Recent advances in nanomaterial-based precision medicine for orthotopic tumor therapy.

Introduction
Cancer remains a leading cause of mortality worldwide, responsible for approximately one in six deaths and accounting for nearly 10 million fatalities annually [1]. According to the GLOBOCAN 2022 report from the International Agency for Research on Cancer, there were an estimated 18.73 million new cancer cases and 9.67 million cancer-related deaths globally in that year [2]. Among newly diagnosed cases, lung cancer was the most prevalent (2.48 million cases), followed by breast (2.31 million), colorectal (1.93 million), prostate (1.47 million), and gastric cancers (0.97 million). Lung cancer also ranked as the primary cause of cancer-related mortality, with 1.82 million deaths, followed by colorectal, liver, gastric, and breast cancers. Alarmingly, the global cancer burden is projected to increase by 77% and reach approximately 35 million cases by 2050.
Conventional treatment modalities, including surgery, chemotherapy, and radiotherapy, remain the cornerstone of cancer management. Chemotherapeutic agents such as DNA alkylating agents, antimetabolites, and mitotic inhibitors have demonstrated clinical utility, as has radiotherapy, which induces DNA damage via reactive oxygen species (ROS) [3–8]. However, these approaches are frequently associated with substantial limitations. Systemic toxicity, off-target effects, and multidrug resistance often lead to poor prognosis and suboptimal survival outcomes [9, 10]. These persistent drawbacks underscore the urgent need for novel, precise, and less invasive therapeutic strategies.
The advent of nanotechnology has opened new frontiers in oncology. Nanomedicine-based approaches, such as photodynamic therapy (PDT) [11], chemodynamic therapy (CDT) [12, 13], and sonodynamic therapy (SDT) [14, 15], have demonstrated promising anticancer efficacy by leveraging ROS-induced tumor cell apoptosis while offering improved selectivity [16, 17]. Nonetheless, most nanotherapeutic strategies are still evaluated in subcutaneous tumor models, which poorly recapitulate the complex organ-specific microenvironments of human cancers [18–20]. Conventional subcutaneous models often fail to predict clinical outcomes due to the absence of tissue-specific barriers. As highlighted by Christopher J. Osgood et al. [21], the absence of lung-specific barriers (e.g., mucus and the air–blood barrier), together with the lack of the intricate tumor microenvironment (TME) characteristic of lung carcinoma, removes many therapeutic challenges encountered in situ. This fundamental limitation extends to other aggressive malignancies and necessitates a decisive shift toward orthotopic model-guided nanomedicine development.
Orthotopic tumor models preserve crucial physiological and pathological features, including local tissue barriers, metastatic patterns, and immune heterogeneity, thereby making them superior platforms for translational research. Their relevance is particularly evident in malignancies such as glioblastoma multiforme (GBM), where the blood–brain barrier (BBB) significantly impedes drug delivery [22–24]. Similar challenges are observed in oral cancer [25], where mucosal integrity and saliva dynamics hinder local therapy, and in thyroid carcinoma [26], which requires ultra-precise interventions due to its proximity to critical anatomical structures such as the trachea and recurrent laryngeal nerves. Furthermore, cancers of the bladder, cervix, and prostate exhibit distinct physiological barriers such as urinary washout, stromal density, and capsular confinement [27]. These site-specific constraints, along with unique metastatic behaviors such as bone tropism in prostate cancer or peritoneal dissemination in gastric cancer, further complicate treatment design.
To address these challenges, this review advocates a paradigm shift in which nanomaterial design is explicitly informed by the unique biological characteristics of the orthotopic TME. This “orthotopic TME–informed design principle” ensures that nanoplatforms are engineered not only to overcome physical and biological barriers but also to balance tumor eradication with functional preservation of healthy tissues. For example, Wang et al. [28] utilized peptide-modified liposomes to enhance BBB penetration in glioma, while CDT has shown liver-targeting potential by leveraging the liver’s naturally high iron content and metabolic activity [29].

