Advances in drug delivery systems for osteoarthritis therapy: exosomes, microparticles, nanoparticles, and hydrogels Authors: Yu-Xin Chen, Fei Zhang, Ya Ding, Ming-Hui Zheng, Zhi-Ang Zhou, Lei Qiu, Yi-Xin Yu, Ying-Ying Gu, Qi-Rong Deng, Kun Guo, Feng Xu, Xiao-Bo Liao, Ling-Qing Yuan, Le-Le Liao Published: 2026 ABSTRACT Osteoarthritis is a debilitating chronic joint disorder characterized by cartilage degradation, inflammation, and pain. It significantly impacts the quality of life, particularly in elderly populations. Traditional treatment methods, such as local and systemic drug administration, are limited by poor drug retention, low therapeutic efficacy, and systemic side effects. Recent advancements in drug delivery systems have emerged as promising strategies to overcome these limitations, offering targeted and sustained therapeutic effects, including promoting chondrocyte proliferation and migration, reducing extracellular matrix degradation, and so on. This review highlights the progress in various innovative drug delivery systems, including exosomes, microparticles, nanoparticles, and hydrogels, which have shown promising results in preclinical studies. Exosomes, particularly those derived from mesenchymal stem cells, demonstrate significant potential in promoting cartilage repair, modulating inflammation, and facilitating cell communication. Microparticles composed of natural or synthetic polymers prolong intra-articular drug retention and provide controlled release. Stimuli-responsive nanoparticles, which react to pathological cues like reactive oxygen species, pH, and enzymes, allow precise drug release within the OA microenvironment. Hydrogels offer versatile platforms for sustained delivery and tissue scaffolding, with responsive systems adapting to joint conditions. Despite their promising preclinical results, challenges remain in standardization, manufacturing scalability, and long-term safety. The continued development and clinical translation of these advanced systems hold great potential to revolutionize osteoarthritis therapy by improving efficacy, reducing side effects, and promoting joint tissue regeneration.Graphical Abstract Introduction Osteoarthritis (OA), a chronic degenerative disease characterized by degeneration, inflammation, and pain of articular cartilage, is one of the joint diseases with the highest disability rate worldwide, and has a serious impact on the quality of life of the elderly population in particular [1]. Historically regarded as a mechanical consequence of articular cartilage degeneration, OA has now been recognized as a progressive disorder affecting the entire joint. Contemporary research posits that its pathogenesis arises from biochemical imbalances and cellular dysregulation within synovial tissues, initiating cascading pathological cascades. These alterations manifest as distinct histological and architectural disruptions: erosion of hyaline cartilage integrity, compromised subchondral bone architecture with osteosclerotic remodeling, synovial membrane hyperplasia accompanied by neoangiogenesis, and progressive laxity of periarticular connective tissues. Ultimately, this multisystem deterioration converges into global joint dysfunction, redefining OA as a complex interplay of molecular, cellular, and biomechanical failures rather than a singular cartilage-centric disease [2]. Current therapeutic approaches for osteoarthritis primarily utilize two drug delivery pathways: localized topical application and systemic distribution through oral intake. In cases of knee OA, clinicians frequently employ direct joint injections as a therapeutic strategy, which offers dual advantages over whole-body drug circulation. Firstly, this targeted method elevates pharmaceutical concentration specifically at the lesion site while simultaneously minimizing exposure to peripheral tissues, thereby decreasing the likelihood of organism-wide adverse reactions. Nevertheless, the clinical efficacy of conventional small-molecule formulations administered through intra-articular injection remains suboptimal, primarily due to their rapid clearance from the injection site and insufficient sustained release characteristics within synovial cavities [3]. In addition to intra-articular delivery, transdermal drug delivery systems have attracted increasing attention as non-invasive strategies for treating arthritis-related joint disorders [4]. Despite its significant potential, the skin barrier’s limiting effect on drug delivery remains a pressing issue that needs to be addressed. In modern therapeutic development, advanced drug delivery systems have emerged as pivotal components owing to their versatility in encapsulating diverse pharmaceutical agents, adjustable release kinetics, and site-specific delivery capabilities. These engineered platforms demonstrate therapeutic enhancement through spatial localization and temporal extension of pharmacologically active compounds within biological systems, thereby optimizing treatment effectiveness while minimizing systemic exposure [5]. Nowadays, the development of drug delivery systems has become a popular direction in the research of osteoarthritis treatment, and a variety of different materials, sizes, and functions of drug delivery systems are constantly being developed. Although several recent reviews have summarized biomaterial-based or nanotechnology-enabled drug delivery strategies for osteoarthritis, the present review provides a distinct and more integrative perspective by systematically comparing multiple delivery platforms—exosomes, microparticles, nanoparticles, and hydrogels—within a unified therapeutic framework. In particular, we highlight the source-, cargo-, and engineering-dependent functions of exosome-based systems, distinguish natural and synthetic polymer microparticles according to their biological and release characteristics, and discuss stimuli-responsive nanoplatforms and hydrogel systems that exploit OA-specific cues such as reactive oxygen species, pH changes, and enzyme activity. Exosomes Exosomes are nanoscale extracellular lipid bilayer vesicles of endocytic origin, 40–160 nm in diameter, secreted by virtually all cell types in physiological and pathological conditions [6]. The biogenesis of exosomes initiates through the invagination of the plasma membrane, leading to the creation of early endosomes. These structures subsequently mature into multivesicular bodies (MVBs), during which the endosomal membrane undergoes inward budding to generate intraluminal vesicles (ILVs). These ILVs encapsulate parent cell-derived biomolecules, including lipids, proteins, and nucleic acids [7]. Two distinct pathways govern ILV biogenesis: The first relies on the endosomal sorting complex required for transport (ESCRT) machinery, which collaborates with auxiliary factors ALIX and TSG101 to cluster ubiquitinated MVB membrane components and facilitate membrane curvature for vesicle formation. Alternatively, ceramide-mediated assembly of lipid raft microdomains triggers spontaneous membrane budding independent of ESCRT. Exosome release occurs via two secretory modes: constitutive secretion through the trans-Golgi network or regulated secretion via MVB-plasma membrane fusion. Spatiotemporal coordination of these processes involves Rab27a/b proteins for MVB docking and SNARE proteins for mediating membrane fusion events [8]. Early on, exosomes were considered to be nonfunctional cellular metabolic waste products, but it has since been recognized to carry proteins, lipids, and nucleic acids, including mRNAs, microRNAs, and long non-coding RNAs, which play a mediating critical role in cell-to-cell communication and cellular immune responses [9]. The unique characteristics of exosomes, particularly their bilayer membrane architecture and lower immunogenicity, enable effective shielding of encapsulated components against immune-mediated elimination mechanisms, ultimately enhancing their circulating half-life and improving their biological activity [10]. Besides, due to their nanoscale dimensions, they effortlessly penetrate organs and navigate through capillaries to specific body locations [11]. Ultracentrifugation, magnetic activated cell sorting, and chemical precipitation are among the most widely used techniques for exosome isolation. Nowadays, to improve efficiency and specificity when processing complex biological fluids, novel approaches—such as microfluidic filtering—have emerged as advanced methods for exosome enrichment [7]. Then, the isolated exosomes are identified mainly by morphological features, particle size, and characteristic proteins such as CD9, CD63, CD81, and TSG101 [12]. For the cargo-loading strategies, the main methods of transmembrane cargo loading are electroporation, sonication, direct transfection, and cell engineering [7]. Clinical and translational studies of exosome or extracellular vesicle therapy for osteoarthritis have shown promising safety and biological plausibility but limited and still inconclusive therapeutic efficacy. Early human studies using mesenchymal stromal/stem cell-derived extracellular vesicles, particularly from placenta or umbilical cord sources, suggest that intra-articular injection is generally well tolerated, with few or no serious adverse events reported [13]. Some small open-label or dose-escalation studies observed improvements in pain scores, functional outcomes, and MRI-based joint assessments after treatment [14]. However, the strongest available randomized, triple-blind, placebo-controlled trial found that placental MSC-derived extracellular vesicles were not superior to saline placebo for pain, function, or MRI outcomes over short-term follow-up [15]. The major limitations of current clinical researches include small sample sizes, short follow-up periods, heterogeneous vesicle sources and manufacturing protocols, inconsistent dosing and characterization standards, lack of large multicenter randomized controlled trials, and uncertainty about optimal patient selection, mechanism of action, long-term safety, and structural disease modification. In summary, exosome-based treatment for osteoarthritis remains an investigational therapy rather than a mature clinical treatment method; therefore, the content of this article focuses primarily on preclinical researches. In recent years, with intensive research, exosomes from various sources have shown excellent potential to be used as drug delivery systems for OA therapy (Fig. 1; Table 1). Fig. 1Roles of cargo carried by exosomes from different cellular sources in OA therapy Roles of cargo carried by exosomes from different cellular sources in OA therapy MSCs-derived exosomes BM-MSCs-derived exosomes Bone marrow mesenchymal stem cells (BM-MSCs) are adult stem cells derived from bone marrow, possessing key advantages—convenient sampling, potent proliferation, low immunogenicity, ease of genetic manipulation, and strong differentiation potential—which position them as the optimal stem cell type in current research and clinical use [11]. Shen et al. investigated that BM-MSCs-derived exosomes loading lncRNA TUC339 improved OA by promoting the polarization of M1-type macrophages to M2-type, inhibiting inflammatory responses, and promoting chondrocyte activity [16]. Lou et al. demonstrated that exosomes derived from fucoidan-pretreated bone marrow mesenchymal stem cells carrying miR−146b−5p significantly enhanced protection against OA compared to untreated MSC-derived exosomes. Mechanistically, miR−146b−5p silences TRAF6, thereby inhibiting the PI3K/AKT/mTOR pathway, resulting in reduction of inflammation and extracellular matrix (ECM) degradation, and promotion of autophagy in chondrocytes [17]. Similarly, Shao et al. found that exosomes derived from parathyroid hormone (PTH)-preconditioned bone marrow mesenchymal stem cells were more effective than those from untreated cells at delaying osteoarthritis progression. The enhanced therapeutic effect was primarily mediated by the microRNA let−7a−5p, which was enriched in PTH-preconditioned exosomes, promoting