Engineered Extracellular Vesicles for Targeted Paclitaxel Delivery in Cancer Therapy: Advances, Challenges, and Prospects

Abstract

Scope

Extracellular vesicles (EVs) have emerged as promising cell-free delivery vehicles for cancer therapy due to their inherent biocompatibility, low immunogenicity, and natural targeting capabilities. EVs derived from various cellular sources offer distinct advantages in drug-loading capacity and therapeutic effectiveness. However, their clinical application is limited by challenges such as poor cargo stability, potential immunogenicity, and off-target effects. These limitations necessitate further surface functionalization of EVs to optimize vesicle stability, targeting precision, and safety of pharmacological cargos. Paclitaxel (PTX), a first-line chemotherapeutic agent effective against multiple cancers, is limited by poor solubility and significant systemic toxicity, highlighting the need for targeted delivery systems.

Methods

A literature search was conducted to identify relevant articles published between 1993 and 2025. This review provides a comprehensive overview of EV biogenesis and cellular origins, highlighting recent advances in engineering strategies for PTX delivery. Current progress in employing engineered EVs for PTX delivery in both in vitro and in vivo cancer models, along with practical challenges and future directions in the clinical translation of EV-based PTX delivery, are discussed

Results

Preclinical studies demonstrate that engineered EVs can effectively encapsulate and deliver PTX to tumor sites, improving therapeutic outcomes while minimizing systemic side effects. Despite these advances, challenges remain in optimizing EV isolation, surface modification, PTX loading efficiency, and precise recognition of tumor cells.

Conclusion

Engineered EVs represent a promising platform for PTX delivery, combining targeted therapeutic potential with reduced systemic toxicity. Continued research to address technical and translational barriers will be critical for advancing EV-based PTX therapies toward clinical application.

Graphical Abstract

Introduction

Cancer remains a leading cause of mortality worldwide, primarily due to late detection, invasion, and poor prognosis. As reported by the World Health Organization, the global incidence of cancer was 18.1 million new cases with 9.6 million deaths in 2018, and it is estimated to reach 16.5 million deaths by 2040 [1]. Consequently, extensive research has focused on developing novel and effective anticancer treatments that offer advantages over traditional therapies (surgery, chemotherapy, immunotherapy, and radiotherapy) in combating cancer recurrence, metastasis, and treatment resistance [2]. Paclitaxel (PTX) is a commonly used chemotherapy agent for treating a wide range of human cancers. Despite its effectiveness, it often causes significant side effects and has limitations such as poor water solubility and substantial toxicity [3]. These challenges highlight the necessity for innovative drug delivery systems that can enhance the therapeutic efficacy of PTX while minimizing its adverse effects. Extracellular vesicles (EVs) are membrane-bound lipid components involved in the bulk transfer of cargos to a wide range of potential recipient cells. Various cell types can release EVs as local and systemic messengers, carrying biomolecules including oncogenic lipids, proteins, and nucleic acids [4]. The composition and content of EVs can vary based on their originating cells, physiological conditions, and environmental stimulation [5]. EVs have emerged as promising drug delivery vehicles due to their low immunogenicity, bypassing lysosomal degradation, and high biocompatibility [6]. The nanoscale size of EVs enables them to effectively target cancerous tissues through both active and passive mechanisms [7]. Passive targeting leverages the enhanced permeability and retention (EPR) effect, while active targeting involves the use of tumor-specific antigens or ligands to direct the EVs. Both approaches enhance the delivery of therapeutic agents, leading to increased efficacy and reduced toxicity in non-targeted tissues [8]. Various techniques have been developed for the effective isolation and purification of EVs, including ultracentrifugation, ultrafiltration, precipitation, size-exclusion chromatography (SEC), and immunoaffinity capture [9]. Nanotechnology advances are increasingly being utilized to engineer EVs by altering surface charge, conjugating various components such as peptides and antibodies, and loading cargo. Depending on their cellular origin, EVs present distinct surface biomarkers serving as ideal tools for functionalization [10–12]. Loading hydrophobic chemotherapeutic agents, such as PTX, curcumin, doxorubicin, cisplatin, and methotrexate, can be achieved through passive diffusion into EV membranes during co-incubation [13]. Conversely, hydrophilic drugs require active loading techniques, such as electroporation or sonication, to effectively penetrate the EV membrane [14]. PTX-loaded EVs (PTX-EVs) have demonstrated encouraging outcomes in targeting and treating various cancers, including lung, breast, and pancreatic tumors. PTX-EVs enabled precise control over drug release in response to specific internal or external stimuli and enhanced tumor targeting, potentially increasing treatment efficacy while reducing side effects [15]. Despite these advantages, several limitations, such as toxicity, poor biocompatibility, low distribution, and degradation, restrict the clinical application of PTX-EVs [16]. Addressing these challenges is crucial for realizing the full potential of EV-based systems for PTX delivery. This review provides an overview of the biogenesis and main sources of EVs, highlighting their potential as PTX delivery vehicles, and discusses strategies for engineering EVs to enhance PTX targeted delivery alongside preclinical studies evaluating the efficacy of PTX-EVs in in vitro and in vivo cancer models. By examining the intersection of engineered EVs and drug delivery, this review aims to provide insights into the practical barriers and future directions for developing more effective and targeted cancer therapies.

EVs Biogenesis

Based on their size and biogenesis, EVs are classified into several subtypes, including exosomes (50-100 nm), ectosomes (100-500 nm), microvesicles (100-1000 nm), and apoptotic bodies (400-1000 nm) (Fig. 1) [17]. The biogenesis of EVs generally occurs through two primary pathways: the endosomal sorting complex required for transport (ESCRT) machinery system or an ESCRT-independent pathway (Fig. 2). This machinery comprises four protein complexes: ESCRT-0, which recognizes ubiquitinated proteins; ESCRT-I and ESCRT-II, which form the membrane bud; and ESCRT-III. ESCRT-I, ESCRT-III, and related proteins such as Rab proteins, Alix, VTA-1, and VPS4 are essential for the formation of multivesicular bodies (MVBs) and the loading of functional cargoes into exosomes [18]. A section of an endosome-limiting membrane invaginates and buds into its own lumen to generate MVBs or intraluminal vesicles (ILVs) [19]. MVBs mature into late-sorting endosomes and can either merge with lysosomes for destruction or with the plasma membrane to release EVs into the extracellular matrix (ECM) [20]. Various Rab GTPases, including Rab11, Rab27a, Rab27b, and Rab35, regulate the trafficking, signaling, and biogenesis of MVBs and ILVs [21]. The ESCRT-independent biogenesis of EVs involves the inward invasion of the plasma membrane, leading to the creation of early-sorting endosomes containing cell surface proteins, fluids, and extracellular components [22]. Finally, cargo loading into exosomes is mediated by the CD63 tetraspanin family [23]. The biogenesis of microvesicles involves several mechanisms, including the outward blebbing and fission of the plasma membrane. This process is mediated by signaling pathways and external elements such as calcium influx or hypoxia. Calcium influx induces microvesicles release through actin-myosin contraction or by stimulating phospholipid redistribution, with phosphatidylserine and phosphatidyl-ethanolamine moving to the outer layer of the plasma membrane [24]. Hypoxia also facilitates microvesicles release through hypoxia-inducible factor-dependent expression of Rab-22A protein [25]. Ectosomes are released through the split or outward budding of the plasma membrane, driven by the translocation of phosphatidylserine to the outer leaflet of the plasma membrane [26]. The small GTPase RhoA, a key regulator of actin cytoskeletal reorganization, is linked to ectosome biogenesis in various cancer cells [27]. Additionally, neutral sphingomyelinase enzymes (nSMase) are essential for the shedding and budding of the plasma membrane, thereby controlling ectosome formation [28]. Overall, the diverse mechanisms underlying the biogenesis of EVs highlight the complexity and specificity of their formation, which is crucial for their function in various physiological and pathological processes.

Fig. 1.

Classification of extracellular vesicle (EVs) subtypes. a Exosomes are released by the fusion of multivesicular bodies (MVBs) with the plasma membrane. Exosomes are packaged in the late endosome and generated through this fusion process. b Microvesicles are generated by the outward budding of the plasma membrane and include ectosomes, membrane particles, and exosome-like vesicles. c Apoptotic bodies are produced by the budding of cells undergoing programmed cell death and are also considered components of EVs with varying sizes

Fig. 2.