In this review, we systematically summarize recent advances in precision nanomedicine for twelve clinically aggressive malignancies, including GBM, oral, thyroid, pancreatic, gastric, liver, lung, breast, colorectal, bladder, cervical, and prostate cancers (Scheme 1). To reinforce our central thesis, each section follows a consistent “Barrier-to-Solution” analytical framework. We first concisely outline the defining orthotopic barriers of each malignancy and then evaluate the corresponding nanomaterial solutions. Using this logic, we examine how nanocarriers are tailored to penetrate organ-specific membranes, target metastatic niches, and respond to TME-specific stimuli. Finally, we discuss translational challenges such as manufacturing scalability and biocompatibility and propose a strategic roadmap for clinical implementation of orthotopic-model-driven nanomedicine.

Glioblastoma multiforme
GBM is an aggressively malignant CNS tumor and remains among the most therapeutically refractory cancers [30]. Despite advances in surgery, immunotherapy, and tumor-treating fields, outcomes remain poor. Recent 2024–2025 statistics indicate that the global 5-year relative survival under standard first-line chemoradiotherapy is ~ 5.0% (95% CI: 4.2–5.8%) [2]. In the United States, although the 5-year survival for all malignant CNS tumors is ~ 30%, GBM-specific survival decreases to ~ 6.9% [31]. In China, brain/CNS tumors show an ASMR of 2.51 per 100,000 in 2022, corresponding to ~ 56,600 annual deaths [32]. These consistently unfavorable outcomes highlight the need for delivery strategies that explicitly address intracranial physiology and the GBM microenvironment [33, 34].
A central translational bottleneck arises from mechanistic mismatch between subcutaneous and orthotopic models. Subcutaneous xenografts grow in relatively well-perfused connective tissue and therefore underrepresent brain-specific constraints. In contrast, orthotopic GBM models reproduce three barriers that dominate treatment failure: (i) the BBB, which excludes the vast majority of small molecules and essentially all large biologics [35]; (ii) a hypoxic, acidic TME that activates HIF-1α–associated resistance programs and P-gp-mediated efflux, reinforcing an MDR cascade; and (iii) diffuse neural invasion along white matter tracts that frustrates margin-based local control [36]. In addition, orthotopic settings capture CNS-specific immune heterogeneity—particularly the prominence of immunosuppressive M2-like microglia and tumor–astrocyte crosstalk—features that are muted or absent in ectopic tumors. Accordingly, advanced nanomedicines increasingly need an orthotopic TME–informed design framework (Table 1), where BBB traversal, high intracranial interstitial pressure, and immune context are treated as first-order design variables rather than afterthoughts.
Ultrasound (US) is widely recognized as a noninvasive physical stimulus capable of transiently increasing BBB permeability through cavitation-induced disruption of endothelial tight junctions, thereby facilitating intracranial drug delivery [37]. Importantly, prior studies have demonstrated that US-mediated BBB opening is reversible and does not cause detectable neurobehavioral abnormalities or structural damage to brain tissue when appropriately controlled [38]. Building on this foundation, Cai et al. [39] developed a biomimetic nanosonosensitizer system (MDNPs) to achieve US-enhanced chemotherapy and SDT against orthotopic GBM (Fig. 1A). MDNPs were constructed using a biodegradable, pH-responsive polyglutamic acid (PGA) core loaded with doxorubicin (DOX), which serves as both a chemotherapeutic agent and a sonosensitizer. The nanoparticles were further camouflaged with human U87 glioblastoma cell membranes, enabling homologous targeting. This membrane-coating strategy allows MDNPs to evade immune clearance, prolong systemic circulation, and preferentially recognize and accumulate in GBM tissue via self-recognition between tumor cell membrane proteins. Following intravenous administration, US irradiation induces cavitation effects that transiently loosen tight junctions of cerebral vascular endothelial cells, enabling MDNPs to cross the BBB and reach intracranial tumor sites. Once within the TME, acidic conditions (pH 5–6) trigger protonation and structural destabilization of the PGA backbone, resulting