chondrocyte proliferation and migration, reducing extracellular matrix degradation. And Let−7a−5p suppressed IL−6expression, thereby reducing STAT3 phosphorylation and inhibiting the IL−6/STAT3 inflammatory pathway [18]. AD-MSCs-derived exosomes Adipose-derived mesenchymal stem cells (AD-MSCs) are another type of stem cells commonly used to extract exosomes, with the advantages of abundant source tissues and a simple extraction process [19]. Using rat OA models and in vitro assays with IL−1β-induced chondrocytes and TGF-β1/LPS-stimulated synovial fibroblasts, Li et al. demonstrated that exosomes derived from human AD-MSCs reduced cartilage degradation, synovial fibrosis, and inflammation. Mechanistically, miR−376c−3p within the exosomes directly inhibited WNT3 and WNT9a, key components of the WNT-beta-catenin pathway, thereby suppressing β-catenin activation. In vivo experiments confirmed that intra-articular injection of exosomes derived from human AD-MSCs mitigated OA progression, while blocking miR−376c−3p reversed these protective effects [20]. Wang et al. found that delivery of miR−486−5p via AD-MSCs-derived exosomes alleviated endoplasmic reticulum stress-induced chondrocyte apoptosis, promoted cartilage matrix regeneration, inhibited inflammatory cytokines, and modulated macrophage polarization to an anti-inflammatory state [21]. In addition, new research has shown that the infrapatellar fat pad located in the knee joint is an important reservoir of mesenchymal stem cells (MSCs) [22]. Recent studies demonstrate that exosomes derived from infrapatellar fat pad mesenchymal stem cells (IPFP-MSCs) exerted chondroprotective effects in osteoarthritis by delivering miR−100−5p to suppress mTOR signaling and activate autophagy flux, thereby restoring cartilage homeostasis and ameliorating gait abnormalities in murine models. Exosomal miR−100−5p directly targeted the 3′UTR of mTOR mRNA, suppressing mTOR signaling and thereby enhancing protective autophagy in IL−1β-stimulated chondrocytes. Through this miR−100−5p/mTOR/autophagy axis, the exosomes reduced chondrocyte apoptosis and catabolic mediators, restored anabolic matrix markers, and ultimately protected articular cartilage and alleviate OA progression [23]. Notably, their team pioneered a TNF-α preconditioning strategy to enhance the potency of IPFP-MSCs exosomes, with TNF-α-primed exosomes exhibiting significantly superior therapeutic efficacy compared to non-modified counterparts. TNF-α pretreatment increased exosome production via activation of the PI3K/AKT–ATG16L1 pathway in IPFP-MSCs, and the resulting exosomes deliver higher levels of protein LRP1 to chondrocytes. Exosomal LRP1 promoted endocytic clearance/degradation of ECM-degrading enzymes, thereby reducing cartilage matrix breakdown, preserving aggrecan, improving synovial inflammation, and exerting stronger chondroprotective effects in osteoarthritis [24]. Furthermore, Yin et al. had also indicated that IPFP-MSCs-derived exosomes exerted a unique efficacy in the treatment of OA compared to other adipose tissue-derived stem cells-derived exosomes, which is associated with the delivery of miR−99b−3. Exosomal miR−99b−3p directly targeted a disintegrin and metalloproteinase with thrombospondin type 1 motifs (ADAMTS) 4, a key ECM-degrading enzyme, thereby suppressing ADAMTS4 expression, reducing cartilage matrix degradation, and restoring anabolic cartilage markers [25]. SMSCs-derived exosomes Besides, synovial mesenchymal stem cells (SMSCs) also drew attention from researchers due to their greater accessibility, higher proliferation rate, lower immunogenicity, and superior chondrogenic differentiation potential [26]. Long et al. found that SMSCs-derived exosomes exerted beneficial therapeutic effects on OA through the delivery of MATN3/IL−17 A, which prevented both ECM degradation and autophagy defects. Exosomal MATN3 interacted with IL−17 A and suppressed IL−17 A-driven activation of the PI3K/AKT/mTOR pathway, thereby reducing downstream inflammatory cytokines such as IL−6 and TNF-α. This inhibition also restored autophagy activity, decreased ECM-degrading enzymes including matrix metalloproteinase (MMP)−13 and ADAMTS5, increased cartilage matrix markers such as aggrecan and collagen II, and ultimately alleviated cartilage degradation and OA progression [27]. Qiu et al. revealed that human SMSCs-derived exosomes loaded with miR−129−5p attenuated chondrocyte dysregulation, inflammation, apoptosis, and extracellular matrix degradation by directly targeting HMGB1 to suppress the HMGB1/TLR4/NF-κB inflammatory pathway [28]. UC-MSCs-derived exosomes Umbilical cord mesenchymal stem cells (UC-MSCs) are noted for their potent regenerative capabilities, and the collection of UC-MSCs is a noninvasive process. Therefore, they have gained favor among researchers [29]. Chen et al. developed a therapeutic strategy for osteoarthritis by priming human UC-MSCs with IL−1β to produce exosomes enriched with anti-inflammatory and cartilage-regenerative microRNAs such as miR-205-5p, miR-766-3p, etc., which were predicted to modulate pathways including NF-κB, MAPK, and PI3K-Akt. Functionally, those exosomes suppressed pro-inflammatory mediators such as IL-1β, IL-6, TNF-α, and MCP-1, inhibited M1 macrophage polarization, and promoted M2 polarization with increased IL-10 and TGF-β. At the same time, they enhanced cartilage matrix regeneration by increasing collagen II, SOX9, aggrecan, and glycosaminoglycan production while reducing MMP13, thereby remodeling the pro-inflammatory OA microenvironment and promoting cartilage repair [30]. In another study, human UC-MSC-derived exosomes mainly acted by delivering miR−199a−3p into IL−1β-stimulated chondrocytes. The delivered miR−199a−3p targeted and suppressed MAPK4, which reduced activation of the NF-κB pathway, as shown by decreased p-NF-κB p65. As a result, the inflammatory response was weakened, with lower production of IL−6, IL−8, and TNF-α, while chondrocyte apoptosis was also reduced. Conversely, activation of NF-κB with PMA or overexpression of MAPK4 partially reversed these protective effects, supporting the mechanism that exosomal miR−199a−3p alleviated osteoarthritis inflammation through the MAPK4/NF-κB axis [31]. Research on UC-MSC-derived exosomes has now entered the clinical trial phase, a study by Wang et al. suggested that exosomes derived from human UC-MSCs may improve symptoms of knee OA and certain MRI parameters; however, further validation is needed through large-scale, placebo-controlled, multicenter randomized clinical trials [14]. Non-MSCs-derived exosomes Macrophage-derived exosomes Exosomes produced by other non-stem cells were similarly employed in the experiments. Qian et al. found that miR−26b−5p in M2 macrophage-derived exosomes protected articular cartilage and ameliorated gait abnormalities in OA mice. After uptake by macrophages, miR−26b−5p directly targeted TLR3, thereby suppressing the downstream TLR3–TRIF/TRAF6–TAK1–IKKα–IκBα–NF-κB p65 inflammatory signaling cascade. This inhibition reduced M1-associated inflammatory responses, including IL−1β, IL−6, TNF-α, PTGS2, and iNOS, and promoted repolarization of macrophages from the pro-inflammatory M1 phenotype toward the anti-inflammatory M2 phenotype. In parallel, miR−26b−5p also targeted COL10A1 in chondrocytes, limiting chondrocyte hypertrophy and cartilage degeneration [32]. Bai et al. also found that M2-polarized macrophages released exosomes containing Sox9 mRNA and protein. These exosomes were transferred to primary chondrocytes, enhancing chondrogenic markers and extracellular matrix production while suppressing dedifferentiation markers, mitigating osteoarthritis damage [33]. Engineering exosomes By implementing targeted membrane engineering and cargo-loading to these naturally occurring nanoscale extracellular vesicles enclosed by lipid bilayers, the therapeutic potential of exosomes could be significantly advanced [10]. At present, lysosome-associated membrane protein 2B (LAMP−2B) is the most widely used exosomal surface protein. Scientific evidence indicates this transmembrane protein exhibits particularly high expression levels in exosomes derived from dendritic cells. Structural analyses have revealed that the extracellular N-terminal domain of LAMP−2B remains exposed on exosomal surfaces, providing a viable platform for functional peptide fusion. Through phage display technology, investigators can identify tissue-specific binding peptides that demonstrate selective affinity for particular cell types. Subsequent genetic engineering enables the precise incorporation of these bioactive sequences into the N-terminal region of LAMP−2B, thereby conferring targeted delivery capabilities to engineered exosomes [34]. Xu et al. transfected dendritic cells with plasmid E7-EGFP-lamp2b to produce E7-labelled exosomes and transplanted synovial fluid MSCs capable of forming cartilage in situ by E7-Exo-delivered kartogenin (KGN) [35]. Similarly, Liang et al. obtained chondrocyte-targeting exosomes from dendritic cells stably transfected with the CAP-GFP-Lamp2b plasmid and achieved excellent chondrocyte-specific targeting and cartilage penetration by altering the surface of the exosomes, thereby delivering miR−140 and rejuvenating OA cartilage. Once inside chondrocytes, exosomal miR-140 suppressed IL-1β-induced catabolic responses by downregulating matrix-degrading enzymes, especially MMP-13 and ADAMTS5, through direct targeting of their 3′UTRs [36]. Besides, they also delivered chondrocyte-targeting exosomes carrying plasmid Cas9 sgMMP−13 into the deep region of the cartilage matrix of intra-articularly administered arthritic rats, delivered plasmid Cas9 sgMMP−13 to chondrocytes, knocked down MMP−13, effectively ablated MMP−13 expression in chondrocytes, and attenuated the cartilage Hydrolytic degradation of ECM proteins [37]. Exosomes from other cellular sources can be similarly engineered for modification. Subcutaneous fat-derived MSC exosomes naturally delivered miR−199a−3p to chondrocytes, inhibiting mTOR and activating autophagy to promote cartilage repair in osteoarthritis. Zhao et al. Found that engineering these exosomes with a cartilage affinity peptide enhanced miR−199a−3p delivery depth and specificity, significantly improving cartilage protection and regeneration in OA animal models compared to unmodified exosomes [38]. Likewise, Zhang et al. developed chondrocyte-targeted exosomes by functionalizing exosome surfaces with a CAP via lipid insertion. These engineered exosomes were loaded with siRNA against MMP−13 using electroporation. Intra-articular delivery of these engineered exosomes in a rat OA model demonstrated enhanced retention in joints and deep cartilage penetration, leading to efficient MMP−13 knockdown and upregulation of cartilage-repair markers. The treatment reduced inflammation, promoted chondrocyte proliferation, and reversed OA-related proteomic changes, significantly ameliorating cartilage damage with no major toxicity [39]. Exosome-based delivery systems have emerged as promising biologically active platforms for OA therapy. Natural exosomes derived from MSCs, macrophages, dendritic cells, and other cellular sources can transfer endogenous regulatory nucleic acids and proteins to chondrocytes, synoviocytes, macrophages, and MSCs, thereby modulating multiple OA-related processes. Mechanistically, these vesicles suppress inflammatory and catabolic pathways, including NF-κB, MAPK, IL-6/STAT3, WNT/β-catenin, and PI3K/AKT/mTOR signaling, resulting in reduced inflammatory cytokine production, decreased MMP- and ADAMTS-mediated ECM degradation, inhibited chondrocyte apoptosis and senescence, enhanced autophagy, and increased macrophage M2 polarization. Compared with natural exosomes, engineered exosomes provide greater therapeutic controllability by incorporating targeting peptides, surface modifications, or defined therapeutic cargoes, which improves cartilage targeting, joint retention, penetration depth, and cargo-delivery specificity. Thus, natural exosomes are attractive because of their