Advances in Drug Delivery Systems for Breast Cancer: From Microenvironment Barriers and Smart Carriers to Clinical Translation Strategies

Abstract

Breast cancer heterogeneity and its complex tumor microenvironment (TME) are key drivers of therapeutic failure and multidrug resistance (MDR). Drug delivery systems (DDS) provide an important approach, but their clinical translation remains limited by biological variability, formulation complexity, and manufacturing challenges. This review delineates the evolution of breast cancer DDS from passive carriers to increasingly functional and responsive platforms, focusing on three core dimensions. First, regarding decisive barriers, we dissect the constraints imposed by the three-dimensional collagen network, pH/glutathione (GSH) gradients, and tumor-associated macrophages (TAMs) on drug penetration. These factors not only act as physical and biochemical obstacles but can also be harnessed as endogenous triggers for responsive design. Second, in terms of diversified platforms, we critically evaluate liposomes, polymeric nanoparticles, hydrogels, extracellular vesicles, and antibody-drug conjugates (ADCs). We highlight the potential of solid lipid nanoparticles to address drug resistance, the development of carrier-free delivery enabled by prodrug self-assembly, and the clinical significance of ADCs (eg, T-DXd) in expanding treatment options for HER2-low breast cancer. Third, we examine inter- and intratumoral heterogeneity, metastatic site-specific delivery barriers in the brain, bone, and lung, the multilayered resistance network from efflux pumps to cancer stem cells, and the sequential barriers nanomedicines must traverse from systemic circulation to intracellular targets. We propose three strategic directions: personalized precision delivery integrating liquid biopsy, organoids, and functional imaging to support patient stratification and treatment adaptation; AI-assisted formulation optimization using machine learning to develop material-structure-function predictive models and support data-driven formulation design; and multi-responsive systems leveraging endogenous signals such as pH, GSH, and enzymes, together with exogenous stimuli including light, magnetism, and ultrasound, for spatiotemporally controlled combination therapy. This review summarizes key considerations from biological barriers to clinical translation, emphasizing that alignment of tumor subtype, metastatic site, microenvironmental characteristics, and patient-specific factors may improve the clinical applicability of drug delivery strategies.

Introduction

The global incidence of breast cancer is steadily rising, surpassing lung cancer to become the most common malignancy among women worldwide. Despite continuous advances in comprehensive treatment modalities, including surgery, radiotherapy, chemotherapy, endocrine therapy, and targeted therapy, breast cancer remains a leading cause of cancer-related mortality in women.1 The root causes of this therapeutic dilemma are twofold: the non-specific distribution of conventional chemotherapeutic agents leads to severe systemic toxicity, while tumor heterogeneity and the overexpression of drug efflux pumps (eg, P-glycoprotein) give rise to multidrug resistance, a critical driver of treatment failure and disease recurrence.2 The complexity of the tumor microenvironment (TME) further exacerbates these challenges. Within the TME, the extracellular matrix, particularly the three-dimensional collagen network, not only constitutes a physical barrier that impedes effective drug penetration but also promotes tumor invasion and immune evasion through mechanotransduction signaling.3 Therefore, a profound understanding of the interplay between TME components and drug delivery systems has become a central scientific issue for overcoming current therapeutic bottlenecks.

Against this background, drug delivery systems have been increasingly investigated as approaches to improve the pharmacokinetics, tumor distribution, and therapeutic index of anticancer agents. By encapsulating therapeutic agents within nanoscale carriers, delivery systems can improve drug pharmacokinetics, prolong circulation time, and facilitate drug accumulation at tumor sites. Liposomal doxorubicin (Doxil®), approved in 1995, is widely regarded as the first FDA-approved nanomedicine.4 Since then, drug delivery research has expanded from passive carrier systems toward actively targeted and stimulus-responsive designs.5 In recent years, diverse delivery platforms have emerged. Solid lipid nanoparticles, owing to their biocompatibility and physicochemical stability, have been investigated for overcoming drug resistance.6 Hydrogels, with their three-dimensional network structures and environmentally responsive properties, have been explored for local drug delivery and postoperative immunotherapy.7 Meanwhile, extracellular vesicle-based delivery systems, leveraging their endogenous biological properties, have been investigated for the treatment of triple-negative breast cancer.8 Despite substantial progress in basic research, only a limited number of nanomedicines have successfully undergone clinical translation, highlighting the persistent gap between preclinical development and clinical application. Bridging this gap therefore remains a central challenge in the field of breast cancer nanomedicine research.

Notably, breast cancer is not a single disease entity but rather a heterogeneous spectrum composed of multiple molecular subtypes. Among these, triple-negative breast cancer, defined by the absence of estrogen receptor, progesterone receptor, and HER2 expression, represents the most aggressive subtype with the most limited treatment options.9 The unique characteristics of this subtype impose higher demands on drug delivery systems, which must not only traverse biological barriers to reach the tumor site but also overcome physical (eg, dense collagen networks) and biochemical (eg, low pH, high glutathione concentration) obstacles within the TME.10 Furthermore, metastatic breast cancer presents organ-specific delivery challenges that are not fully captured by studies of primary tumors. Brain metastases are constrained by the blood–brain barrier and heterogeneous blood-tumor barrier, bone metastases arise within a specialized mineralized microenvironment, and lung metastases are influenced by distinct pulmonary vascular and inflammatory conditions. These differences highlight the need to align delivery strategies with breast cancer subtype, metastatic site, and relevant microenvironmental barriers rather than adopting a uniform approach across metastatic organs.

This narrative review examines recent advances in drug delivery for breast cancer from the perspectives of tumor microenvironmental barriers, representative delivery modalities, therapeutic resistance, and clinical translation. We first discuss how physical, biochemical, and immune features of the tumor microenvironment influence drug distribution and delivery efficiency, and then compare the characteristics, limitations, and translational maturity of major delivery approaches. Particular attention is given to tumor heterogeneity, multidrug resistance, and the factors that limit the transition of promising delivery systems from preclinical studies to clinical use. Finally, we discuss emerging strategies for personalized delivery, formulation optimization, and responsive drug delivery, with emphasis on their practical requirements and translational constraints.

Tumor Microenvironment

The breast cancer tumor microenvironment comprises extracellular matrix components, biochemical gradients, and diverse stromal and immune cells that collectively influence drug penetration, distribution, and retention. Some of these features, including acidic pH, elevated glutathione, reactive oxygen species, and hypoxia, can also be exploited as endogenous triggers for responsive drug delivery. This section discusses three major aspects relevant to delivery efficiency: the three-dimensional collagen barrier, biochemical abnormalities, and interactions between nanoparticles and tumor-associated macrophages (Figure 1).

Figure 1.

Decisive Barriers in the Tumor Microenvironment of Breast Cancer. The breast cancer TME imposes three core barriers to drug delivery: (1) a dense three-dimensional collagen network acting as a physical barrier; (2) biochemical abnormalities (low pH, high GSH, high ROS, and hypoxia) that serve both as obstacles and as endogenous triggers for responsive design; and (3) tumor-associated macrophages that capture nanoparticles but can also be used in macrophage-mediated delivery strategies. The interplay among these barriers influences nanomedicine penetration and therapeutic outcomes.

The Physical Barrier Effect of the Three-Dimensional Collagen Network

The tumor microenvironment is a dynamic environment that contributes to tumor progression and therapeutic resistance. In breast cancer, particularly invasive ductal carcinoma, pronounced fibrotic reaction (ie, desmoplasia) constitutes the primary physical barrier to drug delivery.3 Collagen, as the main structural component of the extracellular matrix (ECM), undergoes excessive deposition and crosslinking that increase tumor tissue stiffness and create a dense physical barrier to nanomedicine penetration. Mechanistically, this dense collagen network impairs drug delivery efficiency through two principal means: on the one hand, high-density and oriented collagen fibers restrict the diffusion of nanoparticles exceeding a certain size via steric hindrance; on the other hand, elevated interstitial fluid pressure resulting from the collagen network and its crosslinking weakens the passive targeting effect mediated by the enhanced permeability and retention (EPR) effect, collapses intratumoral blood vessels, and further hinders drug extravasation from the bloodstream into the tumor parenchyma.11 Therefore, a thorough understanding of the spatial physical characteristics of the collagen network is a prerequisite for overcoming this delivery barrier and designing efficient nanomedicines.

Studies have shown that the content of type I collagen secreted by cancer-associated fibroblasts in breast tumors can be several to tens of times higher than that in normal tissues.12 In addition, chemotherapy-resistant cell lines not only upregulate the expression of drug efflux pumps (eg, P-glycoprotein) but also secrete more collagen and remodel the ECM, thereby establishing combined cellular and microenvironmental resistance.13 This implies that even if nanocarriers successfully evade the drug efflux mechanisms of tumor cells, the dense collagen network still significantly restricts their deep penetration into tumors, preventing drugs from reaching effective concentrations. Notably, recent microphysiological system studies have revealed significant differences in collagen deposition patterns among breast cancer subtypes: for instance, collagen deposition levels in SKBR3 and MDA-MB-468 cell models are considerably higher than those in the MCF7 model, providing important experimental evidence for individualized assessment of physical barriers and the development of tailored therapeutic regimens.14 In summary, the collagen network not only acts as a physical barrier but also intertwines closely with drug resistance, together posing a dual challenge that urgently needs to be addressed in breast cancer therapy.