in rapid intracellular release of DOX from endo-lysosomal compartments (Fig. 1B and C). Released DOX subsequently translocates to the nucleus, intercalates with DNA, and initiates apoptosis via conventional chemotherapeutic pathways. In parallel, US irradiation activates DOX as a sonosensitizer, leading to ROS generation, predominantly singlet oxygen (1O2) and hydroxyl radicals (•OH) (Fig. 1D). These ROS not only induce oxidative damage and apoptosis but also modulate drug resistance pathways. Specifically, elevated ROS suppress heat shock factor-1 (HSF-1), thereby downregulating resistance-associated proteins including mutant p53 and P-gp encoded by MDR1 (Fig. 1E). Through this mechanism, MDNPs inhibit drug efflux, prevent apoptosis escape, and restore chemosensitivity. In vivo near-infrared (NIR) fluorescence imaging confirmed efficient MDNP accumulation in the brains of orthotopic GBM-bearing mice after US treatment, with fluorescence intensity increasing over time, indicating enhanced BBB penetration and tumor retention (Fig. 1F). Therapeutically, MDNPs plus US reduced tumor growth to approximately 20.55% of that in the control group and achieved an 80% survival rate by day 20 post-treatment. These results demonstrate synergy between SDT and chemotherapy and highlight the capacity of US-assisted biomimetic nanoplatforms to improve drug delivery efficiency and therapeutic response in GBM. Notably, efficacy was achieved with a low DOX dose (1 mg/kg), substantially lower than doses commonly employed in oxygen-enhanced chemotherapy or chemotherapy combined with PDT (~ 2.5 mg/kg). This reduced dosage, together with the noninvasive nature of US, resulted in minimal systemic toxicity and negligible cardiotoxicity, allowing repeated treatment without observable damage to major organs. Despite these promising outcomes, clinical translation remains limited by scalability constraints: extraction and coating of U87 cell membranes are complex and labor-intensive, and pilot-scale production showed a 25–30% reduction in yield compared with laboratory-scale synthesis. Addressing these manufacturing limitations will be essential for future large-scale production and clinical application.
One major challenge of photothermal therapy (PTT) in the brain is the potential for nonspecific thermal injury to normal neurons [40]. To address this issue, Zhang et al. [41] designed an intelligent nanomachine, G@IT-R, comprising RVG29 targeting peptides, Ir nanoperoxidases, TMB (photothermal precursor), and Gd2O3 (autophagy inhibitor) (Fig. 1G). A key feature of this system is its high targeting specificity, aiming to minimize effects on normal brain tissue. The RVG29 peptide binds nicotinic acetylcholine receptors (nAChRs), which are widely distributed on glioma cells and brain microvascular endothelial cells, enabling G@IT-R to cross the BBB and accumulate in tumor tissue (Fig. 1H). Upon entering the TME, the acidic environment and elevated H2O2 activate the peroxidase-like (POD) activity of Ir nanoperoxidases, converting TMB into the photothermal agent oxTMB (Fig. 1I). Under 1064 nm laser excitation, oxTMB generates localized hyperthermia. In normal cells, the weaker acidity prevents TMB-to-oxTMB conversion; meanwhile, Ir nanoperoxidases exhibit catalase (CAT)-like activity and scavenge treatment-induced ROS, thereby limiting oxidative stress in healthy cells. Additionally, Gd2O3 serves as an autophagy inhibitor to overcome tumor cell repair mechanisms. By inhibiting autophagosome–lysosome fusion, Gd2O3 enhances tumor sensitivity to PTT, and blockade of autophagic flux was confirmed by upregulation of autophagy markers such as p62 and LC3-II (Fig. 1J). Twelve hours after treatment, in vivo fluorescence imaging revealed substantial accumulation of G@IT-R in glioma tissue with comparatively low signal in normal brain. Combined with NIR irradiation, G@IT-R significantly inhibited tumor growth, achieving a 21.5% tumor suppression rate (Fig. 1K). These results demonstrate synergy between PTT and autophagy inhibition and highlight the potential of differential PTT strategies to enhance therapeutic efficacy (Fig. 1L). Notably, although G@IT-R shows promising therapeutic effects, large-scale translation may be challenged by its multicomponent formulation and batch-to-batch reproducibility, which warrant further optimization.