Biogenesis of EVs. Endosomal sorting complex required for transport (ESCRT)-dependent and ESCRT-independent pathways of extracellular vesicles (EVs): exosomes are EVs released into the extracellular environment when multivesicular bodies (MVBs) fuse with the plasma membrane. In contrast, microvesicles (or microparticles) are EVs that bud directly from the plasma membrane. Besides the ESCRT pathway, EV production also involves ESCRT-independent mechanisms, which require lipid ceramide and neutral sphingomyelinase (nSMase). Inhibiting EV formation and release using nSMase inhibitors could offer a novel therapeutic approach for neurological disorders, including brain injury

Origins of EVs

EVs derived from various cellular sources offer distinct advantages for chemotherapeutic delivery in cancer therapy. The choice of EV origin significantly influences key parameters such as drug loading efficiency, targeting specificity, circulation time, and immunogenicity—making it a critical consideration in the development of EV-based drug delivery systems. Mesenchymal stem cells (MSCs), which can be isolated from diverse tissues such as adipose tissue, umbilical cord blood, liver, placenta, amniotic fluid, and dental pulp, are one of the most extensively used sources for EV production [29]. MSC-derived EVs (MSC-EVs), particularly those from bone marrow (BM-EVs), exhibit functional similarities to their parent cells and are being developed as potential "factories" for drug delivery, offering advantages such as ease of expansion and low immunogenicity, thereby improving the efficiency of drug delivery systems [30–32]. However, MSC-EVs pose risks such as horizontal gene transfer [30–32], as seen in cancer-derived EVs that transfer oncogenic deoxyribonucleic acid (DNA), potentially causing transient phenotypic changes and unexpected immune responses [33]. Immune cells—including natural killer (NK) cells, T cells, B cells, dendritic cells (DC), and macrophages—are valuable sources of EVs for combined chemotherapeutic and immunotherapeutic applications [34–36]. These immune cell–derived EVs (IC-EVs) are inherently suited to evade phagocytic clearance, thereby prolonging systemic circulation and enhancing drug delivery efficiency [37]. Macrophage-derived EVs (M-EVs) exhibit functional variability depending on the phenotype of their parental cells. M1 macrophage-derived EVs (M1-EVs) can enhance chemosensitivity in drug-resistant cancer cells, increasing the efficacy of chemotherapeutic agents. Naïve M-EVs, on the other hand, demonstrate unique abilities to penetrate the blood–brain barrier (BBB), likely through interactions between integrins (LFA-1, ICAM-1) and endothelial receptors (C-type lectin). PTX-loaded M-EVs have shown significantly enhanced cytotoxic effects against drug-resistant tumors, leveraging their natural targeting capacity and intracellular delivery potential [38]. DC–derived EVs (DC-EVs) are enriched in specific phospholipids such as C22:6 docosahexaenoic acid, which contributes to their antigen-presenting capability and anti-tumor activity [39]. EVs from tumor cells (TC-EVs) offer intrinsic tumor-targeting properties, making them promising candidates for delivering chemotherapeutics directly to tumor sites. TC-EVs also possess the ability to promote angiogenesis under hypoxic conditions, which may be harnessed for targeting hypoxic tumor microenvironments (TME). However, these EVs carry risks of promoting tumor progression and are often rapidly cleared from circulation, limiting their clinical utility [40, 41]. Red blood cell–derived EVs (RBC-EVs) represent an attractive, biocompatible platform for drug delivery. They are devoid of nuclear DNA, eliminating concerns related to genetic material transfer, and are amenable to large-scale production. These features make RBC-EVs a safe and scalable option for chemotherapeutic encapsulation [42]. Engineered cell lines such as human embryonic kidney (HEK293) cells are frequently employed to produce EVs with tailored features, including enhanced drug-loading capacities or reduced immunogenicity [43]. More recently, non-human sources like milk-derived EVs have gained attention for their suitability in oral drug delivery. These vesicles protect therapeutic agents during gastrointestinal transit and improve bioavailability. Remarkably, milk-derived EVs have demonstrated substantial drug-loading efficiencies—approximately 30% for PTX and 20% for 5-fluorouracil—highlighting their promise in oral chemotherapeutic applications [44–46,47]. Each EV source presents distinct advantages and limitations. Table 1 provides a comparative summary of EVs from different origins and their potential as drug delivery vehicles. Rational selection of the EV source—based on the therapeutic context, safety considerations, and physicochemical drug properties—is essential for optimizing EV-based strategies in cancer treatment

Table 1.

Various cellular sources of EVs

EV Source	Types	Surface biomarkers	Cargo	Target	advantages	Disadvantages	References
Cancer cells	Stem Cells	CD24, CD44, CD90, CD133	miRNAs, mRNAs, proteins (growth factors, signaling molecules)	Tumors	Tumor-specific targeting, potential for immune priming	Risk of promoting tumor growth due to endogenous cargo	[47]
	Non-Stem cell	CD24, CD44	Proteins, miRNA, mRNA	Various cancers	High cytotoxicity	Potential immunogenicity
MSCs Stem cells	BM-MSCs	CD73, CD90, CD 105	miRNAs, mRNAs, proteins (growth factors, anti-inflammatory cytokines)	Cancers, inflammation	Low immunogenicity, therapeutic effects	Oncogenic potential of source cells	[29, 48]
Immune cells	NK cells	CD56, CD16, NKG2D	Cytotoxic proteins (granzymes, perforin), miRNAs	Tumors	Enhanced tumor targeting and cytotoxicity	Limited persistence in circulation	[37]
	Macrophages	CD68, CD163, CD11b, CD14, F4/80 (in mice)	miRNAs, mRNAs, proteins (cytokines, chemokines)	Inflammation, tumors	Modulation of immune response	Potential for promoting inflammation	[49]
	T cell	CD3, CD4, CD8	miRNAs, mRNAs, proteins (cytokines, chemokines)	Tumors	Immune modulation	Risk of immune activation	[50]
	B cell	CD19, CD20, CD27	miRNAs, mRNAs, antigens	Tumors	Antibody-mediated targeting	Potential for unwanted immune responses	[51]
	DCs	CD11c, MHC class II, CD80	miRNAs, mRNAs, antigens	Tumors	Strong immune activation	Risk of overactivation leading to autoimmune issues	[52, 53]
	Neutrophil	CD15, CD66b	miRNAs, mRNAs, proteins (myeloperoxidase)	Inflammation	Rapid response to infection	Limited specificity for cancer	[54]
Nervous system cells	Glial cells	CD81, CD63, ALIX, GFAP	Proteins, miRNAs, lipids	Neurons, other glial cells	Potential for targeted delivery to brain tissue; crossing BBB	Challenges in large-scale production, purification, and targeting specificity	[55]
Embryonic cells	HEK293 kidney cells	CD9, CD81, CD44	Proteins, mRNAs, miRNAs	Various cell types	Easy to culture and produce EVs; potential for loading with drugs	May not reflect in vivo conditions; potential immunogenicity	[56]
Blood	RBC	CD47, Band 3, Glycophorin A	Hemoglobin, lipids, miRNAs	Various cell types	Abundant source; biocompatible; potential for oxygen delivery with drugs	Limited cargo loading capacity; potential for aggregation	[42]
	Plasma	CD61, CD41, CD31 (platelet-derived); CD14 (monocyte-derived)	Proteins, lipids, miRNAs, coagulation factors	Various cell types	Readily accessible; reflects overall physiological state	Complex mixture; difficulties in isolating specific EV types; potential for off-target effects	[57]
Urine	Mixed population	CD9, CD81, TSG101, uromodulin	Proteins, miRNAs, metabolites	Kidney cells, other cells	Non-invasive collection; contains diverse cargo	Low EV concentration; variability in composition	[58]
Saliva	Mixed population	CD9, CD81, CD63, α-amylase	Proteins, miRNAs, mRNAs, lipids	Oral mucosa, other cells	Non-invasive collection; easy access	Low EV concentration; enzyme activity may degrade cargo	[59]
Milk	Mixed population	CD9, CD81, CD63, MFG-E8	Proteins, miRNAs, lipids, growth factors	Intestinal cells, other cells	Abundant source; contains bioactive molecules; potential for oral drug delivery	Variability in composition; potential for degradation in the GI tract	[60]

Engineering Strategies for EVs

The therapeutic potential of unmodified EVs is often limited by factors such as inadequate targeting specificity, low stability, and inefficient drug delivery. To overcome these deficiencies and harness the full potential of EVs as therapeutic vehicles, nanotechnology advances are increasingly being applied for EV modifications [10]. Engineering EVs often includes modifying surface charge, incorporating targeting ligands and stimuli-responsive elements, and genetic engineering [61]. These advancements offer significant benefits, such as enhanced tumor targeting, leading to higher drug concentrations within tumor cells and reduced side effects [11, 12]. Furthermore, modification of EVs enables drug release in response to specific stimuli, both internal (e.g., pH changes) and external (e.g., temperature, magnetic fields), leading to increased targeting specificity and treatment efficacy [10]. The following sections will explore various engineering strategies and their applications in tailoring EVs for targeted drug delivery, with a focus on chemotherapeutic agents.