From a therapeutic strategy perspective, adjuvant strategies targeting the collagen-rich stroma are emerging. For example, the combination of collagen-degrading enzymes (such as hyaluronidase or collagenase) can temporarily open channels for drug penetration, whereas antifibrotic agents like losartan reduce collagen deposition by inhibiting the transforming growth factor-beta signaling pathway, thereby lowering tumoral interstitial pressure.11,15 Furthermore, recently developed pH- and glutathione-dual-responsive nanoparticles, functionalized with surface thiolated collagenase, have achieved effective penetration of the dense collagen barrier and significantly enhanced the accumulation of chemotherapeutic drugs in deep tumor regions.11 However, it is important to be cautious that excessive degradation of the ECM might disrupt its role as a “natural barrier” restricting tumor cells, thereby inadvertently promoting tumor cell invasion and metastasis. Hence, controlled modulation rather than extensive depletion of the stromal barrier may provide a more appropriate strategy, balancing improved drug penetration against the potential risk of facilitating tumor invasion. Approaches targeting stromal components or incorporating matrix-responsive delivery mechanisms therefore require careful evaluation of both delivery efficiency and biological consequences.

Impact of Biochemical Abnormalities in the Microenvironment on Drug Delivery

In addition to physical barriers, the biochemical abnormalities of the breast cancer microenvironment profoundly influence the behavior of drug delivery systems. Low pH (6.5–6.8), high glutathione concentrations (reaching 2–10 mM of intracellular levels), elevated reactive oxygen species (ROS), and hypoxic conditions constitute the hallmark features of the tumor microenvironment.16 These biochemical characteristics can influence nanocarrier stability, cellular uptake, and drug-release kinetics. At the same time, pH, glutathione, reactive oxygen species, and hypoxia can serve as endogenous triggers for responsive delivery systems.17 Their utility, however, depends on the magnitude, spatial distribution, and interpatient variability of these signals, which may limit the consistency of stimulus-responsive drug release in vivo. Accordingly, exploiting tumor-associated biochemical gradients requires consideration not only of responsiveness but also of signal heterogeneity and pharmacokinetic behavior.

pH-sensitive delivery systems represent one of the most mature strategies in this field. By introducing ionizable groups (eg, tertiary amines, carboxyl groups) or acid-labile chemical bonds (eg, hydrazone bonds, acetal bonds), nanocarriers remain stable in the neutral environment of the bloodstream but undergo conformational changes or chemical cleavage upon entering the acidic tumor microenvironment, thereby releasing the encapsulated drug.18 For example, chitosan-based nanoparticles, owing to their pKa value of approximately 6.5, become protonated under acidic tumor conditions, which not only triggers drug release but also enhances cellular uptake efficiency.19 In addition to hydrogel- and nanoparticle-based delivery systems, pH-responsive mesoporous silica nanoparticles have also been investigated; by surface modification with polymers such as polyethyleneimine, they can achieve “gatekeeping” drug release under acidic conditions.20 Furthermore, calcium carbonate-based lipid-coated nanoparticles exhibit pH-responsive degradation, enabling efficient targeted delivery of chemotherapeutic agents.21 These studies demonstrate the feasibility of pH-responsive release in experimental systems. However, the relatively modest and spatially heterogeneous pH differences observed in tumors may affect release specificity and reproducibility in vivo, and these factors remain important considerations for clinical translation.

Glutathione-responsive systems leverage the highly reductive environment characteristic of tumor cells. Nanocarriers containing disulfide bonds undergo disulfide bond cleavage in the presence of glutathione, achieving “reduction-triggered” drug release.22 Such systems are particularly suitable for delivering therapeutics that require cytosolic release, such as siRNA and certain chemotherapeutic drugs. In addition, some researchers have integrated both pH- and glutathione-responsive mechanisms into a single platform to construct multi-responsive delivery systems that release drugs when both conditions are simultaneously satisfied.11 For instance, a dual-responsive nanoparticle based on maleimide-functionalized polyethylene glycol and polylactide was developed; by surface modification with a protective layer of thiocollagenase and chondroitin sulfate, it achieved deep tumor penetration and CD44 targeting, demonstrating anti-breast cancer efficacy.11 In addition, core–shell nanoparticles with dual responsiveness to ATP and hyaluronidase have been developed to achieve precise drug delivery and apoptosis induction by mimicking the intracellular tumor microenvironment.23 Such multi-responsive systems can improve conditional control over drug release in experimental models. However, the addition of multiple responsive components also increases formulation complexity, characterization requirements, and the potential for batch-to-batch variability, which should be weighed against the incremental benefit of each additional trigger.

In addition to pH and glutathione responsiveness, reactive oxygen species (ROS)-responsive systems have been increasingly investigated. Given the significantly elevated ROS levels in the tumor microenvironment, researchers have designed nanocarriers containing oxalate ester bonds, thioether bonds, or selenium bonds, which undergo cleavage in the presence of ROS, thereby enabling specific drug release.24 For example, co-encapsulation of flurocoxib Q and kenocoxib A into micellar nanoparticles achieved ROS-triggered drug release and fluorescence activation, allowing real-time visualized monitoring of breast cancer therapy.25 Meanwhile, hypoxia-responsive systems, through the introduction of hypoxia-sensitive moieties such as nitroimidazoles or azobenzenes, enable specific drug release in hypoxic regions of tumors, further improving spatiotemporal precision of treatment.26

Together, these systems illustrate how distinct biochemical features of the tumor microenvironment can be exploited as triggers for stimulus-responsive drug release. Overall, biochemical abnormalities in the tumor microenvironment provide useful design cues for stimulus-responsive delivery systems, but their translational value depends on the reproducibility and spatial consistency of these signals in human tumors. Rather than increasing the number of responsive elements per se, future development should prioritize systems in which added complexity provides a demonstrable advantage in release control, pharmacokinetics, safety, or therapeutic efficacy. Detailed characterization of in vivo responsiveness, biodistribution, and manufacturing reproducibility will therefore be necessary before these designs can be translated reliably into clinical use.

Interplay Between the Microenvironmental Immune Status and Drug Delivery Systems

Immune cells infiltrating the tumor microenvironment, particularly tumor-associated macrophages (TAMs), represent another critical factor regulating the behavior of delivery systems.27 In breast cancer, TAMs typically exhibit an M2-polarized state: they not only secrete immunosuppressive cytokines (eg, IL-10, TGF-β) but also physically “trap” nanoparticles, thereby hindering their access to tumor cells. Mechanistically, the cGAS-STING signaling pathway has emerged as a central regulator of macrophage activation and polarization, thereby bridging innate immune sensing with TAM functional plasticity.28 In addition, the cGAS-STING pathway orchestrates a “double-edged sword” in immune-mediated tissue remodeling: while excessive activation drives chronic inflammation and fibrosis, moderate pathway engagement facilitates beneficial macrophage polarization and tissue repair.29 Notably, the non-specific uptake of nanocarriers by TAMs arises not only from their inherent phagocytic function but also from the regulation of chemokines and cytokines within the tumor microenvironment. Moreover, studies have demonstrated that TAM abundance is significantly correlated with poor prognosis in breast cancer patients, rendering them highly attractive therapeutic targets.27 Additionally, the physicochemical properties of nanomedicine delivery systems, such as particle size, surface charge, and ligand modification, largely determine their interaction fate with TAMs, providing crucial guidance for the rational design of delivery vectors. Consequently, a deep understanding of the interaction mechanisms between TAMs and nanocarriers is a prerequisite for achieving precise drug delivery.

This phenomenon presents both challenges and potential opportunities. On one hand, nonspecific nanoparticle uptake by TAMs can reduce the fraction of administered material reaching tumor cells. On the other hand, TAM-directed delivery has been explored to modulate macrophage function, including delivery of agents such as clodronate disodium or zoledronic acid to alter the tumor immune microenvironment.27 In addition, macrophage-mediated relay delivery has been explored in preclinical cancer models, in which nanoparticles are initially internalized by macrophages and therapeutic cargo is subsequently transferred to neighboring tumor cells through extracellular vesicles or tunneling nanotubes.30,31 These findings suggest that macrophage-nanoparticle interactions may be exploited therapeutically, although the efficiency, reproducibility, and in vivo relevance of such transfer mechanisms require further validation.

Beyond macrophages, other immune cells, including natural killer (NK) cells, have also been investigated as potential delivery vehicles. For example, membrane-preserved NK cells generated by controlled freezing and thawing have been reported to retain tumor-directed interactions and to support delivery of therapeutic cargo.32 Engineered macrophages have similarly been explored as cellular carriers.33 Although these approaches exploit endogenous cell-tumor interactions, most evidence remains preclinical, and important issues including cell-source variability, manufacturing consistency, cargo loading, biodistribution, safety, and scalability must be addressed before their clinical utility can be established.

Nevertheless, it must be acknowledged that current understanding of microenvironment–delivery system interactions remains at an early stage. The spatiotemporal heterogeneity of the tumor microenvironment, the distinct microenvironmental characteristics of different breast cancer subtypes, and the dynamics of microenvironmental remodeling induced by therapeutic interventions are all important directions requiring in-depth investigation in future research. In particular, the high plasticity of TAMs and their functional transitions during different stages of tumor progression necessitate the development of more dynamic and precise intervention strategies. These uncertainties highlight the need for dynamic characterization of immune-carrier interactions across tumor subtypes and treatment stages before such strategies can be reliably incorporated into precision delivery approaches.

Diverse Platforms for Drug Delivery Systems

Drug delivery approaches for breast cancer encompass technologically distinct modalities that differ in administration route, carrier composition, targeting mechanism, manufacturing requirements, and level of clinical maturity. Representative examples include lipid-based systems and polymeric nanoparticles as systemic nanocarriers, hydrogels as predominantly local depot systems, extracellular vesicles as biologically derived carriers, and antibody–drug conjugates (ADCs) as molecularly targeted drug conjugates. Rather than representing equivalent technological classes, these modalities provide complementary strategies for addressing different therapeutic and delivery challenges in breast cancer (Figure 2), with their major characteristics and translational maturity summarized in Table 1.

Figure 2.