The immunosuppressive TME of GBM, characterized by a high proportion of M2-polarized microglia, is poorly captured by conventional subcutaneous tumor models [42]. To address this limitation, Wang et al. [36] developed a microglia-mimetic gene delivery system, termed BM@miR-nanosponge, consisting of a self-assembled PLGA–PEG polymeric core loaded with multiple anti-miRNAs and cloaked with BV2 microglial cell membranes (BM). This biomimetic design enables the nanosponge to exploit intrinsic microglial homing properties within the CNS, thereby improving compatibility with the orthotopic GBM immune landscape. Mechanistically, the BM coating preserves key microglial membrane proteins (CX3CR1, CSF1R, and TMEM119) that interact with corresponding ligands (CX3CL1 and CSF1) highly expressed in GBM tissues. Through these receptor–ligand interactions, BM@miR-nanosponge efficiently crosses the BBB and selectively accumulates within the TME. Following tumor localization, the nanosponge releases its PLGA–PEG-based anti-miRNA core, functioning as a molecular “sponge” to simultaneously sequester four oncogenic miRNAs—miR-9, miR-21, miR-215, and miR-221—each contributing to GBM progression via distinct yet interconnected pathways. Specifically, miR-9 promotes angiogenesis; miR-21 drives invasion, apoptosis resistance, and immunosuppressive signaling; miR-215 enhances migration and hypoxia adaptation; and miR-221 regulates cell-cycle progression and survival. Their concurrent neutralization enables the nanosponge to overcome compensatory mechanisms that often limit single-miRNA therapies. At the signaling level, suppression of these miRNAs restores PTEN and inhibits downstream PI3K/AKT and ERK1/2 pathways, reducing proliferation and migration. In parallel, downregulation of miR-9 and miR-215 decreases HIF-1α and VEGF expression, attenuating angiogenesis and partially relieving hypoxia within the TME. Importantly, in an immune-competent orthotopic GL261 GBM model, BM@miR-nanosponge not only inhibited tumor growth but also reprogrammed microglial polarization from an M2-like tumor-supportive phenotype toward an M1-like pro-inflammatory state. This immune modulation translated into a meaningful survival benefit, extending median survival to approximately 30 days, exceeding that achieved with temozolomide in the same model. Key risks include long-term stability in vivo and potential off-target miRNA perturbation in normal brain cells. Future optimization should prioritize controlled release, CNS-restricted distribution, and comprehensive transcriptomic safety profiling.

Oral cancer
Oral cancer, a prevalent malignancy of the head and neck region, remains a significant global health challenge, with the five-year relative survival rate stagnating at approximately 50% [43]. This poor prognosis is primarily driven by delayed clinical diagnosis, aggressive local invasiveness, and a lack of highly effective targeted therapies. According to GLOBOCAN 2022 estimates, lip and oral cavity cancers accounted for 389,846 new cases and 188,438 deaths annually, representing roughly 2.0% of all new cancer cases and 1.9% of total cancer-related mortality globally [2]. Longitudinal data indicate a consistent upward trajectory in global disease burden: new cases rose from 377,713 in 2020 to 422,000 in 2021, while annual deaths increased from 177,757 to 208,000 over the same period [44].
Survival outcomes are critically dependent on stage at diagnosis. The five-year survival rate for localized oral cancer is as high as 88.4%, yet it decreases to 69.4% for regional disease and drops to 36.9% in cases with distant metastasis [45]. Despite these disparities, only 26% of patients are diagnosed at a localized stage [46]. This diagnostic gap underscores the need for nanotherapeutic systems engineered under an “orthotopic TME–informed design principle.” In the complex oral cavity environment, the efficacy of nanomaterials is dictated by their ability to navigate three defining orthotopic barriers: (i) a dense oral mucosal layer that hampers deep tissue penetration; (ii) rapid saliva-mediated drug clearance (the “washout” effect); and (iii) the requirement for structural stability in a persistently moist microenvironment [47–49]. Furthermore, the orthotopic TME poses additional challenges, including localized hypoxia, acidity, and immune evasion, which are often absent in simplified ectopic models.
The transition toward orthotopic-informed design is driven by fundamental limitations of subcutaneous models. Unlike subcutaneous platforms, where tumors reside in simplified and well-perfused connective tissues, orthotopic oral cancer models recapitulate organ-specific physiological features and fluid dynamics. Mechanistically, orthotopic models enable evaluation of mucoadhesive and penetrative properties of nanocarriers under continuous saliva flow [50], a critical factor because subcutaneous settings often overestimate drug residence time. Moreover, orthotopic models better reflect immune heterogeneity of the oral cavity and the clinical necessity of preserving adjacent functional tissues such as salivary glands. While subcutaneous models facilitate initial toxicity screening, they fail to simulate metastatic niches and the mucosal–tumor crosstalk characteristic of oral malignancies. Therefore, orthotopic models are essential to ensure nanomedicine strategies are tailored to physiological demands of the oral cavity, thereby improving translational success (Table 2).