Click Chemistry

Click chemistry has emerged as a leading strategy for functionalizing EV surfaces, enabling precise covalent attachment of targeting ligands or therapeutic agents [62]. This approach employs a two-step process: (1) conjugation of alkyne groups to EV surfaces via condensation reactions (e.g., using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide-N-hydroxysuccinimide, EDC-NHS), followed by (2) copper-catalyzed azide-alkyne cycloaddition with azide-modified molecules to form stable triazole linkages. Click chemistry offers versatile approaches for modifying EVs, supporting a wide range of therapeutic and diagnostic applications. These include targeted delivery of chemotherapeutics and nanoparticles (NPs), as well as in vivo tracking of treatment progress through fluorescent labeling of EVs [63, 64]. This approach is particularly advantageous due to its mild reaction conditions, compatibility with both aqueous and organic solvents, and ability to preserve the native properties of EVs, such as size, morphology, and cellular uptake mechanisms, thereby offering significant improvements over traditional cross-linking techniques [65, 66]. While click chemistry provides a robust and versatile platform for EV functionalization, several practical barriers persist in translation to clinical application. The use of copper catalysts, although highly efficient, raises concerns regarding residual metal toxicity and the potential alteration of EV bioactivity. To address this, copper-free click reactions, such as strain-promoted azide–alkyne cycloaddition (SPAAC), are being explored for biocompatible, catalyst-free surface modification [67]. Furthermore, over-modification can also result in steric hindrance or masking of essential surface proteins, potentially reducing targeting efficacy [65].

Incorporation of Targeting Ligands

The incorporation of targeting ligands into EVs offers a strategic approach to improve drug delivery precision, particularly to challenging sites such as the brain. Surface modifications of EV proteins (e.g., CD63, Lamp2b) with cell-specific ligands enhance localized cargo delivery by leveraging natural or engineered targeting mechanisms. Peptide-based modifications have shown promising results in various applications. For example, the fusion of the rabies virus glycoprotein (RVG) peptide to Lamp2b enables brain-targeted EV delivery, bypassing the BBB through adsorptive transcytosis pathways [68]. Similarly, the conjugation of the c(RGDyK) peptide onto MSC-EVs significantly enhanced their therapeutic efficacy in preclinical models of ischemic stroke. Additionally, conjugation of magnetic molecules and endosomolytic peptides (L17E) onto EV surfaces enhanced their targeting ability. These surface modifications led to significant improvements in tumor accumulation and endosomal escape, enabling the efficient delivery of therapeutic cargoes in both in vitro and in vivo models [69]. Aptamers —short strands of DNA or ribonucleic acid (RNA) with high specificity and affinity for their targets—are increasingly being utilized in surface modifications of EVs [70, 71]. Despite these advancements, targeted synthetic NPs have often faced significant limitations in clinical applications, necessitating further investigations. For example, the use of external peptides like rabies RVG can provoke immune responses, potentially leading to accelerated clearance from circulation and associated immunotoxicity. Additionally, the incorporation of targeting ligands may compromise the structural integrity of EVs, thereby reducing their biocompatibility [72, 73]. The functionalization process itself presents challenges for clinical translation. It is typically complex and costly, making it difficult to standardize and scale up for widespread clinical use. Current methodologies often require genetic or metabolic engineering of donor cells before EV isolation, as traditional ligand functionalization techniques in nanomedicine can be detrimental to EVs. This limitation poses considerable challenges when attempting to introduce exogenous surface ligands to EVs derived from other sources, such as plasma or tissue samples.

Genetic Engineering

Genetic engineering represents one of the most advanced and widely utilized techniques for modifying EVs, allowing the incorporation of various therapeutic molecules, including nucleic acids (DNA, messenger ribonucleic acid (mRNA), micro-ribonucleic acid (miRNA), small interfering ribonucleic acid (siRNA), circular ribonucleic acid (circRNA), and long non-coding ribonucleic acid (lncRNA)) and bioactive proteins or peptides. This approach relies on the abundant expression of transmembrane proteins such as Lamp2b, CD9, CD63, PDGFR, and CD47 on EV membranes to facilitate membrane fusion and enable the targeted delivery of specific proteins and peptides through precise genetic manipulation [74]. Several studies have demonstrated the efficacy of this technique in targeted delivery applications. For example, HEK293 cells were co-transfected with plasmids encoding cardiac-targeting peptides fused to Lamp2b to successfully direct EVs to cardiac cells [75]. Similarly, DCs were genetically engineered to express Lamp2b fused with the iRGD peptide, enhancing targeting efficiency toward breast cancer cells [76]. Despite its advantages, genetic engineering faces several challenges that limit its scalability and clinical translation. Non-specific interactions between engineered ligands and unintended targets can reduce therapeutic efficacy and compromise specificity. Additionally, the complexity of genetic modification processes, particularly those involving plasmid vectors for cell transfection, poses significant obstacles to standardization and large-scale production. Nevertheless, genetic engineering remains a promising strategy for advancing EV-based therapies by enabling precise delivery of chemotherapeutics [74]. As research continues to evolve, optimizing genetic engineering techniques will be essential to enhance the functionality and clinical applicability of engineered EVs while addressing safety concerns related to immunogenicity and off-target effects.

PTX: Mechanism of Action and Challenges in Delivery

PTX is a pseudoalkaloid with a taxane ring nucleus (C₄₇H₅₁NO₁₄, MW 853.93), primarily derived from Taxus species such as Taxus brevifolia. Due to low natural yields and environmental concerns, industrial-scale production now relies on plant cell fermentation and semisynthetic approaches using precursors like 10-deacetylbaccatin III [77]. PTX exerts its anticancer effects by binding to β-tubulin, stabilizing microtubules, and disrupting mitotic spindle assembly and chromosome segregation. This leads to mitotic arrest and ultimately apoptosis. It is clinically effective against a broad spectrum of cancers, including ovarian, breast, lung, and gastric malignancies [78–81]. However, PTX faces several pharmacokinetic and safety limitations. Its poor aqueous solubility and high lipophilicity result in low bioavailability and metabolic instability. Consequently, systemic toxicities such as neutropenia, liver injury, bone marrow suppression, and hypersensitivity reactions are frequently observed [82]. PTX also exhibits non-specific biodistribution, leading to off-target accumulation in healthy organs such as the liver, kidneys, and heart. This can result in dose-limiting toxicities, including peripheral neuropathy and gastrointestinal complications [83, 84]. To improve solubility, conventional formulations like Taxol employ solubilizing agents such as Cremophor EL, which are associated with hypersensitivity reactions and additional toxicity [85]. Moreover, when administered systemically or intraperitoneally, free PTX is rapidly cleared and exhibits non-specific distribution, leading to subtherapeutic concentrations at tumor sites and increased exposure to non-target tissues. Notably, PTX demonstrates limited accumulation in pelvic lymph nodes and tumor tissues, restricting its therapeutic impact [86]. To address these limitations, various PTX formulations have been developed, including NPs, liposomes, micelles, and albumin-bound systems such as Abraxane—an albumin-bound formulation of NP-PTX [87]. These delivery platforms aim to enhance drug solubility, prolong circulation time, and improve tumor targeting.