Diverse Drug Delivery Platforms for Breast Cancer. Five major delivery platforms represent distinct approaches in breast cancer drug delivery. Liposomes (eg, liposomal doxorubicin) pioneered nanodelivery; polymeric nanoparticles (especially PLGA) offer biodegradable and tunable properties, while carrier-free prodrug nanoassemblies provide an alternative strategy with high theoretical drug loading; hydrogels enable local postoperative therapy; extracellular vesicles provide biologically derived carrier properties; and antibody-drug conjugates (ADCs, eg, T-DXd) have achieved substantial clinical maturity. Each platform is annotated with its key advantage and a representative challenge.

Table 1.

Comparative Characteristics and Translational Maturity of Major Drug Delivery Platforms for Breast Cancer

Platform	Typical Administration/Delivery Mode	Representative Cargo	Major Advantage	Major Limitation	Manufacturing Complexity	Clinical Maturity	Representative Example
Liposomes	Systemic	Small-molecule drugs	Established formulation technology; prolonged circulation; surface functionalization	Variable tumor accumulation; limited benefit of active targeting in clinical translation	Moderate	Clinical	Liposomal doxorubicin
Solid lipid nanoparticles	Systemic	Small molecules; nucleic acids	Physicochemical stability; controlled release; combination loading	Lipid polymorphic transitions and drug leakage during storage	Moderate	Preclinical/early translational	Experimental SLN formulations
Polymeric nanoparticles	Systemic	Small molecules; proteins; nucleic acids	Flexible composition, degradation, and surface modification	Formulation complexity; degradation-related instability; scale-up and reproducibility	Moderate–high	Predominantly preclinical	PLGA nanoparticles
Carrier-free/prodrug nanoassemblies	Systemic	Self-assembling prodrugs	High theoretical drug fraction; reduced inert carrier content	Aggregation, pharmacokinetic uncertainty, off-target toxicity, manufacturing reproducibility	Moderate–high	Preclinical	Self-assembled prodrug nanoparticles
Hydrogels	Local/postoperative	Chemotherapeutics; antibodies; immune agonists	Local retention and sustained release	Restricted mainly to local administration; degradation safety; mechanical and release-control requirements	Moderate–high	Predominantly preclinical	Injectable/implantable hydrogels
Extracellular vesicles	Systemic/local	Small molecules; siRNA and other nucleic acids	Biologically derived membrane interface; engineering potential	Heterogeneous composition, low loading efficiency, rapid clearance, difficult GMP-scale production	High	Preclinical/early translational	Engineered exosomes
Antibody–drug conjugates	Systemic	Cytotoxic payload conjugated to antibody	Biomarker-directed delivery; established clinical efficacy	Antigen heterogeneity, resistance, toxicity, linker/payload constraints	High but standardized industrially	Established clinical use	T-DM1, T-DXd, sacituzumab govitecan, datopotamab deruxtecan

Lipid-Based Delivery Systems

As the first nanocarriers to enter clinical application, liposomes have a research history spanning more than half a century. In the context of breast cancer therapy, the success of liposomal doxorubicin not only validated the feasibility of the nanodelivery concept but also revealed several advantages of lipid-based carriers including biocompatibility, the ability to co-encapsulate both hydrophilic and hydrophobic drugs, and facile surface functionalization.4,34 Nevertheless, despite the growing maturity of liposome technology, clinical translation continues to face challenges, most notably, improving targeting efficiency, overcoming tumor microenvironment barriers, and achieving more precise drug release control.35 Consequently, the development of functional liposomes is focused on two major directions.

First, actively targeted liposomes have been developed. By conjugating specific ligands (eg, folate, transferrin, antibody fragments, or aptamers) onto the liposomal surface, researchers aim to improve targeting beyond the passive EPR effect.36 For instance, immunoliposomes modified with the anti-HER2 antibody trastuzumab can improve the therapeutic index in HER2-positive breast cancer through a dual mechanism: active targeting and receptor-mediated endocytosis.37 Furthermore, targeted liposomes engineered using the Kisspeptin/GPR54 system have demonstrated enhanced tumor accumulation and prolonged median survival time, further expanding the scope of active targeting strategies.38 However, despite repeated success in preclinical studies, no actively targeted liposome product has yet been approved for clinical use, suggesting that the pharmacokinetic alterations and potential immunogenicity associated with ligand modification require further optimization.39 Indeed, balancing targeting efficiency with systemic safety during translation from bench to bedside remains a central challenge in the field.

Second, solid lipid nanoparticles (SLNs) have garnered widespread attention as an alternative lipid-based delivery platform to liposomes.6 Unlike liposomes with a liquid lipid core, SLNs use lipids that are solid at room temperature (eg, tristearin, glyceryl monostearate) as the matrix, providing drug encapsulation stability and controlled release characteristics. SLNs can be surface-modified with polyethylene glycol to prolong circulation or conjugated with targeting ligands to promote tumor-directed delivery, and can also co-encapsulate chemotherapeutic agents and nucleic acid drugs for combination therapy.6 Beyond conventional single-agent chemotherapy delivery, SLNs have been investigated in combination with photodynamic therapy, photothermal therapy, immunotherapy, and gene regulation, including strategies aimed at addressing multidrug resistance in breast cancer.6 Nevertheless, polymorphic transitions of lipids during storage can lead to drug leakage, representing an important stability challenge for clinical translation.40 Overall, the development of conventional liposomes, functionalized liposomes, and solid lipid nanoparticles illustrates the expanding range of lipid-based delivery strategies available for breast cancer therapy.

Polymeric Nanoparticles

Polymeric nanoparticles represent a versatile class of platforms in the field of drug delivery. Ranging from natural polymers (eg, chitosan, gelatin, sodium alginate) to synthetic polymers (eg, PLGA, PEG-PLA, PCL), the wide array of available materials affords researchers substantial design flexibility, enabling customization of polymeric nanoparticles to meet diverse therapeutic needs.41,42 This diversity is reflected not only in the source of materials but also in substantial variations in degradation behavior, drug release kinetics, and surface functionalization potential, thereby providing multiple options for addressing the multifaceted barriers in breast cancer therapy. Consequently, an understanding of the properties of different polymeric materials is essential for the rational design of delivery systems.

Poly(lactic-co-glycolic acid) (PLGA) is a widely used polymer in nanoparticle-based drug delivery because of its biodegradability and biocompatibility.43 PLGA nanoparticles can encapsulate various therapeutic agents, ranging from small-molecule drugs and proteins to nucleic acids, and their degradation rate can be adjusted by altering the monomer ratio of lactic acid to glycolic acid, thereby enabling sustained or pulsatile drug release. However, the acidic microenvironment generated during PLGA degradation may cause denaturation and inactivation of encapsulated protein drugs, a limitation that is particularly pronounced in the delivery of biologics, which has spurred the development of more protective carrier designs. To address this challenge, researchers have explored strategies to composite PLGA with other functional materials. For example, loading γ-oryzanol into PLGA nanoparticles can improve its water solubility and enhance its antitumor efficacy through the protective effect of the nanocarrier, providing a potential approach for the delivery of natural products.43 Thus, regulation of PLGA degradation behavior and functional modification are important for delivery performance.

To overcome the limitations of conventional polymeric nanoparticles, researchers have developed a variety of strategies. Among these, the construction of multilayered structures on nanoparticle surfaces using layer-by-layer self-assembly technology enables the orderly encapsulation and time-sequenced release of multiple drugs, a strategy that mimics the hierarchical assembly of biomacromolecules in nature.44 For instance, a co-delivery system for tamoxifen and resveratrol has been reported to enhance antitumor efficacy and improve pharmacokinetic properties and tumor targeting through layer-by-layer self-assembly modification.44 Furthermore, biomimetic membrane coating represents another strategy: by wrapping natural cell membranes (eg, red blood cell membranes, platelet membranes, cancer cell membranes) around a polymeric core, nanoparticles can acquire immune-evasion and tumor-homing properties, thereby prolonging in vivo circulation time and enhancing targeting efficiency.45 This biomimetic strategy utilizes the biomolecular composition of cell membrane surfaces to confer biological functions on synthetic carriers and may help address challenges such as multidrug resistance. The common goal of these strategies is to achieve greater spatiotemporal control over the drug delivery process.

An emerging strategy in polymeric drug delivery is the development of carrier-free or prodrug nanoassemblies. In these systems, chemical modification of drug molecules can generate amphiphilic prodrugs capable of self-assembly without requiring a conventional inert carrier. This approach can increase the theoretical proportion of active drug within the formulation and reduce the amount of additional carrier material.46 However, the absence of a conventional carrier does not eliminate formulation-related risks. The resulting nanoassemblies may still be affected by aggregation, chemical stability, pharmacokinetics, metabolism, off-target toxicity, and manufacturing reproducibility. Multi-drug prodrug assemblies can additionally incorporate complementary therapeutic functions, but increasing structural complexity may complicate characterization, scale-up, and control of drug ratios and release kinetics. Thus, although carrier-free nanoassemblies provide an attractive strategy for increasing drug loading and simplifying carrier composition, their clinical value will depend on reproducible manufacturing, in vivo stability, pharmacokinetic behavior, and demonstration of advantages over established formulations.

Hydrogels

Compared with systemically administered nanoparticles, hydrogels are primarily suited to local drug delivery, particularly for in situ treatment of the postoperative tumor cavity.47 As three-dimensional networks composed of hydrophilic polymer chains, hydrogels can serve as local drug depots that provide sustained release and maintain relatively high drug concentrations at the administration site. This local delivery strategy may reduce systemic exposure while improving drug availability within the surgical cavity, making hydrogels particularly relevant to postoperative treatment settings, including triple-negative breast cancer.48 However, their clinical utility depends on achieving appropriate mechanical properties, predictable degradation, controlled release kinetics, and compatibility with surrounding breast tissue.