Radiotherapy is a central modality in oral squamous cell carcinoma (OSCC); however, its efficacy is substantially constrained by tumor hypoxia, a hallmark of solid tumors that renders cancer cells approximately threefold more resistant to ionizing radiation than normoxic counterparts [51]. Because indiscriminate radiation dose escalation can cause severe damage to surrounding normal tissues, strategies that locally reverse hypoxia and enhance radiosensitivity are of considerable clinical interest. In this context, Liu et al. [52] developed a biomineralized nanoparticle radiosensitizer, CaO2–HSA, inspired by the structural and pharmacokinetic advantages of the clinically approved albumin-based nanodrug Abraxane. CaO2–HSA consists of calcium peroxide nanocrystals stabilized and coated by native human serum albumin (HSA) through a biomineralization-induced self-assembly process. HSA serves two functions: it improves colloidal stability and biocompatibility while preserving protein-shell bioactivity, and it facilitates preferential tumor accumulation through the enhanced permeability and retention (EPR) effect and albumin receptor–mediated transport pathways (including gp60- and SPARC-associated mechanisms) that are commonly upregulated in OSCC tissues (Fig. 2A). Following intravenous administration, this delivery enables efficient accumulation of CaO2–HSA nanoparticles at orthotopic tongue-base tumors within several hours. After endocytic internalization, the CaO2 core reacts with intracellular water to generate molecular oxygen (2CaO2 + 2H2O → 2Ca(OH)2 + O2↑), thereby increasing local oxygen availability and alleviating hypoxia (Fig. 2B). This oxygenation restores oxygen-dependent fixation of radiation-induced DNA damage, markedly enhancing radiotherapy efficacy. In an orthotopic OSCC model, CaO2–HSA treatment increased intratumoral oxygen levels by 3.01-fold compared with controls, translating into significantly elevated tumor cell apoptosis under clinically relevant irradiation doses of 7.5–8 Gy (Fig. 2C). Beyond oxygen generation, the acidic TME accelerates CaO2 decomposition, leading to release of H2O2 and Ca2+ (Fig. 2D). Elevated intracellular Ca2+ induces calcium overload, disrupts mitochondrial membrane potential, and impairs energy homeostasis, thereby triggering mitochondrial dysfunction and apoptosis (Fig. 2E). Concurrently, H2O2 contributes to oxidative stress and further sensitizes tumor cells to radiation-induced damage. Collectively, these synergistic mechanisms underpin the radiosensitizing effect of CaO2–HSA observed in vivo. After intravenous administration, CaO2–HSA demonstrated robust tumor targeting, with substantial accumulation at tongue-base tumors within 4 h (Fig. 2F). This biodistribution guided selection of an 8 Gy radiation dose at 4 h post-injection. The CaO2–HSA group exhibited 3.01-fold higher intratumoral oxygen levels than controls, accompanied by enhanced tumor suppression and superior survival (Fig. 2G). Despite promising performance, translational challenges remain. In particular, scalability of the biomineralization process warrants further optimization, and batch-to-batch variations in HSA coating uniformity may affect formulation consistency and reproducibility. Addressing these manufacturing considerations will be important for clinical advancement.