intrinsic bioactivity, biocompatibility, low immunogenicity, and multi-target effects, whereas engineered exosomes are more suitable for precise and customizable delivery. Nevertheless, natural exosomes are limited by donor variability, heterogeneous cargo composition, and insufficient reproducibility, while engineered exosomes face challenges related to complex fabrication, loading efficiency, potential alteration of biological properties, and regulatory uncertainty. Future studies should therefore integrate the biological advantages of natural exosomes with the targeting and controllability of engineered platforms, while emphasizing standardized manufacturing, mechanistic validation, potency assays, and long-term safety assessment in clinically relevant OA models. Table 1Exosomal sources and their loaded cargoes for the treatment of OAExosomal sourceCargoAnimal modelDosageOutcomeRef.Bone marrow mesenchymal stem cells (BM-MSCs)lncRNA TUC339OA model in rats0.5 mg/mL IAPromoting M1-to-M2 macrophage polarization, inhibiting inflammation, and enhancing chondrocyte activity[16]BM-MSCsmiR−146b−5pOA model in rats10 µL IAInhibiting inflammation, reducing extracellular matrix degradation, and promoting autophagy in chondrocytes[17]BM-MSCsmiR-let−7a−5pOA model in ratslet−7a−5p mimic 20 nM IAPromoting chondrocyte proliferation and migration, reducing extracellular matrix degradation, and alleviating OA pathology by inhibiting the IL−6/STAT3 signaling pathway[18]Adipose-derived mesenchymal stem cells (AD-MSCs)miR−376c−3pOA model in rats100 µg IAReducing cartilage degradation, synovial fibrosis, and inflammation[20]AD-MSCsmiR−486−5pOA model in rats20 µL, 10 µg/mL IAAlleviating chondrocyte apoptosis, promoting cartilage matrix regeneration, and inhibiting inflammatory cytokines[21]Infrapatellar fat pad mesenchymal stem cells (IPFP-MSCs)miR−100−5pOA model in mice10 µL, 1 × 1010 particles/mL IAMaintaining cartilage homeostasis via the mTOR-autophagy pathway, protecting articular cartilage, and improving gait abnormalities[23]IPFP-MSCsmiR−99b−3OA model in mice30 µL IAInhibiting the expression of ADAMTS4, promoting the repair of the ECM in OA[25]Synovial mesenchymal stem cells (SMSCs)MATN3/IL−17 AOA model in mice10 µL IAPreventing extracellular matrix degradation and autophagy defects, exerting therapeutic effects[27]SMSCsmiR−129−5p--Attenuating chondrocyte dysregulation, inflammation, apoptosis, and extracellular matrix degradation[28]Umbilical cord mesenchymal stem cells (UC-MSCs)miR−205−5pmiR−766−3pOA model in rats-Enhancing chondrocyte proliferation, migration, and extracellular matrix synthesis, promoting the polarization of M1 macrophages toward M2 phenotypes[30]UC-MSCsmiR−199a−3pOA model in rats40 µg IATargeting and suppressing MAPK4, reducing activation of the NF-κB pathway[31]M2 macrophagesmiR−26b−5p--Protecting articular cartilage and improving gait abnormalities by targeting TLR3 and COL10A1, influencing macrophage polarization and chondrocyte hypertrophy[32]M2 macrophagesSox9 mRNAOA model in mice-Enhancing chondrogenic markers and extracellular matrix production while suppressing dedifferentiation markers, mitigating osteoarthritis damage[33]Dendritic cellsKartogenin (KGN)OA model in rats100 µL IAPromoting in situ cartilage formation by synovial fluid MSCs via KGN delivery[35]Dendritic cellsmiR−140OA model in rats100 µg IAAchieving chondrocyte-specific targeting and cartilage penetration, delivering miR−140 to rejuvenate OA cartilage[36]Dendritic cellsCas9 sgMMP−13OA model in rats-Delivering Cas9 sgMMP−13 to chondrocytes, knocking down MMP−13 expression, and reducing the hydrolytic degradation of extracellular matrix proteins[37]Subcutaneous adipose mesenchymal stromal cellsmiR−199a−3pOA model in ratsRats: 50 µL,2 × 1010 particles/mL IAmice: 10 µL,1 × 1010 particles/mL IAInhibiting mTOR and activating autophagy to promote cartilage repair in osteoarthritis.[38]Expi293F cellsSiMMP−13OA model in rats100 µL, 1 mg/mL IAReducing inflammation, promoting chondrocyte proliferation, and reversing OA-related proteomic changes, significantly ameliorating cartilage damage with no major toxicity[39] Table 1Exosomal sources and their loaded cargoes for the treatment of OAExosomal sourceCargoAnimal modelDosageOutcomeRef.Bone marrow mesenchymal stem cells (BM-MSCs)lncRNA TUC339OA model in rats0.5 mg/mL IAPromoting M1-to-M2 macrophage polarization, inhibiting inflammation, and enhancing chondrocyte activity[16]BM-MSCsmiR−146b−5pOA model in rats10 µL IAInhibiting inflammation, reducing extracellular matrix degradation, and promoting autophagy in chondrocytes[17]BM-MSCsmiR-let−7a−5pOA model in ratslet−7a−5p mimic 20 nM IAPromoting chondrocyte proliferation and migration, reducing extracellular matrix degradation, and alleviating OA pathology by inhibiting the IL−6/STAT3 signaling pathway[18]Adipose-derived mesenchymal stem cells (AD-MSCs)miR−376c−3pOA model in rats100 µg IAReducing cartilage degradation, synovial fibrosis, and inflammation[20]AD-MSCsmiR−486−5pOA model in rats20 µL, 10 µg/mL IAAlleviating chondrocyte apoptosis, promoting cartilage matrix regeneration, and inhibiting inflammatory cytokines[21]Infrapatellar fat pad mesenchymal stem cells (IPFP-MSCs)miR−100−5pOA model in mice10 µL, 1 × 1010 particles/mL IAMaintaining cartilage homeostasis via the mTOR-autophagy pathway, protecting articular cartilage, and improving gait abnormalities[23]IPFP-MSCsmiR−99b−3OA model in mice30 µL IAInhibiting the expression of ADAMTS4, promoting the repair of the ECM in OA[25]Synovial mesenchymal stem cells (SMSCs)MATN3/IL−17 AOA model in mice10 µL IAPreventing extracellular matrix degradation and autophagy defects, exerting therapeutic effects[27]SMSCsmiR−129−5p--Attenuating chondrocyte dysregulation, inflammation, apoptosis, and extracellular matrix degradation[28]Umbilical cord mesenchymal stem cells (UC-MSCs)miR−205−5pmiR−766−3pOA model in rats-Enhancing chondrocyte proliferation, migration, and extracellular matrix synthesis, promoting the polarization of M1 macrophages toward M2 phenotypes[30]UC-MSCsmiR−199a−3pOA model in rats40 µg IATargeting and suppressing MAPK4, reducing activation of the NF-κB pathway[31]M2 macrophagesmiR−26b−5p--Protecting articular cartilage and improving gait abnormalities by targeting TLR3 and COL10A1, influencing macrophage polarization and chondrocyte hypertrophy[32]M2 macrophagesSox9 mRNAOA model in mice-Enhancing chondrogenic markers and extracellular matrix production while suppressing dedifferentiation markers, mitigating osteoarthritis damage[33]Dendritic cellsKartogenin (KGN)OA model in rats100 µL IAPromoting in situ cartilage formation by synovial fluid MSCs via KGN delivery[35]Dendritic cellsmiR−140OA model in rats100 µg IAAchieving chondrocyte-specific targeting and cartilage penetration, delivering miR−140 to rejuvenate OA cartilage[36]Dendritic cellsCas9 sgMMP−13OA model in rats-Delivering Cas9 sgMMP−13 to chondrocytes, knocking down MMP−13 expression, and reducing the hydrolytic degradation of extracellular matrix proteins[37]Subcutaneous adipose mesenchymal stromal cellsmiR−199a−3pOA model in ratsRats: 50 µL,2 × 1010 particles/mL IAmice: 10 µL,1 × 1010 particles/mL IAInhibiting mTOR and activating autophagy to promote cartilage repair in osteoarthritis.[38]Expi293F cellsSiMMP−13OA model in rats100 µL, 1 mg/mL IAReducing inflammation, promoting chondrocyte proliferation, and reversing OA-related proteomic changes, significantly ameliorating cartilage damage with no major toxicity[39] Exosomal sources and their loaded cargoes for the treatment of OA miR−205−5p miR−766−3p Rats: 50 µL, 2 × 1010 particles/mL IA mice: 10 µL, 1 × 1010 particles/mL IA Microparticles Microparticles have gained prominence as drug delivery systems owing to their small scale, which enables evasion of synovial vascular and lymphatic clearance mechanisms, thereby prolonging intra-articular retention post-injection [5]. Clinically relevant studies of microparticle-based drug delivery for osteoarthritis indicate that the strongest evidence in humans supports intra-articular triamcinolone acetonide (TAA) extended-release, a poly(lactic-co-glycolic acid) (PLGA) microsphere formulation that can provide prolonged local corticosteroid exposure, reduce systemic peak exposure compared with conventional steroid suspensions, and alleviate knee OA pain for several weeks to months [40]. However, the field remains limited by the fact that most platforms beyond triamcinolone microspheres have not yet been validated in humans. Current evidence mainly supports symptomatic pain relief rather than proven disease modification or cartilage regeneration. The clinical translation of microparticle-based drug delivery therapies still requires further exploration and refinement; therefore, this review summarizes the latest preclinical research findings on nanoparticle drug delivery systems, with the aim of providing insights for research aimed at clinical translation. Microparticles can be divided into two types: natural polymer composition and synthetic polymer composition, this review will examine these two types (Fig. 2; Table 2). Fig. 2Different types of microparticles and their advantages as drug delivery systems Different types of microparticles and their advantages as drug delivery systems Natural polymers HA In natural polymers, hyaluronic acid (HA), a naturally occurring macromolecule, serves as a vital constituent in articular cartilage and synovial fluid. Scientific investigations indicate its multifunctional role in joint health, particularly through mechanisms that prevent cartilage deterioration [41]. Due to its widespread presence in the human body, HA is highly biocompatible, biodegradable, and non-immunogenic, making it an ideal material for biomedical applications [42]. Currently, microspheres constructed from HA have been widely used in studies of osteoarthritis treatment. Cao et al. constructed a two-phase release system for the treatment of osteoarthritis by embedding exosomes derived from umbilical cord-derived mesenchymal stem cells with chondrocyte-targeting polymers engineered on the membrane’s surface in thiolated hyaluronic acid microgels. The system improved the therapeutic efficiency and retention time of exosomes in vivo and facilitated the repair of OA cartilage in a rat model [29]. Yang et al. created 2-methylacryloyloxyethyl phosphorylcholine-decorated methacrylate anhydride-hyaluronic acid (HAMA) particles, which exhibited enhanced lubrication and mechanical strength. These particles formed a hydrated lubricating layer on the joint interface, reducing friction. Additionally, the particles served as microcarriers for loading diclofenac sodium (DS) to inhibit inflammatory responses, thereby improving therapeutic outcomes both in vitro and in vivo [43]. Chen et al. developed hybrid exosomes loaded with a CRISPR/Cas9-based FGF18 gene-editing tool, encapsulated in HAMA hydrogel microspheres. This system aimed to address the limitations of current FGF18 protein drugs, such as short half-life and low delivery efficiency. The microgel system provided continuous lubrication and targeted chondrocytes to activate the FGF18 gene, promoting cartilage regeneration and reducing inflammation. In vitro and in vivo experiments demonstrated enhanced chondrocyte proliferation, ECM synthesis, and reduced OA progression, primarily through the activation of the PI3K/AKT signaling pathway [44]. Yang et al. presented a novel approach for the treatment of osteoarthritis using magnetic polysaccharide microcarriers to deliver stem cell-derived exosomes and an anti-inflammatory drug, diclofenac sodium. The proposed microcarriers were composed of modified natural polysaccharides, HAMA, and methacrylate anhydride-chondroitin sulfate, and were generated through microfluidic electrospray combined with a cryogelation process. These microcarriers encapsulated magnetic nanoparticles with spiny structures that efficiently captured exosomes, along with DS. The synergistic release of DS and exosomes from the microcarriers alleviated OA symptoms and promoted cartilage