Injectable hydrogels have evolved from thermosensitive systems toward formulations incorporating multiple responsive mechanisms. Specifically, first-generation thermosensitive hydrogels (eg, Poloxamer 407) exploit body temperature-induced sol–gel phase transition for in situ gelation; however, their insufficient mechanical strength and burst drug release limit their widespread application.49 To overcome these limitations, second-generation responsive hydrogels integrate multiple response mechanisms; for example, in addition to temperature, they respond to endogenous or exogenous stimuli such as pH, enzymes, redox potential, light, and magnetic fields.22 In one illustrative study, a dual-network hydrogel based on N-carboxyethyl chitosan and oxidized chondroitin sulfate was developed, enabling responsive drug release under the acidic pH and glutathione (GSH) conditions of the tumor microenvironment, thereby enhancing antitumor efficacy.22 Furthermore, “micelle-in-hydrogel” composite systems, in which drug-loaded micelles are dispersed within a hydrogel matrix, create a multilevel “hydrogel-encapsulated-micelles” delivery structure. This design exploits the hydrogel for spatial confinement and sustained local release, while the micelles improve the solubility and cellular uptake efficiency of hydrophobic drugs.50 These developments illustrate how hydrogel formulations can be engineered to combine local retention with stimulus-responsive release and improved solubilization of hydrophobic drugs. However, increasing formulation complexity also introduces additional requirements for physicochemical characterization, reproducibility, and control of release kinetics, which should be considered when evaluating their translational potential.

Hydrogels have also been investigated for postoperative adjuvant therapy in breast cancer. Implantable or injectable formulations can maintain therapeutic agents locally within the tumor cavity and may therefore provide an alternative to repeated systemic administration.51 For example, an implantable delivery system containing anti-PD-1 antibodies and a CD40 agonist was reported to promote local antigen-presenting cell activation and systemic antitumor immune responses in experimental models.48 Hydrogels have also been combined with photothermal therapy, gene therapy, and other modalities to explore multimodal control of residual disease and recurrence.52 Nevertheless, these studies remain largely preclinical, and the extent to which such local delivery strategies can provide reproducible clinical benefit has yet to be established.

The clinical translation of hydrogels faces multiple challenges. First, an important issue is the long-term safety of hydrogel degradation products in vivo. In particular, when chemical crosslinkers (eg, glutaraldehyde, carbodiimide) are used to construct the hydrogel network, residual crosslinkers or their degradation by-products may provoke local inflammatory responses and even compromise therapeutic efficacy.53 Second, the physicochemical and mechanical properties of hydrogels require careful optimization, while batch-to-batch consistency, scalable manufacturing, and standardized quality control remain important challenges for clinical translation.47 Finally, controlling the release kinetics of multiple therapeutic agents (eg, chemotherapeutics, immune agonists, antibodies) from hydrogels to achieve appropriate temporal coordination remains an important challenge. Accordingly, clinical translation of hydrogel-based delivery systems will depend on standardized manufacturing, validated degradation and safety profiles, reproducible release characteristics, and demonstration of benefit over existing local and systemic treatment approaches.

Extracellular Vesicles

Extracellular vesicles (EVs), including exosomes, have attracted considerable interest as biologically derived drug carriers.54 Their membrane composition and endogenous cellular origin provide opportunities for transporting small molecules and nucleic acids and for engineering cell- or tissue-directed interactions. However, these characteristics are strongly dependent on the source cell, production conditions, and surface composition of the vesicles. EV preparations can therefore differ substantially in biodistribution, biological activity, immunological properties, and cargo composition, while rapid clearance and inconsistent loading efficiency remain important limitations. These advantages and uncertainties distinguish EVs from conventional synthetic nanocarriers and need to be considered together when assessing their translational potential.

In triple-negative breast cancer (TNBC), exosome-based delivery systems have been investigated as biologically derived carriers.54 Owing to the lack of estrogen receptors, progesterone receptors, and human epidermal growth factor receptor 2 on the surface of TNBC cells, which limits the applicability of receptor-directed strategies, therapeutic options remain relatively restricted.54 Exosomes contain surface adhesion molecules and other membrane-associated components that may influence interactions with tumor cells and the tumor microenvironment. Engineered exosomes have been investigated for the delivery of chemotherapeutic agents and nucleic acid therapeutics, including doxorubicin and siRNA, in TNBC models.55,56 Moreover, strategies such as displaying multiple homing peptides (eg, iRGD and tLyp1) on the exosomal surface can further enhance tumor cell targeting efficiency and uptake.57 Thus, exosomes represent a versatile platform for exploring targeted delivery strategies in TNBC.

Nevertheless, despite encouraging preclinical findings, the clinical translation of exosome-based therapies faces several important challenges. Foremost among these is standardized production: exosomes derived from different cell sources and cultured under varying conditions exhibit substantial heterogeneity in quantity, size, content, and functionality, complicating the establishment of reliable quality-control standards.54 Furthermore, limited and variable drug-loading efficiency remains an important constraint, particularly for passive loading methods such as simple co-incubation. In addition, large-scale production of Good Manufacturing Practice (GMP)-grade exosomes remains costly, while data on in vivo pharmacokinetics and long-term safety are still limited. These factors continue to constrain clinical development.

From a translational perspective, exosome-based systems may be developed either as standalone biological carriers or in combination with synthetic nanomaterials. For example, biomimetic nanoparticles in which synthetic nanoparticles are coated with exosomal membranes aim to combine the drug-loading capacity and controllability of synthetic carriers with selected biological properties of exosomal membranes.58 Such hybrid systems have been reported to improve tumor accumulation, penetration, and antitumor activity in experimental models.59 Similarly, nanoplatforms coated with other cell membranes, such as platelet membranes, have been investigated for tumor-targeted delivery and multidrug resistance.60,61 Overall, hybrid biointerface-based nanoplatforms provide another strategy for integrating biological and synthetic delivery components, but their translational value will depend on reproducible manufacturing, defined composition, pharmacokinetic characterization, and demonstration of benefit over simpler delivery systems.

Antibody-Drug Conjugates

Among drug delivery modalities discussed in this review, ADCs have achieved the highest level of clinical translation. Several ADCs have received regulatory approval for defined breast cancer populations. Ado-trastuzumab emtansine (T-DM1; Kadcyla) was approved by the US Food and Drug Administration (FDA) in 2013 for HER2-positive metastatic breast cancer previously treated with trastuzumab and a taxane. Trastuzumab deruxtecan (T-DXd; Enhertu), initially approved by the FDA for HER2-positive advanced breast cancer, subsequently received approval in 2022 for unresectable or metastatic HER2-low breast cancer and in 2025 for hormone receptor-positive, HER2-low or HER2-ultralow unresectable or metastatic breast cancer after endocrine therapy. Sacituzumab govitecan (Trodelvy), a Trop-2-directed ADC, has received FDA approvals for metastatic triple-negative breast cancer and, in 2023, for unresectable locally advanced or metastatic HR-positive/HER2-negative breast cancer after endocrine-based therapy and at least two additional systemic therapies in the metastatic setting. Datopotamab deruxtecan (Datroway), another Trop-2-directed ADC, was approved by the FDA in 2025 for unresectable or metastatic HR-positive/HER2-negative breast cancer following prior endocrine-based therapy and chemotherapy. These approvals demonstrate that ADCs have progressed beyond preclinical proof-of-concept to established clinical use, although their applicability remains determined by antigen expression, disease subtype, prior therapy, and toxicity profile.

ADCs consist of three principal components: a tumor-targeting antibody, a linker, and a cytotoxic payload.62 Their therapeutic performance depends on coordinated optimization of antigen specificity, linker stability, internalization, intracellular processing, and payload potency. The choice of linker is particularly critical. Cleavable linkers (eg, hydrazone, disulfide, and peptide linkers) exploit characteristic features of the tumor microenvironment (such as low pH, high glutathione concentration, or specific enzyme expression) to release the drug specifically within or near target cells, thereby enabling a “bystander killing effect”, where the drug diffuses from antigen-positive cells to adjacent antigen-negative tumor cells. This bystander effect may help address antigen heterogeneity within tumors. In contrast, non-cleavable linkers require complete degradation of the antibody within lysosomes to release the payload. Although such designs can improve stability in systemic circulation and limit premature payload release, their bystander effect is generally more limited, which may pose limitations when treating heterogeneous tumors. Thus, linker design requires balancing payload release, systemic stability, bystander activity, and safety. Consequently, linker characteristics are important determinants of the therapeutic window and clinical performance of an ADC.

The clinical development of T-DXd has expanded the therapeutic relevance of HER2 expression beyond the conventional HER2-positive category. FDA approval was extended to HER2-low metastatic breast cancer in 2022 and, in 2025, to selected HR-positive HER2-low or HER2-ultralow disease, illustrating how ADC efficacy has reshaped clinically actionable HER2 expression thresholds. This development also increases the importance of accurate and reproducible HER2 assessment. At the same time, HER2-low status should not be interpreted as establishing an entirely new biological subtype, because it is primarily a treatment-relevant biomarker category rather than a distinct molecular entity.

Despite their clinical efficacy, resistance can develop at multiple stages of ADC action. The mechanisms of ADC resistance include downregulation or mutation of tumor antigens leading to reduced antibody binding, aberrant endocytic pathways that impair ADC internalization, lysosomal dysfunction that limits payload release, and upregulation of drug efflux pumps such as P-glycoprotein.63 Current strategies to address these mechanisms include bispecific ADCs that simultaneously target two tumor antigens (eg, HER2 and HER3), dual-payload ADCs that combine cytotoxic agents with different mechanisms of action, and alternative payload classes, including topoisomerase I and RNA polymerase II inhibitors.62 These approaches are intended to address antigen heterogeneity and acquired resistance, but their increased structural complexity may also create additional challenges in characterization, pharmacokinetics, toxicity management, manufacturing consistency, and regulatory development. Accordingly, future ADC development should focus on demonstrating that individual design modifications provide clinically meaningful benefits in appropriately selected patient populations rather than assuming that greater molecular complexity will necessarily improve therapeutic performance.

