serial configuration can complicate system integration and increase costs, which may limit its scalability and practicality in clinical environments. Furthermore, ensuring consistent and reliable performance across all interconnected units, especially when processing diverse clinical samples, remains a substantial challenge. While the platform's flexibility allows for a broad range of analytical tasks, this versatility often comes at the cost of extended processing times, which could hinder its suitability for real-time, point-of-care applications. In contrast, the shared microfluidic platform offers distinct advantages, such as compactness and the ability to perform real-time, in situ analysis. Yet, it faces limitations in accommodating diverse sample types and handling complex workflows. The integration of both enrichment and detection steps within a single unit can compromise sensitivity, particularly when analyzing low-abundance biomarkers.¹⁶⁰ Additionally, the need to optimize both processes simultaneously can introduce potential interference, negatively affecting the quality of detection. This highlights the trade-off between convenience and performance, which must be carefully managed to ensure clinical reliability.¹⁷⁷

From a clinical perspective, both platforms must meet stringent requirements for accuracy, reproducibility, and user accessibility. To gain acceptance, these microfluidic systems must demonstrate sensitivity and specificity comparable to existing gold-standard methods, ensuring their reliability in real-world diagnostic settings.¹³² Moreover, challenges related to regulatory approvals, cost-effectiveness, and the ability to function effectively in point-of-care settings are critical considerations for their widespread adoption. Addressing these technical, logistical, and practical barriers is essential for unlocking the full potential of microfluidic technologies in clinical diagnostics.

IV. CONCLUSIONS

EVs have gained significant attention. These technologies offer notable advantages, including high automation, minimal sample volume requirements, and rapid processing speeds, making them highly promising for early diagnosis and disease monitoring. When compared to traditional methods, microfluidic platforms demonstrate superior sensitivity and lower detection limits. Nonetheless, despite substantial progress, several challenges and limitations continue to hinder their clinical translation. One of the primary barriers to widespread adoption is the issue of standardization. Currently, the design and operational conditions of microfluidic platforms often depend on specific laboratory settings, resulting in poor reproducibility across different laboratories. Furthermore, while microfluidic technologies excel in small sample volumes, scalability remains a challenge, particularly when applied to large-scale, complex biological samples. Sensitivity at low EVs concentrations also needs further improvement, which directly impacts their effectiveness in early-stage disease detection. Finally, regulatory hurdles impede the clinical application of these platforms, as the lack of standardized protocols and certification processes restricts their broader adoption. To overcome these limitations, future research should prioritize the standardization and simplification of microfluidic platforms. Developing more versatile and user-friendly chips will improve both reproducibility and scalability. Additionally, enhancing sensitivity for detecting low-concentration EVs and improving the handling of complex biological samples will be key to advancing their practical application. Additionally, leveraging emerging technologies, particularly AI-driven analytical methods, can accelerate data processing and interpretation, helping to resolve current analytical bottlenecks. By addressing these challenges, integrated microfluidic technologies have the potential to evolve into more precise and efficient tools for clinical diagnostics.

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflicts to disclose.

Ethics Approval

Ethics approval is not required.

Author Contributions

Yang Dai: Investigation (lead); Writing – original draft (lead); Writing – review & editing (lead). Yibo Cui: Investigation (equal); Writing – original draft (equal); Writing – review & editing (equal). Jinwen Li: Investigation (equal); Writing – original draft (equal); Writing – review & editing (equal). Piwu Li: Funding acquisition (equal); Resources (equal). Xiaowen Huang: Funding acquisition (equal); Resources (equal).

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding authors upon reasonable request.