repair. In vitro and in vivo experiments demonstrated the excellent performance of the microcarriers in OA treatment, highlighting their potential for clinical applications [45]. Gelatin Due to its structural similarity to collagen, exceptional biological properties, and inexpensiveness and high availability, gelatin has become an important biomaterial for pharmaceutical delivery applications. This natural polymer demonstrates excellent biocompatibility and undergoes complete metabolic breakdown without generating harmful residues in physiological environments. Its non-antigenic characteristics, combined with predictable degradation patterns, make it particularly suitable for controlled-release formulations. The material’s inherent safety profile stems from sharing identical amino acid sequences and functional groups with endogenous collagen, ensuring seamless integration with biological systems while maintaining therapeutic efficacy [46]. Wang et al. produced a novel low-dose cell therapy system for treating osteoarthritis using MSCs loaded onto gelatin microcarriers. The microcarriers, produced in a spinner flask culture, provided a 3D microenvironment that enhanced MSC retention, survival, and functionality, addressing the challenges of low cell retention and the need for high cell doses in traditional MSC therapies. The study demonstrated that a single injection of MSC-laden microcarriers in a rat model of OA achieved comparable therapeutic effects to repeated high-dose injections of free MSCs, as evaluated through imaging and histological analyses [47]. Peng et al. explored the use of gelatin microcarriers as a delivery system for adipose-derived stem cells (ADSCs) to enhance the treatment of osteoarthritis. The research addressed the limitations of traditional stem cell therapies, such as low retention and reduced paracrine function, by demonstrating that ADSCs cultured on microcarriers exhibited improved paracrine activity and better regulation of macrophages and chondrocytes, which were crucial for OA healing. In vivo experiments in an OA rat model confirmed that microcarrier-based delivery of low-dose ADSCs effectively alleviated OA symptoms, suggesting that this approach could significantly improve the efficacy and reduce the dosage of stem cell therapies for OA [48]. Yang et al. addressed the limitations of conventional OA treatments by developing a system that combined stem cell-derived exosomes with HAMA/Gelatin methacryloyl hydrogel microcarriers to enhance therapeutic effects. The microcarriers were fabricated using microfluidic electrospray technology and liquid nitrogen-assisted cryogelation, resulting in uniform, biocompatible particles. These particles, modified with stem cell-recruiting peptides, encapsulate exosomes within them to promote cartilage repair through sustained release. The system demonstrated significant potential in recruiting endogenous stem cells and enhancing cartilage repair through synergistic effects, as evidenced by in vitro and in vivo experiments [49]. Synthetic polymers PLGA In synthetic polymers, as a widely adopted synthetic polymer in controlled-release drug carriers, PLGA demonstrates remarkable biological advantages, including lower toxicity, higher customizability, great biocompatibility, and the unique ability of sustained drug release [50]. Notably, PLGA’s regulatory status as an FDA-approved biomaterial significantly facilitates its pharmaceutical applications. In osteoarthritis therapeutic development, researchers have engineered diverse PLGA-based microformulations for sustained drug release, with several optimized preparations currently progressing through clinical trial phases [46]. Zhang et al. presented a novel approach to treating partial-thickness cartilage defects in the early stages of osteoarthritis using injectable multifunctional PLGA microspheres. These microspheres were loaded with KGN and modified with an E7 peptide through dopamine self-polymerization to enhance their adhesion to the cartilage matrix. The designed microspheres can adhere to cartilage defects, recruit SMSCs in situ, and promote their differentiation into chondrocytes. Both in vitro and in vivo experiments demonstrated the microspheres’ ability to adhere to the cartilage matrix, recruit SMSCs, and stimulate cartilage differentiation. The treatment resulted in smooth recovery of the cartilage surface and normal glycosaminoglycan content in the model group, while the untreated control group showed significant OA progression [51]. Ye et al. presented a novel therapeutic strategy for osteoarthritis using histone deacetylase inhibitors, specifically Chidamide, encapsulated in PLGA microcapsules to achieve controlled and sustained release of Chidamide, enhancing its therapeutic efficacy. In vitro experiments showed that Chidamide upregulated ECM-related genes and reduced inflammatory factors in chondrocytes. In vivo studies in rats revealed that intra-articular injection of Chidamide-loaded microcapsules significantly alleviated OA symptoms, reduced pain, and protected cartilage from degeneration [52]. Dhanabalan et al. investigated the potential of intra-articular injection of rapamycin-loaded PLGA microparticles to prevent senescence and treat osteoarthritis. The study highlighted the role of autophagy and cellular senescence in OA progression. Rapamycin, known for inducing autophagy and preventing senescence, faced challenges in clinical use due to rapid clearance from joints. The researchers developed these microparticles to prolong rapamycin’s retention in joints, demonstrating that these microparticles induced autophagy, prevented senescence, and sustained sulfated glycosaminoglycans production in human articular chondrocytes from OA patients. In a murine model of post-traumatic OA, the microparticles effectively mitigated cartilage damage and inflammation, both prophylactically and therapeutically [53]. Morille et al. investigated the potential of PLGA-P188-PLGA-based microspheres releasing TGFβ3 to induce chondrogenic differentiation of mesenchymal stem/stromal cells and promote cartilage repair in osteoarthritis. The researchers demonstrated that these microspheres efficiently induced chondrogenic differentiation of human MSCs in vitro, upregulating chondrogenic markers while maintaining the cells’ secretory profile. In vivo experiments using a collagenase-induced OA mouse model showed that MSCs seeded onto the microspheres formed cartilage-like tissue and protected endogenous cartilage from degradation. The study also found that TGFβ3 release by the microspheres enhanced MSC survival in the joint environment, contributing to long-term cartilage protection [54]. PEA Synthesized from alpha-amino acids, aliphatic dicarboxylic acids, and α-ω diols, PEA polymers offer tunable physicochemical properties and enzymatic surface erosion, which allows for predictable and sustained drug release without acidification, in contrast to PLGA-based systems. PEA exhibits structurally adaptable processability, enabling tailored modifications. Key physicochemical properties—such as thermal stability, mechanical strength, solubility, and degradability—can be enhanced by adjusting their chemical compositions and microstructures, all while preserving their excellent biocompatibility [55]. Its distinctive molecular architecture, containing amino acid residues, confers biodegradability through protease-mediated hydrolysis. Notably, synovial fluid contains serine proteases, which are enzymatic agents crucial for mediating inflammatory processes. This enzymatic-environmental responsiveness allows PEA-based drug delivery platforms to synchronize therapeutic release with inflammatory progression in affected tissues, achieving bioactive drug liberation precisely when pathological conditions manifest [46]. PEA microspheres have emerged as a promising platform for controlled intra-articular drug delivery in osteoarthritis therapy. Rudnik-Jansen et al. investigated the efficacy of intra-articular injection of TAA releasing biomaterial microspheres in reducing pain and inflammation in an acute arthritis rat model. The researchers compared a novel PEA microsphere platform with PLGA microspheres and a standard TAA bolus injection. Results showed that TAA-loaded PEA microspheres provided prolonged pain relief and reduced joint swelling more effectively than both PLGA microspheres and the bolus injection. The PEA microspheres also significantly reduced lameness and synovial inflammation over a 42-day period [56]. Tellegen et al. also found when loaded with therapeutics such as triamcinolone acetonide, PEA microspheres exhibit an initial burst release followed by a gradual, long-term release over several months, resulting in prolonged analgesic and anti-inflammatory effects. In preclinical models, a single intra-articular injection of PEA microspheres achieved extended joint retention, reduced synovial inflammation, and improved functional outcomes, while maintaining cartilage integrity. These characteristics highlight PEA microspheres as a versatile and biocompatible nanocarrier for long-acting, localized treatment of osteoarthritis and potentially other joint diseases [57]. Microparticle-based delivery systems have shown considerable potential for OA therapy because they can prolong intra-articular retention and provide sustained release of therapeutic agents. The studies summarized above indicate that natural polymer-based microparticles, including HA and gelatin systems, are advantageous in terms of biocompatibility, biodegradability, lubrication, and biological interactions with cartilage-related cells. These properties make them suitable for delivering exosomes, stem cells, and anti-inflammatory drugs to promote cartilage repair and modulate the joint microenvironment. By contrast, synthetic polymers such as PLGA and PEA provide greater control over particle structure, degradation behavior, drug-loading capacity, and release kinetics, which may improve reproducibility and scalability. However, natural microparticles may be limited by insufficient mechanical strength, rapid degradation, and batch variability, whereas synthetic systems may face challenges related to degradation products, local inflammatory responses, fabrication complexity, and long-term biosafety. In addition, microparticles generally have weaker cartilage penetration than nanoparticles, and their efficacy is highly dependent on particle size, surface modification, degradation rate, and release profile. Moreover, potential burst release, manufacturing complexity, stability issues, and local inflammatory responses should be carefully evaluated before clinical translation. Thus, future microparticle-based systems should aim to integrate the biological advantages of natural polymers with the tunable physicochemical properties of synthetic materials to achieve safer, more durable, and more clinically applicable OA therapy. Table 2Microparticles components and their loaded cargoes for the treatment of OAPolymer componentParticle sizeCargoEncapsulation efficiencyAnimal modelDosagerelease durationIA retentionOutcomeRef.Thiolated hyaluronic acid40–50 μmUmbilical cord-derived mesenchymal stem cell exosomes-OA model in rats-≥ 14 d-Improving therapeutic efficiency, retention time, and cartilage repair[29]methacrylate anhydride-hyaluronic acid (HAMA)47–61 μmDiclofenac sodium (DS)-OA model in rats-≥ 150 h-Enhancing lubrication, reducing friction, and improving anti-inflammatory effects[43]HAMA189.82 ± 19.53 μmCAP/FGF18-hyEXO-OA model in rats100 µL IA≥ 15 d-Promoting cartilage regeneration, reducing inflammation, and activating the PI3K/AKT signaling pathway[44]HAMA/methacrylate anhydride-chondroitin sulfate150–250 μmStem cell-derived exosomes and DSExo: 60%OA model in rats100 µL IA7 d-Alleviating OA symptoms, promoting cartilage repair, and demonstrating potential for clinical applications[45]Gelatin200 ± 7.35 μmMesenchymal stem cells (MSCs)-OA model in rats200 µL,1 × 105 MSCs IA-21 dEnhancing MSC retention, survival, and functionality; achieving therapeutic effects