Current Core Challenges

Breast cancer heterogeneity, multidrug resistance, and sequential biological barriers collectively limit the effectiveness of drug delivery strategies. This section examines how inter- and intratumoral heterogeneity influence delivery efficiency, summarizes the major mechanisms underlying multidrug resistance, and outlines the barriers that nanomedicines encounter from systemic circulation to intracellular targets. It also considers the distinct delivery challenges associated with major metastatic sites, including the brain, bone, and lung. These challenges emphasize the need to align delivery strategies with tumor subtype, resistance mechanisms, metastatic site, and the relevant biological barriers rather than relying on a uniform delivery approach.

Tumor Heterogeneity and Individual Variability in Delivery Efficiency

The heterogeneity of breast cancer represents an important obstacle to achieving precision drug delivery. This heterogeneity is manifested not only in intertumoral heterogeneity among different patients but also in intratumoral heterogeneity between primary and metastatic lesions within the same patient, and even across distinct regions of a single tumor.64 From a molecular subtyping perspective, significant differences exist among hormone receptor-positive, HER2-positive, and triple-negative breast cancers with respect to vascular permeability, stromal composition, and immune microenvironment characteristics, differences that directly influence the tumor distribution and therapeutic efficacy of nanomedicines. Furthermore, even within the same molecular subtype, divergent genetic mutation profiles, epigenetic modifications, and signaling pathway activities can result in markedly distinct responses to an identical delivery system. Recent studies further indicate that alterations in copper homeostasis may contribute to tumor-promoting signaling in TNBC.65 This heterogeneity limits the applicability of uniform nanomedicine strategies, because a single formulation may perform differently across patients or even among lesions within the same individual. Accordingly, delivery-system design should account for molecular subtype, lesion-specific microenvironmental features, and other determinants of drug distribution and response.

From a clinical translational perspective, the actual efficacy of passive targeting strategies, which center on the EPR effect, is highly dependent on the unique vascular permeability and lymphatic drainage functions of tumors. However, the magnitude of the EPR effect varies by orders of magnitude across different molecular subtypes, clinical stages, and individual patients. Clinical imaging studies have demonstrated substantial interpatient and interlesional variability in nanoparticle tumor accumulation. In patients with HER2-positive metastatic breast cancer, ^64Cu-MM-302 PET imaging revealed marked variability in nanoparticle accumulation across tumor lesions.66 The root cause of this variability lies in the heterogeneity of tumor vascular architecture: some tumors exhibit highly leaky, immature blood vessels, whereas others possess relatively intact vascular barriers. Additionally, elevated interstitial fluid pressure, dense extracellular matrix (particularly the collagen network), and differential distribution of tumor-associated macrophages can all profoundly influence the extravasation and intratumoral diffusion efficiency of nanomedicines.3,67 These findings suggest that a universal nanocarrier design strategy is unlikely to achieve uniform performance in clinical practice. This variability supports the development of approaches that can predict delivery efficiency and identify patients most likely to benefit from nanoparticle-based treatment.

Addressing this heterogeneity will require more personalized and predictable delivery strategies. Potential approaches include: identifying biomarkers that can screen “EPR effect responders” to guide patient stratification, for example, assessing tumor vascular permeability via dynamic contrast-enhanced magnetic resonance imaging, or analyzing vascular endothelial growth factor expression levels and stromal characteristics in biopsy samples; designing “preconditioning” strategies that actively modulate the tumor microenvironment, such as using angiotensin II receptor antagonists like losartan to improve vascular permeability and drug distribution,68,69 or employing transforming growth factor-β inhibitors to reduce stromal fibrosis and enhance intratumoral penetration of nanomedicines; and promoting the standardized application of patient-derived xenograft (PDX) models in preclinical nanomedicine evaluation to more accurately predict clinical efficacy. Moreover, liquid biopsy techniques (eg, analysis of circulating tumor DNA and extracellular vesicles) hold promise for real-time monitoring of the dynamic evolution of tumor heterogeneity, thereby providing a basis for adaptive adjustments to delivery strategies.70 Together, these approaches may help identify patients or lesions more likely to benefit from specific delivery strategies. However, their clinical implementation will require prospective validation of predictive biomarkers, standardized imaging or molecular assessment, and evidence that patient stratification improves outcomes compared with unselected treatment. These subtype- and metastatic site-specific considerations are summarized in Table 2.

Table 2.

Breast Cancer Subtype- and Metastatic Site-Specific Considerations for Drug Delivery

Clinical Context	Major Biological/Delivery Barrier	Actionable Feature or Target	Potential Delivery Strategy	Current Evidence Maturity
HR-positive/HER2-negative breast cancer	Variable receptor expression; endocrine resistance; heterogeneous vascular permeability	ER-related signaling; resistance pathways; lesion-specific vascular features	Nanocarrier-based combination delivery; biomarker-guided formulations	Mainly preclinical; conventional systemic therapies remain dominant
HER2-positive breast cancer	HER2 heterogeneity; variable nanoparticle accumulation; acquired resistance	HER2	HER2-directed ADCs; HER2-targeted nanocarriers	High for ADCs; preclinical for most nanocarriers
HER2-low/HER2-ultralow breast cancer	Low and heterogeneous HER2 expression	Low-level HER2 expression	HER2-directed ADCs with bystander activity	Established clinical relevance for selected ADCs
Triple-negative breast cancer	Lack of ER, PR, and HER2; high molecular heterogeneity; immune and stromal barriers	Trop-2; immune pathways; TME features; CD44 and other experimental targets	ADCs; stimulus-responsive nanoparticles; EVs; combination delivery	ADCs clinically established in defined settings; most nanoparticle/EV strategies preclinical
BRCA1/2-mutated breast cancer	DNA-repair deficiency; acquired resistance	Homologous recombination deficiency/PARP pathway	Delivery of DNA-damage or pathway-targeted combinations	Clinical relevance of molecular targeting established; DDS-specific evidence limited
Brain metastasis	BBB/BTB; limited and heterogeneous intracranial drug penetration	BBB/BTB transport pathways; tumor-associated receptors	BBB-penetrating or receptor-mediated nanocarriers; multitargeted carriers; ultrasound-assisted delivery	Predominantly preclinical; limited early translational evidence
Bone metastasis	Specialized mineralized bone microenvironment; inefficient drug delivery; bone remodeling and bone–tumor interactions	Hydroxyapatite; bone-associated targets; tumor-associated receptors	Bisphosphonate-functionalized or other bone-affinitive nanocarriers; responsive nanocarriers	Predominantly preclinical
Lung metastasis	Distributed metastatic foci; limited lesion permeability; pulmonary vascular and inflammatory microenvironment	Pulmonary circulation; inflammatory/metastatic-site homing features	Biomimetic or pulmonary-retentive carriers; targeted nanocarriers; combination delivery	Predominantly preclinical

Metastatic Site-Specific Delivery Barriers

Metastatic breast cancer presents organ-specific delivery challenges that differ substantially from those of primary tumors. In brain metastases, systemically administered agents must traverse the blood–brain barrier (BBB) or blood-tumor barrier (BTB), while drug penetration into intracranial lesions remains limited and heterogeneous. In addition to restricting drug penetration, the brain metastatic microenvironment exhibits distinct neuroimmune interactions and immunosuppressive features that may further influence therapeutic response.71 Nanomedicine-based strategies have therefore been explored to enhance transport across these barriers and increase local drug exposure.72 In a murine model of triple-negative breast cancer brain metastasis, BBB-penetrating amphiphilic polymer–lipid nanoparticles increased docetaxel exposure in the brain and prolonged survival compared with conventional docetaxel.73 Multitargeted terpolymer–lipid nanoparticles have also been designed to cross the BBB and simultaneously target metastatic tumor cells and tumor-associated macrophages, thereby reducing metastatic burden in preclinical models.74 In HER2-positive breast cancer brain metastasis, ferritin-trastuzumab nanoconjugates exploiting transferrin receptor 1-mediated transport enhanced trastuzumab brain delivery and reduced intracranial tumor growth in mice.75 Despite these encouraging findings, the available evidence remains predominantly preclinical, and clinical translation will require reproducible BBB/BTB penetration, consistent pharmacokinetics, and validation in patients with heterogeneous metastatic lesions.

Bone metastases present a different challenge because metastatic tumor cells are embedded within a highly specialized bone microenvironment involving the mineralized matrix, osteoclasts, osteoblasts, and continuous bone remodeling. Nanoparticle-based strategies for breast cancer bone metastasis have therefore included untargeted delivery, bone-targeted delivery, and cancer cell-targeted approaches.76 Bone-targeted nanocarriers commonly exploit the affinity of bisphosphonates for hydroxyapatite to improve localization within metastatic bone lesions. For example, alendronate-functionalized coordination polymer nanoparticles showed enhanced hydroxyapatite binding and preferential accumulation in bone metastatic lesions while reducing osteoclastic bone destruction in vivo.77 Similarly, zoledronate-conjugated mesoporous silica nanoparticles combined bone targeting with photothermal therapy and improved both tumor control and the local bone microenvironment in preclinical models.78 Beyond bone-affinity targeting, responsive systems have also been investigated; hypoxia-cleavable, RGD-modified PLGA nanoparticles enabled tumor-selective delivery of epigenetic drugs and reduced both primary tumor burden and bone metastasis in vivo.79 However, these strategies remain predominantly preclinical, and their clinical translation will require demonstration of selective accumulation in metastatic rather than healthy bone, reproducible pharmacokinetics, and long-term skeletal safety.