Nanocrystal formulations of PTX have shown comparable antitumor efficacy to Taxol while reducing toxicity by minimizing off-target accumulation. Pharmacokinetic studies indicate that while Taxol exhibits higher initial plasma and tumor concentrations, nanocrystals provide more sustained retention in tumor tissues [88]. NP-based formulations also enhance drug delivery to pelvic lymph nodes after intraperitoneal administration, resulting in reduced tumor burden, increased apoptosis, and inhibited tumor proliferation [86]. Liposomal PTX formulations improve pharmacokinetics and reduce systemic toxicity, demonstrating superior efficacy in models of ovarian, metastatic breast, and non-small-cell lung cancers (NSCLC) [78, 89, 90]. Micelle-based systems, such as Genexol-PM—a Cremophor EL-free polymeric micelle formulation—extend circulation time and improve tumor targeting via the EPR effect [87]. Abraxane utilizes albumin to increase solubility and facilitate targeted delivery, offering superior efficacy and safety compared to traditional formulations [91]. Despite these advancements, challenges remain in achieving optimal PTX concentrations at tumor sites while minimizing systemic toxicity. Drug resistance—often mediated by P-glycoprotein overexpression—further complicates PTX therapy and limits long-term efficacy [92]. Therefore, future delivery strategies must focus not only on increasing tumor drug accumulation but also on minimizing exposure to non-target tissues and maintaining steady therapeutic levels. Optimizing drug delivery to reduce toxicity and overcome resistance mechanisms is crucial for improving treatment outcomes. However, the complexity and high production costs of advanced delivery platforms continue to pose barriers to widespread clinical adoption.

Isolation of EVs and PTX Loading

Isolation Techniques

EVs can be differentiated from most serum components based on their density, except for high-density lipoproteins, which share a similar density range (1.06–1.21 g/mL). As a result, size-based separation methods are often required for effective EV isolation. Isolating EVs from serum is particularly challenging due to its high viscosity and the abundance of serum proteins and non-EV lipid particles, which can interfere with particle quantification and biomarker analysis [9]. Several techniques are used for EV isolation, including ultracentrifugation, ultrafiltration, precipitation, SEC, and immunoaffinity capture (Table 2). Among these, ultracentrifugation is the most widely used method (Fig. 3) and is often supplemented with commercial kits. SEC is particularly effective at separating EVs larger than 70 nm from smaller particles and soluble proteins, resulting in lower protein contamination. However, it may yield fewer EVs due to potential loss or damage during processing. Density gradient centrifugation isolates EVs based on buoyant density but typically requires a preliminary ultracentrifugation step for optimal results [9, 93]. Comparative studies have shown that SEC-based methods typically yield higher particle-to-protein ratios than ultracentrifugation or precipitation, suggesting greater EV purity. However, yield can vary depending on the sample type and intended downstream application. The choice of isolation method depends on the intended application of EVs. For example, miRCURY precipitation achieves the highest particle count from serum, followed by Exo-spin and qEV (both SEC-based), ultracentrifugation, and finally exoRNeasy [93, 94]. The choice of EV isolation method should be guided by the intended application, whether it be for high-yield recovery, low protein contamination, or functional studies. A detailed comparison of these techniques is provided in Table 2.

Table 2.

Comparison of EV isolation techniques

Isolation technique	Principle	Types	Recovery	Purity	Samples volume	Processing Time	References
Ultracentrifugation	Sequential separations of particulate components and solutes based on their density, size, and shape	Differential centrifugation	5–25%	Low	100s of mLs	8 h	[95–97]
		Density gradient centrifugation	Higher than UC	Similar to UC	Up to 1 mL	20 h
		Rate-zonal centrifugation	Variable	Moderate	Variable	Variable
Size based techniques	Uses membrane filters, columns, or laminar flow dynamics to separate particles based on size and shape	Ultrafiltration	Moderate	Varies; depends on filter size	5–50 mL	~ 150 minutes	[98]
		Exosome isolation kit	High	High; optimized for specific applications	1–5 mL	Varies by kit
		Sequential filtration	Moderate	Moderate to high	10–20 mL	Hours
		Size exclusion chromatography (SEC)	High	Moderate to high; less protein co-isolation1	1–5 mL	Hours
		Flow field-flow fractionation (FFFF)	High	High; depends on antibody specificity	50-100 μL	Hours
		Hydrostatic filtration dialysis (HFD)	Variable	Low to moderate	Moderate; depends on membrane properties	10–20 mL
Immunoaffinity capture-based techniques	Utilizes antibodies or magnetic beads coated with antibodies to capture specific EV subpopulations	Enzyme-linked immunosorbent assay (ELISA)	Variable	High; depends on antibody specificity	50–100 μL	Hours	[99, 100]
		Magneto-immunoprecipitation	High	Very high; effective for specific targets	1–5 mL	1–3 hours
Precipitation kits	Employs polymers to precipitate EVs from samples	Polyethylene glycol (PEG) precipitation	Moderate	Moderate; can co-precipitate proteins	> from 100 μL up to several mLs	Overnight	[101, 102]
		Lectin-induced agglutination	Variable	Low	> 100 μL	Overnight
Microfluidic-based isolation techniques	Uses an acoustic field or immuno-based microfluidics to isolate EVs	ExoSearch chip	42–97%	Higher than UC	10–100 μL	40–60 min	[103, 104]
		Acoustic nanofilter	Very high (>80%), minimizes damage to EVs	High	50 μL	< 30 min

Fig. 3.

Schematic illustration of extracellular vesicle (EVs) isolation methods. a Ultracentrifugation: large and small EVs are separated by centrifugation at speeds of 10,000–20,000 g and 100,000–120,000 g, respectively. Density gradient ultracentrifugation can further purify EVs. b Filtration: membrane filters with varying pore sizes are used to isolate EVs of specific sizes. c Size exclusion chromatography (SEC): A SEC column with porous beads allows smaller particles, like proteins, to enter. EVs, being larger, migrate faster and are thus isolated. d Precipitation: precipitants in the solution cause less soluble particles, including EVs, to exceed their solubility limit and precipitate. e Immunoaffinity capture: antibodies targeting EV surface proteins are used to selectively isolate specific EV subpopulations. f Microfluidics: this method involves the use of microfluidic devices for EV isolation

Drug Loading Strategies

Drug loading strategies are broadly categorized into pre-loading and post-loading methods. Pre-loading involves incorporating drugs into parental cells via transfection or co-incubation, allowing drugs to be packaged into EVs during their biogenesis. Post-loading methods include direct incorporation of drugs into isolated EVs using techniques such as incubation, electroporation, sonication, extrusion, freeze/thaw cycles, or surfactants [105]. The hydrophobic lipid bilayer of EVs provides an optimal microenvironment for encapsulating and stabilizing hydrophobic drugs such as PTX. Unlike polymeric NPs or albumin-bound formulations, EVs incorporate PTX within their membrane structure, improving solubility and protecting the drug from premature degradation. This enhances the pharmacokinetic profile of PTX and allows for higher loading efficiencies without surfactants or organic solvents, potentially reducing systemic toxicity [106]. Comparative studies have shown that PTX-EVs exhibit increased tumor accumulation and reduced off-target distribution relative to synthetic nanocarriers, highlighting their potential as effective drug delivery vehicles [38, 107]. PTX can passively diffuse into EV membranes during co-incubation, though this typically results in low loading efficiencies [13, 106]. Reported encapsulation efficiencies of PTX in EVs vary considerably depending on the cell source and loading methodology, with values ranging from approximately 4.2% in EVs derived from human pancreatic ductal carcinoma cells via passive incubation, to 14.23% in human umbilical cord MSC-EVs using extrusion, and up to 19.55% in M1 macrophage-derived EVs following sonication [108–110]. Haney et al. confirmed the efficiency of sonication in intracellular accumulation and delivery of PTX via M-EVs to triple-negative breast cancer cell (TNBC) [111]. However, sonication may compromise EV structural integrity due to mechanical stress. Performing sonication in an ice bath can help mitigate this risk. Kim et al. demonstrated that electroporation achieved superior PTX loading compared to both incubation and sonication [38], while Fuhrmann et al. reported that using saponin or hypotonic dialysis increased efficiency up to 11-fold over traditional methods like incubation or electroporation, highlighting the significant impact of the loading method on drug encapsulation [112]. Advances in microfluidic technology have led to the development of integrated systems capable of both isolating and loading EVs. A lab-on-a-chip platform using surface acoustic waves and miniaturized electroporation demonstrated comparable purity to ultracentrifugation but with improved drug loading and cellular uptake of PTX-EVs, highlighting the potential of this technology to streamline EV-based drug delivery systems for cancer treatment [113]. These findings underscore the importance of selecting drug-loading techniques based on the physicochemical properties of the drug and the lipid composition of EVs. While significant progress has been made, further optimization is essential for clinical translation. Future efforts should focus on refining current techniques and developing innovative approaches to enhance EV purity, loading efficiency, and therapeutic functionality.