comparable to high-dose MSC injections[47]Gelatin320.93 ± 30.39 μm175.21 ± 21.45 μmAdipose-derived stem cells (ADSCs)-OA model in rats1 × 105 ADSCs IA--Improving paracrine activity, reducing cartilage friction, and enhancing joint lubrication[48]Gelatin methacryloyl/HAMA130–160 μmStem cell-derived exosomes90%OA model in rats100 µL IA16 d-Enhancing cartilage repair and recruiting endogenous stem cells[49]poly(lactic-co-glycolic acid) (PLGA)352.37 ± 11.32 μmKGN and E7 peptide35–40%OA model in rats100 µL,10 mg/mL IAKGN: 28 d E7:>240 h≥ 7 dAdhering to cartilage defects, recruiting SMSCs, and promoting chondrocyte differentiation[51]PLGA2.91 μmChidamide-OA model in rats--8 wkReducing inflammation, alleviating pain, and protecting cartilage from degeneration[52]PLGA1 μmRapamycin32.6 ± 3.007%OA model in mice1.8 µg rapamycin IA30–45 d35 dInducing autophagy, preventing senescence, and mitigating cartilage damage and inflammation[53]PLGA-P188-PLGA-TGFβ3-OA model in mice7 µL IA-12 dInducing chondrogenic differentiation of MSCs, promoting cartilage repair, and protecting endogenous cartilage[54]Poly(Ester Amide) (PEA)23.8 μmTriamcinolone acetonide (TAA)92%OA model in rats25 µL IA42 d-Providing prolonged pain relief, reducing joint swelling, and decreasing synovial inflammation[56]PEA8–50 μmTAA-OA model in dogs0.5 mL IA175 d-Reducing synovial inflammation, and improving functional outcomes, while maintaining cartilage integrity[57] Table 2Microparticles components and their loaded cargoes for the treatment of OAPolymer componentParticle sizeCargoEncapsulation efficiencyAnimal modelDosagerelease durationIA retentionOutcomeRef.Thiolated hyaluronic acid40–50 μmUmbilical cord-derived mesenchymal stem cell exosomes-OA model in rats-≥ 14 d-Improving therapeutic efficiency, retention time, and cartilage repair[29]methacrylate anhydride-hyaluronic acid (HAMA)47–61 μmDiclofenac sodium (DS)-OA model in rats-≥ 150 h-Enhancing lubrication, reducing friction, and improving anti-inflammatory effects[43]HAMA189.82 ± 19.53 μmCAP/FGF18-hyEXO-OA model in rats100 µL IA≥ 15 d-Promoting cartilage regeneration, reducing inflammation, and activating the PI3K/AKT signaling pathway[44]HAMA/methacrylate anhydride-chondroitin sulfate150–250 μmStem cell-derived exosomes and DSExo: 60%OA model in rats100 µL IA7 d-Alleviating OA symptoms, promoting cartilage repair, and demonstrating potential for clinical applications[45]Gelatin200 ± 7.35 μmMesenchymal stem cells (MSCs)-OA model in rats200 µL,1 × 105 MSCs IA-21 dEnhancing MSC retention, survival, and functionality; achieving therapeutic effects comparable to high-dose MSC injections[47]Gelatin320.93 ± 30.39 μm175.21 ± 21.45 μmAdipose-derived stem cells (ADSCs)-OA model in rats1 × 105 ADSCs IA--Improving paracrine activity, reducing cartilage friction, and enhancing joint lubrication[48]Gelatin methacryloyl/HAMA130–160 μmStem cell-derived exosomes90%OA model in rats100 µL IA16 d-Enhancing cartilage repair and recruiting endogenous stem cells[49]poly(lactic-co-glycolic acid) (PLGA)352.37 ± 11.32 μmKGN and E7 peptide35–40%OA model in rats100 µL,10 mg/mL IAKGN: 28 d E7:>240 h≥ 7 dAdhering to cartilage defects, recruiting SMSCs, and promoting chondrocyte differentiation[51]PLGA2.91 μmChidamide-OA model in rats--8 wkReducing inflammation, alleviating pain, and protecting cartilage from degeneration[52]PLGA1 μmRapamycin32.6 ± 3.007%OA model in mice1.8 µg rapamycin IA30–45 d35 dInducing autophagy, preventing senescence, and mitigating cartilage damage and inflammation[53]PLGA-P188-PLGA-TGFβ3-OA model in mice7 µL IA-12 dInducing chondrogenic differentiation of MSCs, promoting cartilage repair, and protecting endogenous cartilage[54]Poly(Ester Amide) (PEA)23.8 μmTriamcinolone acetonide (TAA)92%OA model in rats25 µL IA42 d-Providing prolonged pain relief, reducing joint swelling, and decreasing synovial inflammation[56]PEA8–50 μmTAA-OA model in dogs0.5 mL IA175 d-Reducing synovial inflammation, and improving functional outcomes, while maintaining cartilage integrity[57] Microparticles components and their loaded cargoes for the treatment of OA 200 µL, 1 × 105 MSCs IA 320.93 ± 30.39 μm 175.21 ± 21.45 μm 100 µL, 10 mg/mL IA Nanoparticles Nanoparticles, owing to their small size, offer dual advantages in joint therapy: enhancing drug retention time and enabling precise targeting. Their small dimensions not only improve the efficiency of targeting moieties but also promote deeper penetration and diffusion through the dense cartilage matrix [5]. Clinical research on nanoparticle-based drug delivery for osteoarthritis has shown promising but still limited therapeutic effects, mainly in symptom relief rather than proven disease modification. The strongest clinical evidence comes from intra-articular liposomal dexamethasone, TLC599, which demonstrated sustained reductions in knee osteoarthritis pain and improvements in WOMAC outcomes for up to several months, with generally acceptable safety in phase II and reported phase III studies [58]. Oral nanocurcumin or bioavailability-enhanced curcumin formulations also showed short-term improvements in pain, stiffness, and physical function in small randomized trials and meta-analyses [59]. However, the field remains at an early translational stage: most nanoparticle systems for osteoarthritis are still supported mainly by preclinical data, sample sizes in human studies are often modest, formulations are heterogeneous, follow-up durations are relatively short, and there is insufficient evidence that these therapies can slow cartilage degeneration or alter structural disease progression. Therefore, this review focuses on the latest preclinical research on nanomedicine delivery systems. At present, nanoparticle targeting is commonly enhanced through biomimetic membrane encapsulation or surface modification with targeting ligands [60]. A number of nanomaterials with certain physiological properties, such as reactive oxygen species (ROS) responsiveness, pH responsiveness, and enzyme responsiveness, have been developed in the field of osteoarthritis therapeutics [61]. Nanoparticles constructed using these materials will provide new solutions for OA treatment (Fig. 3). Fig. 3Illustration of stimulus-responsive drug delivery systems for treatment of OA Illustration of stimulus-responsive drug delivery systems for treatment of OA ROS-responsive nanoparticles The pathogenesis of osteoarthritis involves oxidative stress mediated by ROS, contributing to articular cartilage deterioration through apoptosis and dysfunction of chondrocytes and degradation of the ECM in cartilage [62]. Li et al. developed an innovative ROS-responsive polymeric system capable of co-assembling with astaxanthin and the autophagy activator rapamycin into functional nanoparticles. The nanoparticles served as both a drug carrier and a therapeutic ROS-scavenging platform: their thioketal-containing polymer structure responded to the high-ROS microenvironment of M1 macrophages, causing nanoparticle dissociation and controlled release of rapamycin, while astaxanthin and the polymer backbone further depleted ROS. In vitro, this system was efficiently taken up by M1 macrophages, markedly reduced intracellular ROS, restored mitochondrial membrane potential and ATP production, increased antioxidant capacity, activated autophagy, and inhibited NLRP3 inflammasome-related inflammatory signaling. As a result, it shifted macrophages from the pro-inflammatory M1 phenotype toward the anti-inflammatory M2 phenotype, reducing inflammatory cytokine release and protecting chondrocytes from apoptosis and matrix degradation. In an ACLT-induced osteoarthritis mouse model, intra-articular injection of the system improved drug retention and delivery at the synovial inflammatory site, reduced joint inflammation, promoted M2 macrophage repolarization, lowered cartilage degeneration scores, and protected articular cartilage [61]. Jiang et al. presented a cartilage-targeted, ROS-responsive nanocarrier system by loading the antioxidant small molecule oltipraz into positively charged mesoporous silica nanoparticles and modifying them with methoxy polyethylene glycol (PEG)–thioketal linkers. High ROS conditions disrupt the system architecture and open the nanoparticle pores, allowing accelerated oltipraz release inside ROS-stressed chondrocytes. In vitro experiments demonstrated that the system enhanced ROS scavenging, activated the Nrf2/HO−1 antioxidant pathway, reduced chondrocyte apoptosis, and mitigated extracellular matrix degradation compared to free oltipraz. In vivo studies using a rat OA model showed prolonged retention of the system in joints, effective cartilage penetration, and significant alleviation of OA progression by preserving cartilage integrity, reducing subchondral bone remodeling, and improving gait parameters [62]. Kou et al. developed a novel ROS-responsive nanoadsorbent, engineered to selectively target M1 macrophages and facilitate their phenotypic transition for osteoarthritis treatment. Its ROS-responsive drug release relied mainly on bilirubin: under the high oxidative-stress environment of activated M1 macrophages, hydrophobic bilirubin in the bilirubin-grafted polylysine core was oxidized and converted into hydrophilic biliverdin, disrupting the nanoparticle structure and accelerating berberine release. In vitro evaluations revealed the nanoparticle’s specific internalization by activated M1 macrophages, triggering a phenotypic shift towards anti-inflammatory M2 polarization through dual mechanisms: effectively scavenging reactive oxygen species and suppressing NF-κB signaling activation. This macrophage reprogramming mechanism was shown to create a regenerative microenvironment conducive to chondrocyte viability and function. Substantiating these findings, in vivo experiments demonstrated significant enhancement in articular cartilage restoration following therapeutic application of the engineered nanoparticle system [63]. Some other researchers had also conducted experiments using ROS-responsive nanoparticles, again with good results [64, 65]. By reacting to elevated reactive oxygen species, the nanoparticles release the drugs they carry, producing favorable therapeutic effects on osteoarthritis. pH-responsive nanoparticles The acidic microenvironment is a typical hallmark of OA, and synovial inflammation is closely associated with the acidic microenvironment of OA. Yan et al. developed an innovative nanotherapeutic system through the integration of carbonic anhydrase IX (CA9)-targeted siRNA and nitric oxide scavenging components. These engineered particles demonstrated effective siRNA encapsulation and structural stability while exhibiting pH-sensitive degradation characteristics that facilitated controlled release of therapeutic siRNA and simultaneous nitric oxide neutralization. The pH-responsive drug release mainly depended on the calcium phosphate core: after the nanoparticles were internalized by inflammatory macrophages, they entered acidic endosomes and lysosomes, where the lower pH weakened and disassembled the phosphate/siRNA structure, causing the particles to become loose and irregular and thereby rapidly releasing siCA9 into the cytoplasm. This pH-triggered release enabled efficient CA9 gene silencing, while the NAHA shell simultaneously scavenged NO through its o-phenylenediamine groups. Experimental results revealed that the dual-functional nanoparticles could effectively suppress CA9 expression in synovial macrophages while eliminating excess nitric oxide, thereby achieving efficient and safe therapeutic outcomes for osteoarthritis management [66]. In addition, relevant studies have shown that metal-organic framework-based systems can reduce premature drug leakage, enable stimulus-responsive drug release, and enhance drug loading capacity [67]. Sun et al. developed a nanoparticle formulation through covalent functionalization of HA on a zeolitic imidazolate framework−8 (ZIF−8) metal-organic framework loaded with KGN. This nanocomposite demonstrated the ability to overcome challenges related to drug penetration