Lung metastases represent another distinct delivery environment characterized by distributed metastatic foci, limited permeability of metastatic lesions, and a pulmonary vascular and inflammatory context that can influence carrier localization and retention. Micro-combinatorial hydrogel particles have been developed to exploit the pulmonary circulation, increasing the colocalization and retention of docetaxel-loaded nanoparticles within metastatic foci and improving tumor control in preclinical models.80 Biomimetic approaches have also been explored. Macrophage–cancer hybrid membrane-coated PLGA nanoparticles showed enhanced accumulation at inflammatory and metastatic sites and improved targeting of breast cancer-derived lung metastases,81 whereas macrophage-derived exosome-coated nanoparticles exhibited prolonged circulation and preferential distribution to pulmonary metastatic nodules.82 More recently, degradable cRGD-modified DNAzyme nanocapsules have been developed for gene therapy of pulmonary metastatic breast cancer and suppressed metastatic lesions in vivo.83 Nevertheless, these strategies remain preclinical, and improved pulmonary accumulation in animal models should not be interpreted as evidence of clinically validated organ-specific targeting.

These site-specific differences indicate that delivery strategies developed for primary breast tumors cannot be assumed to perform similarly across metastatic organs, and anatomical barriers, local microenvironmental features, and the maturity of supporting evidence should all be considered when selecting or evaluating delivery platforms (Table 2).

Multidrug Resistance

Multidrug resistance (MDR) represents a primary obstacle to successful chemotherapy in breast cancer. Fundamentally, MDR is a cross-resistance phenomenon wherein tumor cells become resistant to chemotherapeutic agents with diverse structures and mechanisms of action. Classical MDR mechanisms involve the overexpression and functional upregulation of ATP-binding cassette (ABC) transporters, such as P-glycoprotein, breast cancer resistance protein (BCRP), and multidrug resistance-associated protein 1 (MRP1). These efflux pumps harness energy from ATP hydrolysis to actively extrude intracellular chemotherapeutic drugs, thereby maintaining sub-therapeutic intracellular drug concentrations and substantially diminishing the cytotoxic effects of chemotherapy.84,85 Thus, efflux pump-mediated drug efflux represents an important mechanism of multidrug resistance, although it is not the sole contributor.

However, drug resistance involves multiple mechanisms beyond drug efflux. First, cancer stem cells (CSCs) represent an important component of drug resistance. These cells, endowed with self-renewal and multilineage differentiation capacities, inherently overexpress various drug efflux pumps and often reside in a relatively quiescent cell cycle state. Consequently, their relative quiescence and high expression of drug-efflux mechanisms can reduce sensitivity to conventional chemotherapies that preferentially target proliferating cells, thereby contributing to tumor persistence, relapse, and acquired resistance. Second, epigenetic alterations, including DNA methylation, histone modifications, and non-coding RNA regulation, can dynamically modulate the expression of efflux pumps and related resistance genes without altering the DNA sequence, thereby contributing to the establishment and maintenance of drug resistance86 and providing a flexible molecular basis for tumor adaptation to unfavorable microenvironments. Furthermore, pervasive features of the tumor microenvironment (TME), such as hypoxia, low pH, high interstitial pressure, and specific cytokine signaling (eg, TGF-β, IL-6), can also induce resistance phenotypes in cancer cells,87,88 further exacerbating the complexity of MDR and posing substantial challenges to clinical treatment (Figure 3). Collectively, CSCs, epigenetic changes, and TME factors constitute a multilayered, dynamically evolving resistance network, rendering interventions targeting a single mechanism frequently ineffective.

Figure 3.

Multilayered Drug Resistance and Delivery Strategies. Multidrug resistance in breast cancer arises from a multilayered network: (A) efflux pumps (P-gp, BCRP, MRP1) expel drugs; (B) cancer stem cells (CSCs) exhibit quiescence and high efflux activity; (C) epigenetic changes and TME factors (hypoxia, low pH, cytokines) further promote resistance. Nanomedicine countermeasures include co-delivery with efflux inhibitors, non-canonical endocytosis, mitochondrial targeting, and ferroptosis induction.

Nanoscale drug delivery systems have been investigated as a means of addressing several components of MDR through co-delivery, altered cellular uptake, subcellular targeting, and modulation of cell-death pathways. These strategies primarily include: first, the co-delivery of chemotherapeutic agents with P-gp inhibitors (eg, verapamil, D-α-tocopherol polyethylene glycol succinate, TPGS) to restore chemosensitivity in resistant cells by inhibiting efflux pump function;60,89 second, the design of nanoparticles capable of bypassing efflux pump recognition, such as those entering cells via ligand-mediated non-canonical endocytic pathways, thereby evading recognition and extrusion by efflux pumps;90 third, targeting subcellular organelles like mitochondria to induce irreversible damage, including mitochondrial membrane potential collapse and release of apoptotic factors, thus circumventing traditional drug efflux pathways and promoting tumor-cell death.91 Nanoplatforms that induce non-apoptotic cell-death pathways, including ferroptosis, have also been explored as strategies to bypass resistance associated with apoptosis evasion. Ferroptosis is characterized by iron-dependent lipid peroxidation and impaired antioxidant defense, including GPX4 inactivation, and may therefore provide an alternative vulnerability in some therapy-resistant tumor cells.92–95 However, the extent to which ferroptosis-based nanotherapies can overcome clinically established MDR remains largely supported by preclinical evidence. Moreover, cell membrane biomimetic systems (eg, platelet membrane-coated nanoparticles) and stimulus-responsive designs (eg, pH- or glutathione-responsive systems) have also been investigated to address TME-associated resistance.11,60 Overall, MDR in breast cancer arises from multiple interacting mechanisms, making single-mechanism interventions unlikely to provide uniform benefit. Combination delivery strategies that address efflux, cancer stem-cell phenotypes, intracellular survival pathways, and microenvironmental influences are therefore biologically plausible, but their added complexity must be justified by reproducible pharmacokinetic, safety, and clinical efficacy data.

In vivo Fate of Delivery Systems

Following intravenous administration, nanomedicines must traverse a complex series of biological barriers before reaching their intended targets in breast cancer tissues, a process often described as a cascade of “nano–bio interface” interactions.96 Specifically, nanoparticles initially encounter opsonization in the bloodstream, leading to rapid clearance by the mononuclear phagocyte system (MPS). Subsequently, the vascular endothelial barrier must be crossed via extravasation to reach the tumor interstitium. Once in the interstitial space, the dense extracellular matrix (particularly the collagen network) severely impedes their diffusion.97 Upon contacting tumor cells, nanoparticles must further traverse the cell membrane barrier via endocytosis and, finally, escape from lysosomal degradation to avoid inactivation.98,99 Each of these barriers can substantially reduce the fraction of administered nanomedicine that ultimately reaches its intended site of action. Accordingly, understanding the sequential biological barriers encountered in vivo is essential for interpreting delivery efficiency and for designing strategies that improve systemic exposure, tumor accumulation, tissue penetration, and intracellular release.

PEGylation remains a widely used strategy to circumvent MPS clearance; however, the generation of anti-PEG antibodies can lead to the “accelerated blood clearance” (ABC) phenomenon, substantially altering the pharmacokinetics and therapeutic efficacy of PEGylated nanoparticles after repeated injections. The clinical relevance of anti-PEG immune responses and accelerated blood clearance remains an important consideration, particularly for repeated administration. Emerging biomimetic approaches include CD47-derived peptides, cell membrane coatings derived from red blood cells or platelets, and substitution of PEG with zwitterionic polymers such as poly(carboxybetaine).60,100 Among these, platelet membrane-coated nanoparticles not only exhibit prolonged circulation times but also leverage the innate tumor tropism of platelets for active targeting, and have therefore been investigated as an alternative strategy for extending circulation and modifying tumor interactions.60 Collectively, these approaches aim to reduce premature clearance and improve systemic exposure, although their effects on biodistribution, immunogenicity, and manufacturing reproducibility require careful evaluation.

For solid tumors, the depth of nanoparticle penetration into tumor tissue represents another important issue. Most studies report only overall drug accumulation within the tumor, neglecting the heterogeneous distribution of drugs across different tumor regions. In fact, owing to the heterogeneous nature of tumor vasculature and elevated interstitial fluid pressure (IFP), nanoparticles tend to accumulate in perivascular regions and struggle to penetrate into hypoxic tumor cores.14 Moreover, the three-dimensional collagen network in the tumor microenvironment not only constitutes a physical barrier but also promotes tumor invasion and immune evasion through biomechanical signaling.3 To overcome this deep-penetration challenge, synergistic optimization of nanoparticle size, shape, and surface charge (eg, employing rod-shaped or flexible nanoparticles) is required, along with combination with tumor microenvironment modulators (eg, collagen-degrading enzymes or IFP-lowering agents).101 Of note, the use of IFP-lowering drugs such as losartan has been demonstrated to significantly improve the homogeneous distribution of nanomedicines within tumors.15 These findings suggest that improving intratumoral penetration may require coordinated optimization of nanoparticle physicochemical properties and, where appropriate, modulation of the tumor microenvironment.

In addition to physical barriers, biochemical and metabolic features of the tumor microenvironment can influence intracellular drug release and may be incorporated into stimulus-responsive delivery strategies. For example, acidic pH and elevated intracellular glutathione have been used to trigger pH-sensitive or reduction-responsive release.18,21,102 However, as discussed above, the heterogeneity and temporal variability of these signals can limit the reproducibility of response in vivo. Overall, the in vivo performance of nanomedicines reflects the cumulative effects of systemic clearance, vascular transport, tissue penetration, cellular uptake, and intracellular processing. Optimization should therefore consider the entire delivery pathway rather than focusing on a single barrier or responsive feature in isolation.