Preclinical Studies of PTX-EVs

In Vitro Examinations

Unmodified PTX-EVs

Unmodified or minimally modified EVs represent the foundational approach to PTX delivery, leveraging the intrinsic targeting capabilities and biocompatibility of EVs derived from tumor cells or MSCs. These EVs typically undergo drug loading without extensive surface engineering, maintaining their natural biological functions and tumor-homing properties. Kanchanapally et al. delivered PTX using exosomes (110 to 125 nm) derived from MDA-MB-231 and MCF-7 breast cancer cells and compared their efficacy against these cell lines with free PTX and liposomal PTX. The results showed that PTX delivered with exosomes was more efficient in inhibiting cancer cell growth compared to both free and liposomal PTX. In MBA-MB-231 cells, exosomal PTX exhibited a significantly lower half-maximal inhibitory concentration (IC₅₀) of 2.87 nM compared to liposomal PTX (IC₅₀ 8.11 nM), and free PTX (IC₅₀ 12.14 nM). Similarly, in MCF-7 cells, exosomal PTX demonstrated a lower IC₅₀ value of 2.36 nM while liposomal PTX and free PTX had IC₅₀ values of 2.91 nM and 3.83 nM, respectively (p < 0.05). This enhanced efficacy in inhibiting cell growth was attributed to the higher intracellular accumulation of PTX-EVs. The study also showed that exosomal PTX induced apoptosis more effectively by modulating p21, Bax, Bcl-2, and Bcl-xL proteins (p < 0.05). However, the study does not fully elucidate the mechanisms by which exosomes facilitate drug delivery across cell membranes, necessitating further investigation [114]. Similarly, Bi et al. enhanced the efficacy of hepatocarcinoma treatment by using PTX-loaded TC-EVs, which leverage both the EPR effect and the inherent tumor-homing properties of TC-EVs for targeted drug delivery. Accordingly, the TC-EV-PTX formulation (EVPA) exhibited superior antitumor efficacy compared to free PTX. Notably, EVPA administration led to enhanced immune responses, characterized by increased infiltration of CD4 + and CD8 + T cells into the tumor site. While TC-EVs leverage inherent tumor-homing properties, the risk of pro-tumorigenic effects and the potential for transferring drug resistance mechanisms necessitate modification strategies [115]. Expanding the scope to explore the potential of EVs in modulating cellular processes, a study explored the potential of PTX-EVs as a therapeutic strategy for cancer treatment, focusing on the modulation of autophagy pathways. EVs were isolated from A549 lung cancer cells and loaded with PTX. The PTX-EVs effectively induced apoptosis in A549 and MCF-7 breast cancer cells while inhibiting autophagy, suggesting a promising approach for effective chemotherapy through targeted delivery [116]. Loading PTX into autologous EVs derived from Lymph Node Carcinoma of the Prostate (LNCaP) and Prostate Cancer-3 (PC-3) cells significantly enhanced cytotoxicity against cancer cells, increasing the effect from 16% to 40% after 24 hours across various EV populations. Both exosomes and microvesicles delivered PTX to recipient cells through endocytosis, further aiding intercellular drug release. While unloaded EVs increased cancer cell viability, PTX-EVs significantly enhanced cytotoxicity. The increased cancer cell viability with unloaded EVs indicates that EVs are not simply inert carriers and must be carefully evaluated for biological activity [106].

MSC-EVs present a more refined approach with the potential for targeted delivery and reduced immunogenicity. Various studies have explored the synergistic potential of combining MSC-EVs with PTX to enhance efficacy and reduce toxicity in cancer treatments. For instance, a recent study demonstrated the synergistic effect of PTX with the secretome from human uterine cervical MSCs (CM-hUCESC) in targeting TNBC. This combination showed promising results by decreasing tumor cell proliferation and invasiveness in vitro. Specifically, the combination of PTX (1 µM, 3 µM, and 5 µM) with CM-hUCESC (1 µM, 3 µM, and 5 µM) significantly inhibited the proliferation of MDA-MB-231 cells and primary breast cancer cells, allowing for superior effects at lower PTX concentrations compared to higher doses of free PTX (p < 0.0001). Additionally, the treatment altered cell cycle progression and induced apoptosis, with tissue inhibitor of metalloproteinases-1 (TIMP-1) and TIMP-2 identified as key mediators of the anti-tumor effects (p = 0.004) [117]. While EVs are being explored as drug delivery vehicles, their role in promoting chemoresistance has also been reported. In TNBC, PTX treatment elicited C-X-C motif chemokine ligand 1 (CXCL1) growth factor-enriched EVs from apoptotic cancer cells, promoting chemoresistance and invasion of co-cultured TNBC cells through M2 macrophage polarization. These findings highlight the complex role of EVs in cancer treatment, where they can serve as both drug delivery vehicles and mediators of resistance mechanisms. Future strategies may need to combine drug delivery with approaches to mitigate EV-induced drug resistance [118]. Additional studies on the application of PTX-EVs in in vitro cancer models are summarized in Table 3. Despite their promising intrinsic tumor-targeting capabilities and biocompatibility, unmodified PTX-EVs face several critical limitations. These include limited drug loading efficiency, rapid clearance by the reticuloendothelial system, insufficient targeting specificity, off-target effects, and challenges with in vivo stability.

Table 3.

Application of exosomes for PTX delivery in in vitro cancer models

EV source	Isolation method	Tumor model/cell line(s)	IC50 (PTX-EVs)	IC50 (Free PTX)	Apoptosis/growth inhibition	Mechanism/key findings	References
Macrophage-derived exosomes	Total Exosome Isolation Kit	3LL-M27 (mouse lung carcinoma) MDCK wt (wild-type kidney cells) MDCK MDR1 (multi drug-resistant)	3LL-M27: 13.57 ± 1.33 MDCK wt: 23.33 ± 3.77 MDCK MDR1: 187.5 ± 38.65	3LL-M27: 126.41 ± 31.31 MDCK wt: 428.77 ± 63.37 MDCK MDR1 > 10,000	> 50-fold increase in cytotoxicity compared to free PTX; 40% viability reduction (MDR1-MDCK); ↑apoptosis	Significant increase in cytotoxicity and apoptosis in MDCKMDR1 (Pgp + ) MDR cells	[38]
LNCaP & PC-3 (prostate)	Differential centrifugation	LNCaP, PC-3	50 nM	10 nM	40% decrease in PC-3 cell viability after 24 h; 80% decrease in LNCaP cell viability after 24	Reduced microvesicle-mediated cytotoxicity by ~ 30%;	[106]
MSCs (human)	Ice-cold extraction solvent	SK-OV-3, A549, MDA-hyb1	Not specified	Not specified	80–90% cytotoxicity; > 60% tumor reduction, ~ 50% metastasis reduction	Synergistic effect at lower PTX concentrations; PTX-EV contained a 7.6-fold lower PTX concentration than the amount of free PTX needed for equivalent cytotoxicity	[119]
U-87 MG (glioblastoma)	Exo-spin™ kit	U-87 MG	59.92%	80.70%	60% cell death at 10 μg/mL (sonication method)	Sonication > incubation for drug loading/efficacy	[120]
Pancreatic ductal cells	Total exosome isolation kit	Pancreatic ductal adenocarcinoma/PANC-1, MIA PaCa-2 and BxPC-3 cells	10 to 100 nM	Not specified; higher than PTX-EV IC50	Enhanced apoptosis	Clathrin-mediated endocytosis	[121]
MSCs (human)	Ultracentrifugation	CFPAC-1 (pancreatic cancer)	2.73 ng/mL	2.54 ± 1.75 ng/mL	60% proliferation inhibition at 48h; ↑apoptosis	Higher efficacy than free PTX	[122]