barriers and controlled release kinetics in dense cartilage matrices, enabling pH-responsive KGN delivery specifically at osteoarthritic lesions. The pH-responsive release mainly relies on the acid-sensitive degradation of the ZIF−8 framework: after uptake by chondrocytes or exposure to the weakly acidic osteoarthritic/lysosomal microenvironment, the ZIF−8 structure gradually decomposes, leading to sustained release of KGN and Zn²⁺. The gradual release of KGN from the composite stimulated extracellular matrix production, which is essential for articular cartilage repair. Simultaneously, Zn²⁺ ions liberated during degradation contributed to anti-inflammatory effects through M2 macrophage polarization modulation, thereby interrupting the pathological cascade of OA. As a result, this therapeutic system effectively promoted cartilage regeneration and mitigated OA progression within a three-month intervention period [68]. In addition, many other researchers have also studied pH-responsive nanoparticles [69–72]. Researchers develop pH-responsive nanoparticles to achieve precise drug release, enabling highly effective treatment for osteoarthritis. Enzyme-responsive nanoparticles Overexpression of matrix metalloproteinases is another feature of the osteoarthritic microenvironment [73]. Thus, matrix metalloproteinase is a key biomarker for early OA diagnosis and a central factor in initiating OA pathogenesis and continuing to promote OA progression [74]. Zhou et al. presented the development of MMP−13-targeted siRNA-loaded micelles for theranostic applications in early-stage posttraumatic osteoarthritis (PTOA). These micelles incorporated a MMP−13-cleavable peptide linker, cyanine 5 for fluorescence imaging, and cyclic Arg-Gly-Asp peptide (cRGD) ligands for targeted delivery. Upon encountering MMP−13-overexpressing diseased chondrocytes, the micelles shed their PEG shell, exposing cRGD to enhance cellular uptake and releasing siRNA to silence MMP−13, thereby reducing collagen degradation. Simultaneously, MMP−13-triggered cyanine 5 fluorescence restoration enabled real-time monitoring of disease severity and treatment response. In murine PTOA models, the micelles demonstrated prolonged joint retention, effective MMP−13 downregulation, cartilage protection, and osteophyte reduction, while exhibiting minimal toxicity [74]. Lu et al. developed hierarchical functional nanoparticles for osteoarthritis therapy. After intra-articular injection, the nanoparticles first targeted cartilage via an outer collagen-binding sequence. In the OA microenvironment, MMP−2 cleaved the spacer, exposing an inner peptide that recruited joint-resident MSCs. The nanoparticles then promoted chondrogenesis of MSCs into chondrocytes. In a rat OA model, these nanoparticles significantly reduced cartilage destruction by generating hyaline cartilage and inhibiting the PI3K/AKT/mTOR pathway [75]. By utilizing MMP-responsive nanoparticles, the sustainability of osteoarthritis treatment has been achieved, enhancing drug utilization. Multifactorial responsive nanoparticles Multifactorial responsive nanoparticles are designed to achieve more precise treatment than single-factor response nanocarriers. Chen et al. designed a cartilage-targeting, dual MMP−13/pH-responsive theranostic nanoprobe for OA imaging and therapy. Utilizing biocompatible ferritin nanocages modified with a collagen II-targeting peptide, the nanoprobes specifically localized to cartilage and responded to OA-specific stimulations: overexpression of MMP−13 and acidic pH. The MMP−13-sensitive cleavage of a conjugated peptide substrate activated near-infrared fluorescence for real-time imaging of OA severity, while the acidic microenvironment triggered sustained release of the anti-inflammatory drug hydroxychloroquine. In vitro and in vivo studies demonstrated enhanced drug retention, reduced synovial inflammation, and improved cartilage repair in OA mice [73]. Lan et al. constructed a theranostic nanoplatform designed for OA treatment, leveraging the OA-specific acidic microenvironment and overexpressed MMP−13 to enable targeted drug delivery and diagnostic imaging. The nano-micelles incorporated a collagen II-targeting peptide for joint retention, an MMP−13-responsive fluorescent probe to report OA severity, and psoralidin, an anti-inflammatory drug from traditional Chinese medicine, encapsulated within a pH-sensitive polymer for controlled release. In osteoarthritic joints, overexpressed MMP−13 cleaves the GPLGVRGC peptide, separating Cyanine 5.5 from Black Hole Quencher-3 and restoring fluorescence, thereby indicating the presence of OA lesions and the biodistribution of the micelles. At the same time, the acidic microenvironment associated with osteoarthritis disrupts the micellar structure and accelerates the release of psoralidin. In vitro studies demonstrated the system’s biocompatibility, MMP−13-triggered fluorescence activation, and pH-dependent drug release, alongside reduced inflammatory markers and rescued chondrocyte viability via inhibition of MAPK, NF-κB, and PI3K/Akt pathways. In vivo experiments in OA-induced mice showed prolonged joint retention, reduced cartilage erosion, and diminished MMP−13 expression, with histological improvements in glycosaminoglycan deposition and joint morphology [76]. Nanoparticle-based delivery systems have emerged as promising platforms for OA therapy because they can improve drug stability, prolong intra-articular retention, and enhance penetration into the dense cartilage matrix. The studies summarized above indicate that responsive nanoparticles can exploit OA-specific pathological cues, including elevated ROS levels, acidic pH, and increased MMP activity, to achieve selective and controlled drug release. Mechanistically, ROS-responsive nanoparticles couple therapeutic cargo release with oxidative stress scavenging, mitochondrial protection, inflammatory signaling inhibition, and macrophage M2 polarization. pH-responsive systems promote cargo liberation in acidic inflammatory or lysosomal environments, whereas enzyme-responsive nanoparticles use disease-associated proteases to trigger structural changes, targeting ligand exposure, imaging activation, or therapeutic cargo release. Multifactorial responsive systems further improve specificity by integrating multiple OA-related signals, thereby enhancing both therapeutic accuracy and diagnostic potential. However, nanoparticles still face important translational barriers, including potential cytotoxicity, immunogenicity, long-term accumulation, unclear clearance pathways, complex fabrication, and difficulties in large-scale quality control. Future studies should therefore focus on biodegradable materials, simplified and reproducible preparation methods, optimized particle size and surface modification, and comprehensive safety evaluation after repeated intra-articular administration. Hydrogels Hydrogels are three-dimensional hydrophilic polymer networks capable of absorbing significant amounts of water—sometimes exceeding 90% of their dry weight—without dissolving, thereby maintaining their structural integrity. This unique behavior arises from hydrophilic functional groups on the polymer backbone, which enable water absorption and retention, and crosslinks between polymer chains, which prevent dissolution and ensure stability in aqueous environments. Hydrogels can respond to stimuli such as pH or ROS changes, triggering significant transformations in their network structure, swelling behavior, permeability, and mechanical strength [77]. Currently, hydrogels are used in various medical fields, the most mature of which is aesthetic medicine. In recent years, hydrogels have also made significant progress in the treatment of osteoarthritis, spinal fusion, advanced heart failure, type 2 diabetes, chronic kidney disease, and other diseases [78]. Clinically relevant research on hydrogel-based drug delivery for osteoarthritis has advanced rapidly at the preclinical level, but true human clinical evidence remains limited. Most clinical studies involve hydrogel-like intra-articular injectables such as hyaluronic acid derivatives or polyacrylamide hydrogels, which mainly act as viscosupplementation, cushioning, or joint-environment modifiers rather than as classic drug-loaded delivery systems. In contrast, drug-loaded, stimuli-responsive, injectable, or mechanically resilient hydrogels have shown strong potential in animal and translational studies by prolonging joint retention, enabling sustained local release, reducing inflammation, and protecting cartilage. This review summarizes the research findings in this part. Similar to nanoparticles, modern biomedical research has developed responsive hydrogel systems capable of adapting to OA pathologies through multiple biochemical triggers. These injectable platforms exhibit reactions to pathological ROS levels, acidic microenvironmental pH variations, and abnormally elevated MMP concentrations [79]. ROS-responsive hydrogels In the study of ROS-responsive hydrogels, a ROS-responsive hydrogel was designed by Wang et al. to attenuate OA progression by integrating MMP−13 gene silencing, ROS scavenging, prolonged joint retention, and hydrogel-based mechanical support. In this system, Fe₃O₄-polyethyleneimine-PEG nanoparticles were used to load and protect MMP−13 siRNA, forming si-Fe nanocarriers that enhanced intracellular siRNA delivery and gene-silencing efficiency; these nanocarriers were then encapsulated in the hydrogel composed of a matrix of hyaluronic acid-phenylboronic acid and polyvinyl alcohol, which scavenged H₂O₂ and released si-Fe nanoparticles in response to elevated ROS levels in the inflammatory OA microenvironment. High H₂O₂ concentrations accelerated hydrogel degradation and si-Fe release, whereas lower inflammatory conditions prolonged nanoparticle retention, enabling an intelligent and sustained drug-delivery system. Mechanically, its HA-based network also supported long-term intra-articular retention and may help provide local lubrication and mechanical protection within the joint [80]. Chen et al. constructed a novel collagen hydrogel, which alleviated inflammation and promoted cartilage regeneration in a rat cartilage-defect model by combining staged nanoparticle delivery with a mechanically supportive, biocompatible collagen scaffold. In terms of drug delivery, the collagen matrix directly embedded CeO₂ nanoparticles and PLGA microspheres encapsulating Fe₂O₃ nanoparticles, enabling ROS-responsive sequential release: rapid CeO₂ release under high-ROS inflammatory conditions to reduce early-stage oxidative stress and inflammation, followed by sustained Fe₂O₃ release to support cartilage regeneration during the proliferative phase. For mechanical support, the sulfhydryl-crosslinked collagen network showed enhanced mechanical strength and injectability, and it functioned as an ECM-mimicking scaffold that supported cell adhesion, proliferation, and chondrogenic differentiation [81]. Lu et al. developed a novel dual-layer hydrogel system that enables the staged release of drugs through a structurally supportive hydrogel matrix, thereby improving the inflammatory microenvironment and promoting cartilage regeneration, which enhances the repair of cartilage defects. For drug delivery, the upper phenylboronic acid-crosslinked polyvinyl alcohol hydrogel rapidly released sodium diclofenac under high-ROS conditions to reduce oxidative stress and reprogram macrophages toward a pro-healing phenotype, while the lower HA-based hydrogel provided sustained release of BMSC-derived exosomes to promote BMSC recruitment, migration, and chondrogenic differentiation [82]. In addition, improvement of mitochondrial function using a ROS-responsive hydrogel platform by Liu et al. opened up a new approach to OA treatment [83]. pH-responsive hydrogels pH-responsive hydrogels are also a hot topic of research. Such pH-responsive characteristics are particularly desirable for drug delivery applications, because they help prevent premature drug leakage during systemic circulation