Future Directions

Future development of breast cancer drug delivery systems will depend not only on improving carrier performance but also on addressing patient heterogeneity, formulation reproducibility, and clinical feasibility. Several emerging directions are particularly relevant, including biomarker-guided personalized delivery, AI-assisted formulation optimization, stimulus-responsive systems, and combination delivery strategies. Although these approaches may improve the matching of delivery systems to tumor biology and therapeutic needs, their translational value will ultimately depend on prospective validation, manufacturing consistency, pharmacokinetic predictability, and demonstration of clinically meaningful benefit (Figure 4).

Figure 4.

Future Directions: Toward Intelligent and Personalized Drug Delivery. Future breast cancer DDS may integrate three strategic directions: (1) personalized precision delivery using liquid biopsy, organoids, and functional imaging to support patient stratification and treatment adaptation; (2) AI-assisted formulation optimization using experimental data and material-structure-function predictive models to support data-driven formulation design; and (3) multi-responsive logic-gated systems combining endogenous signals (pH, GSH, enzymes) with exogenous stimuli (light, magnetism, ultrasound) for spatiotemporally controllable combination therapy. Clinical translation also requires biomarker validation, prospective validation, manufacturing/CMC development, reproducibility, and clinical evaluation.

Personalized Precision Delivery

The principles of precision medicine are increasingly being incorporated into drug delivery research, with growing interest in formulations tailored to patient-specific molecular subtypes, biomarker profiles, and pharmacokinetic characteristics.103,104 This approach is motivated by the heterogeneity of breast cancer, as different patients, and even distinct tumor lesions within the same patient, may exhibit different molecular features and therapeutic responses.103 Conventional nanomedicine development has generally focused on standardized formulations, which may not fully account for interindividual differences in the tumor microenvironment, receptor expression, and resistance mechanisms.105 Accordingly, personalized delivery requires not only adaptable carrier systems but also validated biomarkers and analytical tools that can identify patients most likely to benefit from a particular formulation. The practical challenge is therefore to integrate molecular subtype, biomarker status, pharmacokinetic characteristics, and treatment history into clinically applicable patient-stratification strategies. Such stratification should also account for metastatic site-specific barriers and the maturity of supporting evidence, as summarized in Table 2.

Implementing personalized delivery will require integration of multiple analytical and experimental approaches. First, liquid biopsy-based real-time monitoring technologies can dynamically assess the evolution of tumor molecular profiles, thereby guiding adaptive adjustments of delivery systems. Analyses of circulating tumor DNA and exosomes offer feasible routes for non-invasively monitoring resistance mutations and tumor evolution.106 Tumor-derived extracellular vesicles may provide protein and nucleic acid signatures associated with tumor molecular status and could complement circulating tumor DNA for longitudinal monitoring.107 Furthermore, dynamically tracking the emergence of resistance mutations during treatment using liquid biopsies may support adjustment of therapeutic agents or delivery strategies as tumor molecular features evolve.108 Second, patient-derived xenograft models and organoid technologies provide complementary preclinical platforms for evaluating personalized nanomedicines, allowing comparison of different nanoformulations against patient-derived tumor models.14 Patient-derived tumor organoids can preserve important features of tumor heterogeneity and may provide a platform for comparing candidate formulations or combination regimens before clinical evaluation. However, incomplete representation of stromal, vascular, and immune components limits their ability to reproduce the full in vivo delivery environment. Third, advanced imaging techniques may enable pre-treatment assessment of tumor vascular permeability and EPR-related nanoparticle accumulation, providing a potential basis for identifying patients more likely to benefit from passive targeting strategies.66 Specifically, dynamic contrast-enhanced magnetic resonance imaging or near-infrared fluorescence imaging can evaluate tumor vascular characteristics and may help identify patients more likely to benefit from passive targeting strategies. Together, liquid biopsy, patient-derived models, and functional imaging may provide complementary information for patient stratification and treatment adaptation. Their clinical value, however, will require standardized assays, prospective validation, and evidence that these measurements improve the selection or modification of delivery strategies.

Theranostic platforms represent a potential approach for integrating assessment of drug distribution with treatment delivery. By incorporating diagnostic probes and therapeutic agents within the same platform, these systems may enable visualization of biodistribution, treatment response, or drug-release behavior and thereby support treatment adjustment. For example, indocyanine green-labeled nanoparticles have been used for near-infrared fluorescence-guided photothermal therapy while simultaneously providing information on nanoparticle distribution.109–111 Multifunctional nanoplatforms have also been developed to integrate magnetic resonance imaging or computed tomography with therapeutic delivery.112–114 In addition, photoacoustic imaging and positron emission tomography have been combined with therapeutic nanocarriers to monitor drug distribution and treatment response.115 However, integrating diagnostic and therapeutic functions within a single platform increases formulation complexity, characterization requirements, manufacturing burden, and regulatory challenges. Therefore, the clinical value of theranostic delivery systems will depend on whether the added imaging component provides actionable information that improves patient selection, treatment adaptation, or therapeutic outcomes. More broadly, personalized precision delivery will require validated biomarkers, standardized patient-stratification criteria, and prospective evidence that liquid biopsy, patient-derived models, and functional imaging can meaningfully guide the selection or modification of delivery strategies.

Artificial Intelligence-Assisted Formulation Optimization

Artificial intelligence (AI) and machine learning are increasingly being explored as tools to support nanomedicine formulation development. More broadly, machine-learning approaches, including graph neural networks, have also been applied to drug-interaction prediction and other drug-development tasks, illustrating the expanding role of data-driven methods across the pharmaceutical pipeline.116 By integrating information on material composition, physicochemical properties, process parameters, and biological performance, data-driven models may help identify relationships between formulation variables and delivery outcomes that are difficult to capture through conventional trial-and-error screening.117 Such approaches have the potential to prioritize candidate formulations, reduce the number of experimental iterations, and support more systematic optimization. However, their value depends strongly on the quality, diversity, and comparability of the underlying datasets, and evidence that AI-guided optimization improves downstream clinical development remains limited.118 Similar challenges have been recognized in other AI-assisted drug-development applications, where model interpretability, data quality, and regulatory acceptance remain important determinants of clinical utility.119

AI-based approaches are increasingly being applied to predictive tasks in drug delivery research. During formulation development, machine learning models can be used to estimate critical quality attributes such as nanoparticle size, drug loading capacity, and release behavior from material properties, drug characteristics, and process parameters.120 Deep learning methods have also been explored to identify physicochemical factors associated with nano–bio interface interactions from high-throughput screening data. In efficacy assessment, automated image-analysis models can facilitate quantitative evaluation of nanoparticle uptake and subcellular localization. In addition, quantitative structure-activity relationship (QSAR) models may assist in the selection of surface ligands and optimization of modification density during preclinical formulation development.121 These applications illustrate the potential of AI to support experimental prioritization and formulation optimization, although their predictive performance depends strongly on data quality, model generalizability, and validation in independent experimental settings.

Several limitations currently restrict the generalizability and translational value of AI-assisted formulation optimization. Available nanomedicine datasets are often small, heterogeneous, and generated using different experimental protocols, analytical methods, biological models, and reporting standards.122 Batch effects, selective publication of positive results, and the limited availability of negative or failed formulations can introduce substantial bias during model training. In addition, inconsistent material descriptors and incomplete reporting of formulation and process parameters may reduce reproducibility across laboratories. Models developed from narrowly defined datasets may therefore perform poorly when applied to new materials, experimental systems, or patient populations. Beyond predictive accuracy, uncertainty estimation and model interpretability are important for identifying unreliable predictions and for understanding the formulation variables that drive model outputs.117 Standardized data reporting, broader inclusion of negative results, external validation using independent datasets, and prospective experimental testing will be necessary before AI-guided formulation design can be incorporated reliably into translational development. Accordingly, AI should currently be viewed as a decision-support tool for formulation optimization rather than as a substitute for experimental and clinical validation.

Stimulus-Responsive Drug Delivery Systems

Early stimulus-responsive delivery systems primarily relied on a single stimulus (eg, pH or temperature) to trigger drug release. However, given that certain abnormalities, such as reduced pH in inflamed sites or elevated oxidative stress levels, may also exist in normal tissues, the tumor specificity of single-stimulus strategies may be limited, which can lead to premature drug leakage in non-target tissues and off-target toxicity.123 The heterogeneity of the tumor microenvironment, including spatial variation in pH, glutathione concentration, enzyme expression, and ATP levels, has motivated the development of delivery systems that respond to more than one signal. Multi-responsive designs may improve conditional control of drug release by requiring combinations of tumor-associated cues rather than relying on a single trigger. However, their potential advantage depends on whether the selected signals are sufficiently distinct, stable, and reproducible in vivo.

Logic-gated multi-responsive systems are being investigated as a means of increasing the conditional specificity of drug release. Such systems can be designed to respond only when multiple predefined microenvironmental conditions are present. For example, nanoparticles responsive to both acidic pH and elevated glutathione exploit two features of the tumor microenvironment to reduce nonspecific release under other conditions.11 Other designs incorporate sequential or temporally controlled responses, while ATP- and hyaluronidase-responsive core–shell nanoparticles illustrate how intracellular biochemical cues can be combined to regulate drug release.23,124 These studies support the feasibility of multi-signal control in experimental models, but the extent to which such designs improve therapeutic selectivity in heterogeneous human tumors remains to be established.