Surface Functionalized PTX-EVs

To overcome limitations of natural EVs, surface functionalization with targeting ligands has been extensively explored to enhance drug delivery and therapeutic efficacy. For example, a multifunctional nanosystem—referred to as EPM—was developed by co-extruding TC-EVs, PTX-albumin (PA), and melanin, enabling combined chemo-photothermal-immunotherapy for breast cancer. The EPM was efficiently taken up by 4T1 breast cancer cells, with uptake rates reaching 84.1% after 1 hour and nearly 100% after 3 hours. The internalization process was found to be energy-dependent, primarily mediated by lectin pathways. EPM exhibited enhanced cytotoxicity compared to free PA, inducing 71.6% apoptosis in 4T1 cells. When exposed to near-infrared (NIR) laser irradiation, the apoptotic rate increased to 87.7%, highlighting the synergistic effect of photothermal therapy with EV-mediated combinational chemo-immunotherapy [123]. Kim et al. engineered M-EVs isolated from RAW 264.7 cells with aminoethylanisamide-polyethylene glycol (AA-PEG) for targeted PTX delivery. Engineered EVs specifically bind to sigma receptors overexpressed on 3LL-M27 lung cancer cells and enable precise tumor targeting. By leveraging the innate biocompatibility of exosomes and integrating active targeting strategies, this approach overcame the rapid clearance and inefficient tumor penetration of EVs, providing a promising platform for precision oncology. However, the impact of sigma receptor expression levels in different cancer types on the binding affinity and specificity of AA-PEG-modified EVs requires further exploration [124]. Folic acid (FA) modification has also been explored in engineering EVs for PTX delivery. The use of FA-functionalized bovine milk exosomes for co-delivery of PTX and 5-fluorouracil (5-FU) in breast cancer therapy showed promising in vitro outcomes. The practical drug loading of Exo-PTX was notably higher (28 ± 3.9%) than that of Exo-5-FU (19 ± 4.6%), while folate-targeted formulations, FA-Exo-PTX and FA-Exo-5-FU, demonstrated slightly lower loading efficiencies at 26 ± 2.6% and 17 ± 2.9%, respectively. Despite these differences in loading capacities, all formulations achieved comparable entrapment efficiencies within the range of 75–85%. The enhanced cellular uptake and sustained drug release profiles (25% for up to 48 h) of FA-modified exosomes in MCF-7 and MDA-MB-231 cell lines were attributed to folate receptor-mediated targeting. The Exo-5-FU and Exo-PTX + Exo-5-FU formulations demonstrated a significant reduction in IC50 values, decreasing from 2.2 ± 0.4 μM and 0.28 ± 0.12 μM, respectively, when compared to free PTX (0.6 ± 0.5 μM) and unloaded EVs (13 ± 1.6 μM). These treatments also suppressed cell migration and increased apoptotic indices, achieving values of 0.92 in MCF-7 and 0.91 in MDA-MB-231 cells. Notably, FA functionalization amplified cytotoxicity, enhanced tumor-specific accumulation, and improved intracellular delivery. These effects are likely attributed to enhanced intracellular PTX delivery facilitated by the strong interaction between FA and folate receptors, which promotes improved cellular uptake and therapeutic efficacy [125]. However, experimental models are required to characterize the synergistic efficacy, systemic biodistribution, toxicity profiles, and TME interactions of FA-functionalized EVs. Similarly, in the study conducted by Kandimalla et al., an innovative FA-Exo-PTX formulation was developed utilizing exosomes derived from bovine colostrum to enhance the therapeutic efficacy of PTX in NSCLC treatment [3]. Another approach to enhance targeting efficiency involves engineering EVs with specific peptides. Exosomes derived from embryonic stem cells were utilized to deliver PTX across the BBB for glioblastoma treatment. Exosomes were engineered with the c(RGDyK) peptide and loaded with PTX (cRGD‐Exo‐PTX). Comparative analysis demonstrated that the cRGD‐Exo‐PTX exhibited superior efficacy in delivering PTX to glioblastoma cells compared to unmodified exosomes or free PTX. The designed system represented considerable antiproliferation effects on U87 and U251 human glioblastoma cell lines [126]. Surface functionalization markedly improves the targeting specificity and therapeutic efficacy of PTX-EVs, enabling enhanced cellular uptake and overcoming biological barriers. Despite promising in vitro outcomes, further in vivo studies are necessary to fully understand biodistribution, toxicity, and TME interactions, alongside the need to optimize functionalization techniques for scalable manufacturing.

Hybrid PTX-EVs Systems

Hybrid EV systems represent a significant advancement in drug delivery by combining the natural biocompatibility and targeting abilities of EVs with the customizable features of synthetic NPs like liposomes or lipid NPs [127]. These hybrids are created using various techniques-such as incubation, electrostatic interactions, PEG-mediated fusion, co-extrusion, and freeze-thaw cycles-to merge EVs with synthetic carriers, resulting in delivery vehicles that offer improved cargo loading, stability, and surface modification capabilities while maintaining low immunogenicity [128]. Kim et al. developed hybrid exosomes (HE) by fusing exosomes with liposomes for PTX delivery (PTX-HE) to TNBC cells. The PTX-HE had improved drug loading capacity (6.20 ± 0.79%) and encapsulation efficiency (86.79 ± 11.07%) compared to unmodified liposomes (48.60 ± 3.68%). In vitro studies demonstrated higher cellular uptake efficiency and stronger anti-migration effects of PTX-HE. This biohybrid nanosystem exhibited a significantly lower IC50 value (4.32 ± 0.48 μg/mL) and enhanced anti-migration effects while reducing cell mobility to 16% compared to PTX-liposomes (6.32 ± 3.08 µg/mL) [129]. Wang et al. developed such a hybrid system by combining MSC-derived exosomes with liposomes, which significantly enhanced the PTX loading capacity-achieving a 120% increase compared to natural exosomes alone (2.2% vs. 1.0% loading content). This improvement was primarily attributed to the hybrid structure, where the EV component facilitated hydrophobic drug encapsulation, while the liposomal fusion contributed to a sustained drug release profile over 48 hours, overcoming the rapid burst release seen with natural exosomes. The hybrid EVs also demonstrated markedly improved cellular uptake and targeting specificity towards tumor cells, including CT26 colon carcinoma, mouse melanoma B16 cell line, and human ovarian cancer A2780 cells, which overexpress folate receptors. Furthermore, the engineered HE exhibited excellent stability, maintaining consistent particle size in serum-containing media for up to four weeks, whereas unmodified exosomes showed significant aggregation under similar conditions. These findings underscore the potential of hybrid EV systems as stable, efficient, and targeted nanocarriers for cancer drug delivery [130]. Hybrid EV systems combining natural vesicles with synthetic NPs offer superior drug loading, stability, and multifunctionality, representing a promising direction for effective PTX delivery. Nevertheless, their increased complexity introduces challenges related to reproducibility, immunogenicity, off-target effects, and large-scale production, necessitating standardized protocols and comprehensive in vivo validation.