and facilitate rapid drug release upon arrival at the target cells [84]. Zhou et al. fabricated an injectable hydrogel that alleviated OA through the combined effects of MnO₂-mediated ROS scavenging, platelet-rich plasma-mediated cartilage repair, and HA-mediated mechanical support, leading to reduced oxidative stress, suppressed cartilage matrix degradation, and improved cartilage protection in vivo. As a delivery platform, the hydrogel encapsulated bovine serum albumin-stabilized MnO₂ nanozymes and platelet-rich plasma, enabling faster release of MnO₂ nanoparticles, total platelet-rich plasma proteins, and the representative growth factor TGF-β1 under the weakly acidic OA-like condition than under normal physiological pH, thereby supporting on-demand antioxidant and pro-regenerative therapy in the inflammatory joint microenvironment. Mechanically, the Schiff-base-crosslinked HA network showed rheological properties, shear-thinning injectability, and self-healing behavior, allowing it to function as a viscosupplement that could provide lubrication and mechanical dissipation within the joint [85]. Jiang et al. constructed a novel hydrogel system enabling cartilage repair by providing localized immunomodulatory neobavaisoflavone delivery, supporting macrophage polarization toward an anti-inflammatory M2 phenotype, enhancing BMSC chondrogenesis, and mechanically maintaining the cartilage defect space for tissue integration and neocartilage formation. For drug delivery, neobavaisoflavone was encapsulated in ZIF-8 metal-organic frameworks and incorporated into a GelMA/HA hydrogel network, enabling release in response to the weakly acidic and enzymatic inflammatory microenvironment, facilitating on-demand drug release at inflamed defect sites. For mechanical support, polydopamine-decorated ZIF-8 metal-organic frameworks were uniformly dispersed in the hydrogel and acted as nanoreinforcing agents, endowing the hydrogel with high resilience and toughness [86]. Not only that, but the research conducted by Liang and colleagues is equally refreshing: they utilized self-assembled rhein hydrogels both as a bioactive drug delivery platform and as an intra-articular scaffold for the treatment of osteoarthritis [87]. Enzyme-responsive hydrogels Enzyme-responsive hydrogels are equally effective in controlling drug release. Miao et al. developed an injectable, MMP−13-responsive bilayer hydrogel microsphere. The gelatin methacryloyl outer layer demonstrated therapeutic efficacy by responding to elevated MMP−13 concentrations characteristic of osteoarthritic environments. Through rapid enzymatic degradation of this hydrogel shell, controlled release of celecoxib from encapsulated liposomes was achieved, effectively inhibiting pro-inflammatory mediator production. Concurrently, the exposed chondroitin sulfate methacryloyl core gradually degraded and released chondroitin sulfate to promote chondrocyte anabolism, inhibit catabolism, and support cartilage repair. This dual release mechanism synergistically enhanced cartilage regeneration through both anti-inflammatory action and chondroprotective effects. On the other hand, hydrogel microspheres have a porous spherical structure and an optimal particle size, which helps them roll within the joint and produce a “ball-bearing”-like lubricating effect; meanwhile, liposomes form a hydrated lubricating layer under the shear stress of the joint, helping to reduce mechanical friction within osteoarthritic joints [88]. Zhou et al. developed an injectable, hypoxia- and MMP-13-responsive hydrogel microsphere system for the treatment of osteoarthritis. in the OA microenvironment, excessive MMP−13 promoted microsphere degradation, while hypoxia triggered hydroxychloroquine release, enabling on-demand anti-inflammatory drug delivery in inflamed joints. The methacrylate-modified sulfonated azocalix [4]arene component also acted as a therapeutic carrier by scavenging ROS, reducing oxidative stress, and synergizing with hydroxychloroquine to suppress macrophage inflammation. Post-injection evaluation demonstrated the therapeutic system’s multifunctional efficacy in mitigating oxidative stress, suppressing hypoxia-inducible factor−1α activation, and downregulating pro-inflammatory cytokine production. Furthermore, the intervention exhibited chondroprotective effects by significantly delaying articular cartilage degeneration [89]. Multifactorial responsive hydrogels Multifactorial responsive hydrogels enable more precise treatment. Zhang et al. developed a lubricating injectable hydrogel with dual environmental responsiveness. This hydrogel alleviated OA by integrating pH/ROS-responsive bioactive release, antioxidant and immunomodulatory effects, and lubrication-based mechanical protection. Under the weakly acidic and ROS-rich OA microenvironment, the methionine-modified carboxymethyl chitosan shell collapsed and ZIF−8 degraded, enabling sustained release of Quercetin and Zn²⁺ to scavenge ROS, inhibit inflammation, promote macrophage polarization from M1 to M2, and support cartilage repair. Furthermore, this hydrogel acted as a physical lubricating barrier within the joint cavity to reduce friction between cartilage surfaces [90]. Xia et al. developed a novel dual-responsive hydrogel system, employing microsphere architecture for targeted drug delivery. These engineered particles demonstrated stimuli-triggered drug release characteristics specifically activated by elevated concentrations of MMP−13 enzymatic activity and ROS. In the OA microenvironment, ROS triggered the early release of dihydromyricetin during the first week, while MMP-mediated degradation promoted sustained release in the following weeks, thereby enabling sequential and context-dependent regulation of damaged cartilage. This hydrogel system activated endogenous SIRT3, suppressed mitochondrial apoptosis, promoted PINK1/Parkin-LC3B-mediated mitophagy, restored ECM anabolism–catabolism balance, reduced cartilage abrasion and subchondral bone sclerosis, and improved motor function in a PTOA mouse model by combining responsive dihydromyricetin delivery with lubrication-based joint protection [91]. Other researchers who had adopted the same strategy had likewise made breakthroughs [92, 93]. Hydrogel-based delivery systems have emerged as versatile platforms for OA therapy because they combine localized retention, sustained release, microenvironmental regulation, and mechanical support. The studies summarized above demonstrate that responsive hydrogels can sense pathological features of the OA microenvironment, such as excessive ROS, acidic pH, hypoxia, and elevated MMP activity, and subsequently release anti-inflammatory agents, antioxidants, exosomes, growth factors, or nucleic acid therapeutics in a controlled manner. These systems can reduce oxidative stress and synovial inflammation, promote macrophage M2 polarization, enhance chondrocyte anabolic activity, inhibit ECM degradation, and facilitate cartilage regeneration. Beyond biochemical modulation, hydrogels provide unique biomechanical advantages by filling cartilage defects, improving lubrication, dissipating local stress, maintaining defect space, and acting as extracellular matrix-like scaffolds for cell adhesion, migration, chondrogenic differentiation, and matrix deposition. Nevertheless, their therapeutic efficacy depends strongly on gelation time, injectability, swelling ratio, degradation rate, crosslinking density, adhesive strength, and load-bearing capacity. Insufficient mechanical robustness, poor integration with wet cartilage surfaces, burst release, mismatched degradation, complex fabrication, and uncertain long-term biosafety remain important translational challenges. Future studies should therefore focus on developing hydrogels that integrate disease-responsive drug release with cartilage-mimetic mechanical properties, durable joint retention, reproducible preparation, and validated safety in clinically relevant OA models. Conclusions and prospects Osteoarthritis remains a significant cause of disability and societal burden, particularly among older adults. Current treatment methods, such as local and systemic drug administration, have difficulties in achieving the expected efficacy due to limitations in terms of drug retention, efficacy, and potential side effects. Recent advances in innovative drug delivery systems, including exosomes, microparticles, nanoparticles, and hydrogels, have opened new avenues for not only improving the efficiency by delivering drugs or biomolecules to target pathological tissues or cells directly, but also having favorable biocompatibility and less immunogenicity. Growing evidence has demonstrated the great potential of these drug delivery systems in OA therapy. Although exosomes, microparticles, nanoparticles, and hydrogels all aim to improve local drug availability in osteoarthritic joints, they differ substantially in loading capacity, intra-articular persistence, cartilage penetration, targeting precision, and translational feasibility (Table 3). Table 3Comparative perspective on exosomes, microparticles, nanoparticles, and hydrogelsDelivery systemDrug Loading EfficiencyRetention TimePenetration into CartilageTargeting SpecificityScalabilityClinical Translation PotentialExosomesModerate; can be enhanced via electroporation, sonication, or engineered cargo loadingShort-to-moderate; naturally cleared quickly unless modifiedExcellent; nanoscale size enables deep tissue penetrationHigh when engineered with tissue-specific peptidesLimited; donor variability and complex isolation hinder large-scale productionPreclinical and early clinical trials; translational challenges due to heterogeneity and reproducibilityMicroparticlesModerate-to-high; depends on polymer type and fabrication methodLong; can sustain release for days to monthsModerate; larger size limits penetration into dense cartilageModerate; can be improved with surface modificationHigh; manufacturing processes for PLGA/PEA are establishedSeveral formulations (e.g., PLGA- or PEA-based) under clinical evaluationNanoparticlesVariable; dependent on material and surface chemistryModerate; stimuli-responsive designs allow controlled releaseExcellent; small size allows diffusion into dense ECMHigh; can be functionalized with targeting ligands or responsive moietiesModerate; scalable synthesis is feasible but complex for multifunctional designsPreclinical stage; limited clinical evidence, mostly for symptom reliefHydrogelsHigh; 3D matrix allows large payload encapsulationVery long; depot effect prolongs drug retentionLimited; penetration into cartilage is restricted unless combined with carriersLow-to-moderate; usually requires embedded carriers or ligands for targetingHigh; hydrogel fabrication is scalablePromising; especially for combination therapies with cells or nanoparticles for regenerative applications Table 3Comparative perspective on exosomes, microparticles, nanoparticles, and hydrogelsDelivery systemDrug Loading EfficiencyRetention TimePenetration into CartilageTargeting SpecificityScalabilityClinical Translation PotentialExosomesModerate; can be enhanced via electroporation, sonication, or engineered cargo loadingShort-to-moderate; naturally cleared quickly unless modifiedExcellent; nanoscale size enables deep tissue penetrationHigh when engineered with tissue-specific peptidesLimited; donor variability and complex isolation hinder large-scale productionPreclinical and early clinical trials; translational challenges due to heterogeneity and reproducibilityMicroparticlesModerate-to-high; depends on polymer type and fabrication methodLong; can sustain release for days to monthsModerate; larger size limits penetration into dense cartilageModerate; can be improved with surface modificationHigh; manufacturing processes for PLGA/PEA are establishedSeveral formulations (e.g., PLGA- or PEA-based) under clinical evaluationNanoparticlesVariable; dependent on material and surface chemistryModerate; stimuli-responsive designs allow controlled releaseExcellent; small size allows diffusion into