Beyond endogenous stimuli, the use of exogenous physical stimuli (light, magnetic fields, ultrasound, and radiation) adds another dimension of control to smart delivery. First, photothermal/photodynamic therapy not only exerts direct antitumor effects but also enhances blood perfusion via local temperature elevation, improving nanoparticle penetration into tumors; meanwhile, reactive oxygen species generated by photosensitizers can further modulate the tumor microenvironment.125 Second, ultrasound-mediated drug delivery exploits the sonoporation effect to transiently disrupt cell membranes and the blood–brain barrier, which has been investigated as an approach to facilitate drug delivery across the blood-brain barrier in brain metastases.126 Combining exogenous stimuli with endogenous response mechanisms may provide additional spatial and temporal control. However, the clinical applicability of these systems depends on factors such as tissue penetration of the external stimulus, treatment accessibility, device requirements, and reproducibility of the biological response. Moreover, increasing the number of responsive components can complicate formulation, characterization, pharmacokinetics, manufacturing, and regulatory evaluation. Therefore, multi-responsive systems should be judged by whether each additional responsive element provides a measurable improvement in selectivity, safety, or therapeutic efficacy rather than by design complexity alone.

Combination Therapy Strategies

Combination delivery strategies aim to coordinate therapeutic agents with complementary mechanisms of action while controlling their relative exposure in tumors. Nanocarriers can co-encapsulate or co-deliver multiple agents and may help reduce differences in pharmacokinetic behavior between individual drugs.127 However, therapeutic synergy depends on dose ratio, release kinetics, tumor distribution, and treatment sequence, and co-delivery does not necessarily guarantee improved efficacy. These variables therefore need to be optimized and validated for each combination rather than assuming an inherent advantage over separately administered agents.

Chemo-immunotherapy combinations have been widely investigated as a strategy to integrate direct tumor-cell killing with immune activation. Chemotherapy-induced immunogenic cell death (ICD) can promote the release of damage-associated molecular patterns (DAMPs), thereby facilitating antigen-presenting cell activation and subsequent T-cell responses.32 Agents such as doxorubicin and oxaliplatin have been reported to induce ICD and promote dendritic cell maturation and T-cell activation.128 Nanocarriers have therefore been explored for the co-delivery of chemotherapeutic agents and immunomodulators, including anti-PD-1 antibodies and STING agonists. In this context, the cGAS-STING pathway provides a mechanistic link between chemotherapy-induced DNA damage and antitumor immune activation, and co-delivery of STING agonists with chemotherapeutic agents may enhance immunogenic responses in preclinical models.28 This approach may be particularly relevant to triple-negative breast cancer, although the extent and therapeutic significance of STING pathway suppression vary across tumors. Ferroptosis-inducing agents have also been combined with immunomodulators to explore whether ferroptotic stress can further promote antitumor immune responses.129 For example, FBR-NDs (Fe³⁺-BMS-1-R848 nanodrugs) were reported to induce ferroptosis while promoting DAMP release, dendritic-cell and M1-macrophage infiltration, and T-cell activation.95 Cell-based carrier systems, including freeze-thaw-treated natural killer cells, have similarly been investigated for combined drug delivery and immune modulation.32 Collectively, these approaches illustrate the potential of coordinated chemo-immunotherapy delivery, but most evidence remains preclinical, and their clinical value will depend on reproducible immune effects, manageable toxicity, and appropriate patient selection.

Chemo-targeted therapy combinations have been explored as a strategy to address adaptive resistance to chemotherapy. Tumor cells can activate compensatory signaling pathways under therapeutic stress, and co-delivery of chemotherapeutic agents with pathway inhibitors may help suppress these adaptive responses. For example, paclitaxel and the mTOR inhibitor rapamycin have been co-delivered to simultaneously target tumor-cell proliferation and PI3K/AKT/mTOR signaling.130 Ratio-controlled formulations have also been developed to maintain defined drug proportions during delivery, because the relative exposure of individual agents can influence pharmacological synergy. Rapaxane, a polymeric micelle co-loading paclitaxel and a rapamycin prodrug, was reported to inhibit tumor growth and lung metastasis in preclinical triple-negative breast cancer models while maintaining a 5:1 drug ratio.130 Prodrug design and site-specific conjugation may provide additional approaches for controlling drug stoichiometry and release. Similar strategies have been investigated for multidrug resistance, including co-delivery of chemotherapeutic agents with P-glycoprotein inhibitors such as TPGS to reduce drug efflux.60 These studies suggest that control of drug ratio and release kinetics can influence combination performance, but clinical translation will require evidence that such formulation control provides reproducible pharmacokinetic, safety, or efficacy advantages over conventional combination dosing.

Chemo-photothermal and chemo-photodynamic combinations have been investigated as approaches that combine pharmacological treatment with externally controlled activation. Photothermal and photodynamic modalities offer local control through light-triggered energy conversion or reactive oxygen species generation, although their applicability is influenced by tissue penetration and treatment accessibility. For example, the ODPDC injectable hydrogel co-loaded doxorubicin- and paclitaxel-containing nanomedicines and enabled microenvironment-responsive drug release under acidic and glutathione-rich conditions; in combination with photothermal therapy, this formulation showed enhanced antitumor activity in experimental models.22 Similarly, the CuS@mSiO₂-Pc(DOX)@HA nanoplatform targets CD44-overexpressing tumor cells through hyaluronic acid and combines chemotherapy with photothermal and photodynamic effects under near-infrared irradiation.60 These multimodal systems illustrate the feasibility of integrating external physical stimuli with drug delivery, but their clinical translation will depend on factors such as light penetration, treatment accessibility, device standardization, formulation complexity, and reproducibility.

Combination delivery introduces additional design variables beyond those encountered with single-agent formulations, including drug ratio, release sequence, exposure duration, and delivery order. Differences in these parameters can substantially influence whether pharmacological synergy is achieved. Mathematical modeling and systems-level approaches may assist in exploring this multidimensional design space, although their predictions require experimental validation. In addition, the performance of combination regimens is likely to depend on tumor heterogeneity, resistance mechanisms, and microenvironmental characteristics. Future development should therefore prioritize combinations supported by a clear mechanistic rationale and demonstrate that coordinated delivery provides reproducible pharmacokinetic, safety, or efficacy advantages over conventional administration. Increasing the number of therapeutic components without such evidence may instead increase formulation complexity, manufacturing burden, and regulatory challenges.

Conclusion and Perspectives

Drug delivery systems provide a diverse set of approaches for modifying the pharmacokinetics, biodistribution, local retention, and intracellular release of anticancer agents in breast cancer. Liposomes, polymeric nanoparticles, hydrogels, extracellular vesicles, antibody–drug conjugates, and other delivery modalities differ substantially in technological complexity, mechanism of action, manufacturing requirements, and level of clinical maturity. Although many platforms have shown improved delivery or therapeutic activity in experimental models, only a limited proportion have progressed to established clinical use. This translational gap reflects not only biological barriers and tumor heterogeneity but also challenges related to formulation reproducibility, pharmacokinetic variability, toxicity, scale-up, and regulatory evaluation.

Drug delivery systems provide a diverse set of approaches for modifying the pharmacokinetics, biodistribution, local retention, and intracellular release of anticancer agents in breast cancer. Liposomes, polymeric nanoparticles, hydrogels, extracellular vesicles, antibody-drug conjugates, and other delivery modalities differ substantially in technological complexity, mechanism of action, manufacturing requirements, and level of clinical maturity. Outside established approaches such as liposomal doxorubicin and clinically approved ADCs, many smart, stimulus-responsive, and biologically derived delivery systems discussed in this review remain predominantly preclinical. Although these platforms may show improved delivery or therapeutic activity in experimental models, their clinical adoption is limited by interpatient and tumor heterogeneity, variable pharmacokinetics and biodistribution, formulation complexity and reproducibility, safety and immunogenicity concerns, scale-up and manufacturing constraints, and regulatory requirements. In addition, promising performance in small-animal models may not reliably predict behavior in patients because human tumors exhibit greater heterogeneity in vascular permeability, stromal architecture, immune composition, and metastatic-site-specific barriers. For increasingly complex multifunctional systems, the translational burden is further increased by the need to demonstrate that each added component provides a reproducible and clinically meaningful advantage over simpler formulations. These factors collectively contribute to the persistent gap between promising experimental performance and routine clinical application.

Future development should therefore focus on matching delivery strategies to clinically relevant biological contexts rather than increasing formulation complexity alone. Tumor microenvironmental features, including stromal density, vascular permeability, immune-cell composition, pH, redox state, and hypoxia, may influence both delivery efficiency and the performance of responsive systems. Strategies that modulate the extracellular matrix or exploit microenvironmental signals may improve delivery in selected settings, but their biological consequences and interpatient variability require careful evaluation. Cell-based carriers and other biologically derived delivery systems also remain of interest, although issues related to cell-source variability, biodistribution, manufacturing consistency, safety, and scalability currently limit their translational readiness.

Clinical translation will require greater attention to chemistry, manufacturing, and controls (CMC), together with regulatory requirements. Critical quality attributes such as particle size, surface properties, drug loading, release kinetics, stability, sterility, and batch-to-batch consistency need to be defined and controlled throughout scale-up. Biodistribution, immunogenicity, long-term toxicity, and interactions with companion diagnostics or patient-selection biomarkers should also be evaluated using standardized and clinically relevant models. For complex or multifunctional systems, the added benefit of each component should be demonstrated against the additional burden it introduces in characterization, manufacturing, and regulatory assessment.

Progress in the field will therefore depend on integrating mechanistic understanding with evidence maturity and translational feasibility. Biomarker-guided patient stratification, prospective validation of predictive tools, standardized manufacturing, and appropriately designed clinical studies will be essential for determining which delivery systems provide meaningful advantages over existing treatments. A clinically useful breast cancer delivery platform will ultimately be defined not by technological sophistication alone, but by reproducible performance, acceptable safety, manufacturability, and demonstrated benefit in the patient population for which it is intended.

Funding Statement

This work was funded by the National Natural Science Foundation of China (Nos. 82401633), the Natural Science Research Projects in Universities of Jiangsu Province (No. 24KJA310007, No. 24KJB310013), and Nantong Science and Technology Program (No. 20241118084).

Disclosure

The authors report no competing interests.