In Vivo Tumor Models

Minimally Modified PTX-EVs

The transition from in vitro to in vivo models is crucial for validating the therapeutic potential of EV-based PTX delivery systems. For example, studies using zebrafish models demonstrated that EV-PTX and doxorubicin effectively cross the BBB and target brain tumors, highlighting their potential for treating central nervous system malignancies [131]. Studies in mice have shown that EV-encapsulated drugs exhibit superior anticancer effects compared to free drugs. This enhanced efficacy is attributed to the ability of EVs to improve drug bioavailability, reduce systemic toxicity, and enhance targeted delivery to tumor sites [110, 132]. Similarly, milk-derived exosomes loaded with PTX exhibited markedly enhanced efficacy against lung tumor xenografts in mice. Orally administered PTX-loaded milk exosomes significantly inhibited tumor growth compared to intraperitoneal injections of free PTX and showed considerably lower systemic and immunological toxicities [133]. Furthermore, PTX-loaded milk exosomes displayed considerably lower systemic and immunological toxicities than free PTX. However, the mechanisms underlying the improved oral bioavailability of PTX-loaded milk exosomes remain unclear and warrant further investigation [134]. Aquil et al. investigated the antiproliferative effects of berry anthocyanidins (Anthos) and exosomal formulations in an animal model of ovarian cancer and demonstrated that the combination of ExoAnthos and EV-PTX significantly enhanced antitumor activity compared to individual treatment in cervical tumor xenografts (61%; p < 0.01). Oral administration of EV-PTX achieved higher therapeutic efficacy compared to free PTX, highlighting the potential of exosomal delivery in enhancing the bioavailability and tolerability of PTX [135]. The in vivo efficacy of combining MSC secretomes with PTX was confirmed in the NMRI fox1 nu/nu mouse tumor xenograft model. Mice treated with the combination of PTX and CM-hUCESC exhibited significant reductions in tumor growth. Notably, the combination therapy showed enhanced efficacy in reducing tumor growth, particularly evident at day 14 post-treatment (p = 0.025) [117]. Despite the benefits of EVs in targeting tumor sites and penetrating deep into tumor tissues, concerns have been raised regarding their potential to activate tumors when derived from cancer cells [136]. While natural and minimally modified EVs have demonstrated significant in vivo efficacy in various tumor models, challenges such as understanding long-term safety, tumor-promoting risks, and mechanisms of enhanced oral delivery remain to be addressed.

Surface-Engineered PTX-EVs

Engineering EVs can result in a notable suppression of tumor development without noticeable side effects in experimental models [69]. For instance, modification of human umbilical vein endothelial cells (HUVECs)-derived EVs with biotin-avidin resulted in improved cancer targeting activity in a mouse model of HepG2 human hepatocellular carcinoma, increasing the bloodstream maintenance of PTX up to 48 hours. However, the long-term effects of biotin-avidin modification on EV immunogenicity and potential off-target effects need to be further investigated [137]. Regarding the increased cytotoxicity against drug-resistant cancer cells, Kim et al. further optimized the encapsulation of PTX into macrophage exosomes with an AA-PEG vector moiety to specifically target the sigma receptor in vivo. This approach was tested in C57BL/6 mice with pulmonary metastases established by injecting 3LL-M27 cells. Exosomes were isolated from autologous macrophages, labeled with DiL (red), and functionalized with the PEG-AA vector composed of the lipid molecule 1,2-distearoyl-sn-glycero-3-phosphoethanolamine–polyethylene glycol–ascorbic acid. Mice with established lung metastases were intravenously injected with these vectorized exosomes (DiL-AA-PEG-exo) or non-vectorized exosomes (DiL-exo) as the control group. The antineoplastic effects of PTX-EVs were evaluated using 3LL-M27 cells transduced with lentiviral vectors encoding the optical reporter mCherry. The results showed superior antineoplastic efficacy of AA-PEG-exoPTX compared to non-vectorized exoPTX or Taxol, leading to the potent eradication of pulmonary metastases. Additionally, confocal imaging revealed close co-localization of fluorescently labeled AA-PEG-exosomes with cancer cells in the mouse liver, suggesting potential applications for pulmonary metastases therapy [124]. Kandimalla et al. reported that FA-Exo-PTX demonstrated significant efficacy in vivo, achieving over 50% tumor reduction and 55% growth inhibition after subcutaneous injection in a xenograft mouse model of lung cancer. These results surpassed the efficacy of free PTX injected intravenously and matched that of Abraxane. Importantly, the oral administration of FA-Exo-PTX demonstrated a significantly improved safety profile. This approach effectively avoided the systemic toxicities commonly associated with solvent-based PTX administration. However, it should be noted that the enhanced efficacy and safety profile of FA-Exo-PTX may be specific to certain lung cancer subtypes that express high levels of folate receptors [3]. To address the challenge of targeting collagen-rich pancreatic TME, researchers designed a formulation (rmExo-PTX) by conjugating RGD ligands and magnetic NPs (MNPs) with autologous EVs from human pancreatic cancer cells (PANC-1). Engineered EVs demonstrated enhanced therapeutic efficacy in PANC-1 xenograft mice, resulting in significant tumor reduction compared to free PTX. The RGD modification facilitated targeted delivery by binding to αvβ3 integrins, which are overexpressed on pancreatic cancer cells. Notably, autologous EVs derived from PANC-1 cells exhibited superior homing properties compared to heterologous EVs, such as those derived from U937 or HT29 cells, due to shared integrin β3 expression between the PANC-1 cells and their EVs. Furthermore, rmExo-PTX mediated ECM remodeling by reducing α-smooth muscle actin and collagen type I, thereby enhancing tumor penetration. Although MNPs were incorporated into the EVs, external magnetic guidance did not provide additional therapeutic benefits, proving the intrinsic targeting capabilities of the EVs. These findings underscore the potential of autologous EVs as a potential tool to cross stromal barriers in pancreatic cancer, leveraging both active targeting and intrinsic homing mechanisms to optimize drug delivery [110]. These surface-functionalization strategies demonstrate significant improvements in targeting specificity, therapeutic efficacy, and safety profiles of PTX-EVs in vivo. However, long-term safety and immunogenicity require further evaluation.

Hybrid and Stimuli-Responsive PTX-EV Systems

Hybrid and stimuli-responsive PTX-EV nanosystems combine EVs with other nanomaterials or external triggers to enhance drug delivery and therapeutic outcomes. Studies on experimental models have demonstrated that encapsulation of EVs within NPs achieved high drug loading capacity, resulting in enhanced cytotoxicity and improved drug accumulation at the tumor site. For example, erythrocyte membrane-coated PTX spherical crystals demonstrated superior efficacy in inhibiting tumor growth compared to free PTX in tumor-bearing murine models [138]. Wang et al. evaluated the therapeutic efficacy of their HE-liposome system loaded with PTX (ELP) using a CT26 colon carcinoma tumor-bearing mouse model. Mice treated intratumorally with ELP at 6 mg/kg PTX showed significantly greater tumor growth suppression compared to those receiving free PTX or the HE-liposome vector without drug. Survival analysis demonstrated that ELP treatment notably prolonged the lifespan of tumor-bearing mice. Importantly, ELP administration did not cause significant changes in body weight or induce pathological damage in major organs such as the heart, lungs, spleen, and kidneys, indicating a favorable safety profile. Additionally, analysis of the TME revealed that ELP treatment activated CD4 + and CD8 + T cells, decreased immunosuppressive regulatory T cells, reduced M2-type tumor-associated macrophages (TAMs), and promoted polarization of TAMs toward the pro-inflammatory M1 phenotype. These findings highlight that, beyond enhanced drug delivery, the hybrid EV system modulates the tumor immune microenvironment to potentiate antitumor immunity and improve therapeutic outcomes in vivo [130]. While this approach shows promise, the complexity of manufacturing EV-NP hybrids may limit their scalability and cost-effectiveness. In vivo studies demonstrated that PTX-HE exhibited 1.9-fold higher tumor retention in delivering PTX to a TNBC mouse model compared to liposomes. Enhanced therapeutic efficacy was confirmed by a tumor inhibition rate of 60% and a significant reduction in average tumor weight in the PTX-HE group (0.37 ± 0.21 g) compared to the control group (0.95 ± 0.24 g). Safety evaluations revealed minimal systemic toxicity, as indicated by lower levels of liver and kidney function markers relative to free PTX, with no significant organ damage observed. Nonetheless, long-term studies are necessary to evaluate potential risks associated with PTX-EVs, including drug resistance and the promotion of metastasis [129]. While this approach shows promise, the complexity of manufacturing EV-NP hybrids may limit their scalability and cost-effectiveness. In vivo studies using an orthotopic 4T1 breast cancer mouse model further validated the efficacy of the EPM nanosystem for PTX delivery. Following peritumoral injection, photoacoustic imaging revealed the highest EPM accumulation at the tumor site within 4 to 8 hours. Upon NIR laser activation, the EPM + laser group showed significant tumor growth inhibition without recurrence throughout the observation period, significantly higher than controls (PBS, PBS + laser, PA, and EM groups). Notably, EPM treatment did not cause significant changes in body weight, indicating minimal systemic toxicity. Fluorescence imaging after intravenous injection demonstrated rapid tumor targeting and clearance, with minimal retention in other tissues. However, the requirement for NIR laser activation limits the use of this formulation in tumors that are not easily accessible to light [123]. Table 4 summarizes the application of exosomes for PTX delivery in various in vivo cancer models. Collectively, these in vivo experiments underscore the remarkable potential of EV-based drug delivery systems to enhance therapeutic efficacy, reduce systemic toxicity, and overcome biological barriers while highlighting the need for further research to address challenges such as scalability and clinical applicability.