dense ECMHigh; can be functionalized with targeting ligands or responsive moietiesModerate; scalable synthesis is feasible but complex for multifunctional designsPreclinical stage; limited clinical evidence, mostly for symptom reliefHydrogelsHigh; 3D matrix allows large payload encapsulationVery long; depot effect prolongs drug retentionLimited; penetration into cartilage is restricted unless combined with carriersLow-to-moderate; usually requires embedded carriers or ligands for targetingHigh; hydrogel fabrication is scalablePromising; especially for combination therapies with cells or nanoparticles for regenerative applications Comparative perspective on exosomes, microparticles, nanoparticles, and hydrogels Exosomes possess intrinsic biological activity and can naturally carry proteins, lipids, mRNAs, miRNAs, and other regulatory molecules; however, their drug-loading efficiency is often cargo- and method-dependent, and active loading strategies such as electroporation, sonication, or cell engineering may compromise vesicle integrity or introduce batch variability. By contrast, microparticles, especially PLGA-, PEA-, hyaluronic acid-, and gelatin-based systems, generally provide higher and more controllable loading capacity for small molecules, proteins, cells, or extracellular vesicles, while nanoparticles offer tunable encapsulation and surface adsorption owing to their large surface-area-to-volume ratio. Hydrogels can accommodate a broad range of therapeutic agents, including drugs, cells, exosomes, and nanoparticles, but their loading efficiency and release profile are highly dependent on network density, crosslinking chemistry, and cargo–matrix interactions. In terms of retention time, microparticles and hydrogels are usually more advantageous than free exosomes or nanoparticles because their larger size or three-dimensional polymeric network reduces rapid clearance from the synovial cavity and enables sustained release over days to months. Hydrogels are particularly suitable when prolonged local residence and mechanical support are required, as they can act simultaneously as drug depots and tissue scaffolds. Microparticles also provide durable intra-articular retention and controlled release, but their relatively large size limits direct penetration into dense cartilage. Nanoparticles and exosomes, owing to their nanoscale dimensions, show superior diffusion into cartilage matrix and are more suitable for reaching chondrocytes within deeper cartilage regions. Nevertheless, very small nanoparticles may be cleared more rapidly unless optimized through charge modulation, biomimetic coating, or incorporation into hydrogel or microparticle depots. Targeting specificity also differs among these systems. Natural exosomes exhibit inherent cell-communication properties and low immunogenicity, while engineered exosomes modified with cartilage-affinity peptides or chondrocyte-targeting ligands can further improve cartilage localization and cargo-delivery specificity. Nanoparticles are highly customizable and can be functionalized with targeting ligands, biomimetic membranes, or stimuli-responsive motifs that respond to pathological cues such as reactive oxygen species, acidic pH, or disease-associated enzymes. Microparticles and hydrogels mainly achieve targeting through local intra-articular retention and controlled release rather than deep cellular targeting, although their specificity can be improved by incorporating targeting peptides, magnetic components, or responsive linkers. Therefore, exosomes and nanoparticles are generally superior for cell-specific or cartilage-penetrating delivery, whereas microparticles and hydrogels are more suitable for depot-based, long-acting, and structurally supportive therapy. From the perspective of scalability and clinical translation, synthetic microparticles currently have the most mature translational foundation, particularly because polymers such as PLGA have established manufacturing experience and regulatory familiarity. Hydrogels also have considerable clinical potential, especially when based on biocompatible materials such as hyaluronic acid, gelatin, or PEG derivatives, bu t their translation requires careful control of injectability, sterilization, mechanical properties, degradation behavior, and reproducibility. Nanoparticles offer high design flexibility and disease-responsive release, yet their clinical development is constrained by formulation heterogeneity, long-term biosafety concerns, and the need for scalable, quality-controlled production. Exosomes are biologically attractive but face the greatest manufacturing and regulatory challenges, including donor variability, heterogeneous cargo composition, low and variable loading efficiency, complex isolation and purification, potency testing, storage stability, and batch-to-batch reproducibility. Overall, microparticles and hydrogels may be more readily scalable as sustained-release platforms, nanoparticles may provide the best balance between penetration and engineerability, and exosomes may offer the highest biological sophistication but require the most rigorous standardization before broad clinical translation. Despite the encouraging therapeutic potential of these drug delivery systems for osteoarthritis, their immunogenicity and long-term biosafety must be systematically evaluated before clinical translation. Because most current studies rely on short-term animal models, the long-term biodistribution, intra-articular persistence, and extra-articular migration of injected carriers remain incompletely understood. After intra-articular administration, nanoparticles, microparticles, exosomes, and hydrogel fragments may be retained within cartilage, synovium, meniscus, or subchondral bone, but they may also drain through lymphatic routes or enter systemic circulation. Therefore, quantitative tracking of carrier distribution in joint tissues and major organs, including the liver, spleen, kidney, lung, and lymph nodes, is necessary to clarify whether repeated administration leads to unintended accumulation or delayed clearance. Particle accumulation is particularly relevant for inorganic, cationic, or slowly degradable systems, which may persist in synovial macrophages or cartilage matrix and induce local inflammatory responses. Although prolonged joint retention is often considered beneficial for sustained drug release, excessive persistence may disturb synovial homeostasis, alter macrophage polarization, or impair cartilage cell viability. Similarly, polymeric systems such as PLGA, PEA, hyaluronic acid, gelatin, and hydrogel-based matrices should be assessed not only for carrier biocompatibility but also for the biological effects of their degradation products. Acidic degradation products, residual crosslinkers, unreacted monomers, inorganic residues, or by-products generated under ROS-, pH-, or enzyme-responsive conditions may cause cytotoxicity, oxidative stress, synovitis, or matrix catabolism if they accumulate locally. From a pharmacokinetic/pharmacodynamic perspective, prolonged intra-articular retention should not be interpreted as therapeutic superiority unless it is linked to well-defined drug release kinetics, tissue exposure, and biological response. Future studies should move beyond qualitative retention imaging and establish quantitative PK profiles for both carriers and released cargoes, including burst release, sustained-release phases, degradation-dependent liberation, synovial fluid half-life, cartilage and synovial tissue penetration, lymphatic drainage, and systemic leakage. This is particularly important because the joint cavity is not a closed compartment: synovial fluid turnover, vascular and lymphatic clearance, phagocytosis by synovial macrophages, and transport barriers within the dense cartilage extracellular matrix can markedly alter the actual concentration reaching chondrocytes, synoviocytes, or immune cells. In addition, carrier size, surface charge, stiffness, hydrophilicity, targeting ligands, and degradation rate may differentially influence synovial distribution, cartilage diffusion, cellular uptake, and extra-articular clearance. Therefore, rational design of osteoarthritis drug delivery systems should integrate release kinetics with PK/PD modeling to determine whether local drug exposure remains within the therapeutic window for a sufficient duration without causing excessive accumulation or off-target toxicity. Pharmacodynamic endpoints, such as inhibition of inflammatory cytokines, suppression of MMPs and ADAMTS, restoration of anabolic cartilage markers, macrophage polarization, pain relief, and structural cartilage preservation, should be correlated with time-resolved intra-articular drug concentrations rather than assessed only at terminal time points. Establishing exposure–response relationships, minimum effective concentrations, dosing intervals, and safety margins will be essential for comparing different platforms, optimizing repeat-administration regimens, and improving the clinical translatability of exosomes, microparticles, nanoparticles, and hydrogels for osteoarthritis therapy. Immune responses represent another critical safety issue. Natural exosomes are often described as low-immunogenic vesicles; however, their immunological behavior depends strongly on donor source, cell culture conditions, isolation method, surface proteins, and cargo composition. Allogeneic or engineered exosomes may carry immunomodulatory proteins, nucleic acids, or targeting ligands that activate complement pathways, alter macrophage phenotypes, or induce unexpected cytokine release. Biomimetic nanoparticles coated with cell membranes also require careful evaluation because membrane-derived antigens may trigger immune recognition, particularly after repeated injections. In addition, surface modifications such as PEGylation, targeting peptides, cationic coatings, or gene-editing components may improve targeting efficiency but may also introduce new immunogenic or inflammatory risks. Repeated intra-articular injection safety should therefore be regarded as a core translational requirement rather than a secondary endpoint. OA is a chronic disease, and clinically useful delivery platforms may require multiple administrations over months or years. Repeated dosing studies should evaluate joint swelling, pain behavior, synovial inflammation, cartilage integrity, subchondral bone remodeling, systemic hematology and serum biochemistry, anti-carrier antibody production, complement activation, and cytokine profiles. Particular attention should be paid to cumulative toxicity, accelerated clearance after repeated exposure, immune sensitization, and possible interference with normal cartilage repair. Future studies should incorporate standardized long-term safety protocols, clinically relevant large-animal models, sex- and age-related variability, and head-to-head comparisons between natural and engineered carriers. Only through comprehensive assessment of biodistribution, accumulation, degradation, immunogenicity, and repeated-dose toxicity can these advanced delivery systems be safely advanced toward durable OA therapy. Taken together, the comparative profiles summarized in Table 3 indicate that no single delivery platform is universally optimal for osteoarthritis therapy. Exosomes offer biological sophistication, nanoparticles provide deep tissue penetration and high engineerability, microparticles enable durable depot-like release, and hydrogels integrate sustained delivery with structural support. Rational integration of these platforms, such as exosome-loaded hydrogels, nanoparticle-embedded hydrogels, or microparticle-based sustained-release composites, may provide synergistic therapeutic benefits. With continued progress in material design, targeting strategies, manufacturing standardization, and rigorous clinical evaluation, advanced drug delivery systems hold great potential to transform osteoarthritis treatment from short-term symptom control toward precise, durable, and regenerative joint therapy.