Table 4.

Application of exosomes for PTX delivery in vivo cancer models

EV source	Modification	Tumor Model	Tumor Model	Quantitative Outcome	Mechanism/ Comparative Advantage	Ref
Macrophage	Unmodified	C57BL/6 mice with metastatic Lewis lung carcinoma	Free PTX and PTX-EV: 107 particles/10 µL x 2; injected intra tail vein	Tumor volume reduction: 60% with PTX-EV, 25% with PTX	Increased cellular uptake, reverse multi-drug resistance	[38]
Macrophage	Nanoformulation with aminoethyl anhydride-polyethylene glycol carrier	Mice model of lung cancer	PTX-EV: 4 × 1011 particles/100 μl; injected intra tail vein	60-70% fewer metastatic nodules with PTX-EV	Inhibition of pulmonary metastasis and enhanced lung-targeting capabilities	[124]
M1 and M2-macrophages	Unmodified	Xenograft tumor-bearing Balb/c mice	specifiedFree PTX and PTX-EV: 5 mg/kg; intravenous injection	2-fold reduction in tumor size with PTX-EV	Activated proinflammatory M1 signaling/ Immunomodulation improved anti-tumor properties in PTX-EV	[109]
Brain neuronal glioblastoma-astrocytoma U-87 MG, endothelial bEND.3, neuroectodermal tumor PFSK-1, and glioblastoma A-172 cell line	Unmodified	Zebrafish (Danio rerio)	0.2 mg/ml PTX-EV; delivered across BBB	Tumor inhibition by 40-60% with PTX-EV	Crossing BBB with intact PTX payload/ Localized delivery	[139]
Pancreatic ductal cells	Exo/PTX saucer-like double-layer nanoformulation	Lung adenocarcinoma in nude mice	Free PTX and PTX-EV: 10 mg/kg	Tumor inhibition: 50% with PTX-EV, 30% with PTX	Enhanced retention in tumor site/ Prolonged circulation half life	[140]

Optimizing PTX-EVs for Clinical Use: Emerging Strategies and Practical Barriers to Clinical Translation

The clinical translation of engineered EVs for PTX delivery faces several practical challenges. Maintaining the integrity and functionality of EVs during large-scale, good manufacturing practice (GMP)-compliant EV production is limited due to the need for precise control over reaction conditions and purification steps, batch-to-batch variability, low yields, and difficulties preserving EV integrity during isolation, modification, and PTX loading [141, 142]. Although techniques such as tangential flow filtration and SEC are being optimized for scalability and reproducibility, regulatory frameworks for EV-based therapeutics remain under development [143]. Comprehensive preclinical and clinical toxicology evaluations-including immunogenicity, biodistribution, and off-target effects-are essential to ensure safety and efficacy prior to clinical adoption [144]. The performance of stimuli-responsive EVs depends heavily on TME conditions. Variability in TME pH (6.2 to 7.0) can lead to inconsistent drug release, as some pH-sensitive linkers may not activate in less acidic regions [145]. Temperature-responsive systems also face limitations due to modest temperature differences between tumor and normal tissues [146]. Tumor heterogeneity further complicates the predictability and reproducibility of EV-based systems for PTX delivery, highlighting the need for robust, clinically feasible stimuli-responsive systems validated in vivo [106, 147]. Fabrication methods such as electroporation, incubation, sonication, and freeze-thaw cycles have been employed, with electroporation-enhanced hybridization yielding the highest PTX loading efficiency [148]. The development of exosome–liposome hybrid NPs, which utilize both the natural targeting properties of EVs and the customizable features of liposomes, exemplifies a promising direction for enhancing drug accumulation at tumor sites [149, 150]. Comparative analysis of recent research on PTX delivery highlights several strategies from liposomal formulations that could be leveraged to advance EV-based systems. PTX-EV–liposome hybrids demonstrate improved solubility, enhanced tumor accumulation, and reduced systemic toxicity compared to EVs or liposomes alone [151]. Liposomal PTX carriers can be extensively modified with functional moieties such as PEGylated folate, glutamic hexapeptide-FA derivatives, estrone, and pH-responsive peptides. These modifications have led to enhanced tumor targeting, prolonged systemic circulation, and improved therapeutic outcomes, as demonstrated in preclinical models of breast, ovarian, and brain cancer [146, 152–154]. Notably, the co-delivery of chemotherapeutics and siRNA in dual-modified cationic liposomes, as well as the use of hybrid nanosystems combining liposomes and EVs, has shown synergistic effects against tumor growth and resistance [154]. Another promising approach is the encapsulation of PTX prodrugs within EVs to enable stimuli-activated drug release. These prodrugs remain stable during circulation and activate specifically within the TME, where elevated glutathione levels or acidic pH trigger drug release. This strategy enhances PTX stability, minimizes premature leakage, and allows controlled, site-specific activation. Optimized loading conditions—including pH, temperature, and ultrasound—have produced EV formulations with high drug-loading capacity, efficient tumor cell accumulation, and potent antiproliferative effects [155]. Additionally, biomimetic decoy EVs co-loaded with PTX and immune checkpoint inhibitors offer a novel means to overcome the immunosuppressive TME. These EVs sequester immunosuppressive signals such as programmed death-ligand 1 (PD-L1) while delivering PTX directly to tumor cells. Preclinical studies demonstrate that co-delivery of PTX and immune checkpoint inhibitors via EVs enhances antitumor immunity, reduces tumor growth, and overcomes resistance to monotherapy. This synergistic approach is especially promising for combination immuno-chemotherapy regimens targeting cancers lacking tumor-specific markers [156]. These findings suggest that surface engineering, combination therapy, and hybrid nanosystems can be adapted to improve the specificity, stability, and translational potential of EV-based PTX delivery platforms. Standardized protocols and rigorous analytical methods remain critical to improve reproducibility and unlock the full therapeutic potential of PTX-EVs.

Conclusion and Future Prospects

EVs have emerged as promising vectors for cancer therapy due to their inherent targeting capabilities and biocompatibility. Nonetheless, several obstacles must be overcome to fully realize their clinical potential. The choice of EV source critically influences therapeutic efficacy, as EVs retain characteristics from their parent cells that affect cargo capacity, biodistribution, and anticancer activity. Despite advances, a deeper understanding of EV composition and biomolecular mechanisms is needed. Scalable, GMP-compliant production remains a significant hurdle, complicated by complex and costly protocols. Enhancing targeting specificity through engineering will reduce off-target toxicity and improve therapeutic outcomes. Combining EV-PTX with immunotherapy, photothermal therapy, or other modalities may yield synergistic benefits. Furthermore, the development of EV analogs and hybrid systems offers promising avenues to overcome current limitations in stability, targeting, and drug-loading efficiency. Realizing the clinical promise of PTX-EVs will require multidisciplinary collaboration, standardized methodologies, and continued innovation..

Author Contributions

Conceptualization, literature search, and drafting: Mohamed J. Saadh; Critical revision of the manuscript: Hanan Hassan Ahmed, Radhwan Abdul Kareem; writing—review and editing: Mohamed J. Saadh, Ashishkumar Kyada, H. Malathi, Deepak Nathiya, Deepak Bhanot, Waam Mohammed Taher, Mariem Alwan, Mahmood Jasem Jawad, Atheer Khdyair Hamad; Funding acquisition, supervision: Hanan Hassan Ahmed.

Funding

This study was funded by the Researchers Supporting Project (ANUI2024M111), Alnoor University, Mosul, Iraq.

Data Availability

No new data were generated or analyzed in this study.

Declarations

Conflict of interest

Authors have no conflicts of interest or fundings.