To overcome saliva-induced clearance and insufficient local retention, injectable hydrogels have emerged as an effective platform for sustained intratumoral therapy. Gu and colleagues developed a TME-responsive radiosensitizing nanocomposite hydrogel, Ta@PVP-DAA, for orthotopic OSCC treatment [53]. This system integrates polyvinylpyrrolidone-modified tantalum nanoparticles (Ta@PVP NPs) with a dopamine-functionalized alginate hydrogel matrix, enabling localized retention, enhanced radiosensitization, and minimal off-target exposure. After intratumoral administration, the alginate component responds to elevated divalent cations, particularly Ca2+ and Mg2+, characteristic of the OSCC microenvironment. This ionic interaction induces rapid in situ crosslinking, transforming the formulation from a sol to a stable gel and anchoring the nanocomposite within tumor tissue. In parallel, catechol groups confer mussel-inspired adhesive properties, enabling strong hydrogen and covalent bonding with extracellular matrix (ECM) components and tumor cell surfaces. This dual gelation–adhesion mechanism immobilizes Ta@PVP nanoparticles and prevents extrusion under high interstitial pressure, a common limitation in oral tumors. Once retained, Ta@PVP NPs function as efficient radiosensitizers. Owing to tantalum’s high atomic number, these nanoparticles exhibit strong X-ray absorption and energy deposition capacity (4.30 cm2/g at 100 keV), surpassing that of the clinically approved radiosensitizer HfO2. Under X-ray irradiation, deposited energy is converted into secondary high-energy electrons, leading to amplified ROS generation. The resulting oxidative stress induces extensive DNA double-strand breaks, calcium overload, mitochondrial dysfunction, and ultimately apoptosis. In addition to physical dose enhancement, Ta@PVP nanoparticles display high photothermal conversion efficiency (~ 44.83%). Localized hyperthermia alleviates tumor hypoxia, a major contributor to radioresistance, and further sensitizes cells to radiation-induced damage. In an orthotopic OSCC model, this synergistic radiotherapeutic effect resulted in pronounced tumor growth inhibition, with substantial suppression by day 14. Importantly, biosafety of Ta@PVP-DAA was rigorously evaluated: histological analyses of adjacent normal tissues, including oral mucosa, skin, and salivary glands, showed no detectable pathological damage, highlighting the advantage of localized gel retention in minimizing systemic toxicity. Nevertheless, a potential limitation is reliance on local divalent cation concentrations for gel formation; variability in ionic composition across OSCC subtypes or disease stages may influence gelation efficiency and therapeutic consistency, warranting further investigation.
Mitochondrial metabolic plasticity is a major barrier to effective therapy in OSCC [54, 55]. To achieve spatiotemporally controlled metabolic intervention, Ning et al. [56] designed a dual pH- and photo-responsive “Energy NanoLock” nanocomplex consisting of cyclic RGD–modified carboxymethyl chitosan (CyclicRC; pI = 6.7) encapsulating indocyanine green (ICG) and KLA peptide–functionalized gold nanoparticles (IK-AuNPs) (Fig. 2H). Multivalent cyclic RGD promotes active accumulation in OSCC and tumor neovasculature via integrin αvβ3 recognition (Fig. 2I). NanoLock remains colloidally stable at pH 7.4, whereas the mildly acidic TME (pH = 6.5) induces CyclicRC charge reversal and electrostatically driven disassembly, releasing smaller CyclicRC and cationic IK-AuNPs (Fig. 2J). This microenvironment-triggered “size/charge switch” supports deeper tumor penetration and enhances cellular uptake. Therapeutically, the two released modules act in parallel to block exogenous and endogenous energy supply. CyclicRC competitively antagonizes integrin signaling, suppressing angiogenesis and restricting nutrient/oxygen delivery. Meanwhile, IK-AuNPs deliver the amphipathic α-helical KLA peptide to mitochondria, destabilizing mitochondrial membranes and reducing membrane potential (Fig. 2K), thereby impairing respiration and ATP production. Upon 808 nm NIR irradiation, ICG-mediated photothermal heating further aggravates mitochondrial collapse, promotes cytochrome c release, attenuates HSP-70/90 upregulation, and activates caspase-dependent apoptosis (Fig. 2L). Intravenously administered Energy NanoLock exhibited progressive tumor accumulation, as evidenced by time-dependent fluorescence intensification at oral tumor sites (Fig. 2M). In vivo, tumor-localized photothermal elevation (50 ℃) (Fig. 2N) and mitochondrial disruption cooperatively achieved near-complete tumor ablation in orthotopic OSCC within 24 days while preserving oral function and exhibiting minimal systemic toxicity. Remaining translational considerations include protocol standardization for dual-responsive systems and potential low-grade vascular off-target interactions from RGD–integrin binding that warrant long-term safety evaluation.

Thyroid carcinoma
Anaplastic thyroid carcinoma (ATC) is an exceptionally rare yet highly aggressive malignancy. Although it accounts for less than 2% of thyroid cancer cases globally, it is responsible for the vast majority of thyroid cancer-related deaths. The clinical outlook for ATC remains dismal due to rapid progression, extensive local invasion, and a high propensity for systemic metastasis [57]. Median overall survival (OS) is typically limited to 4–6 months, with a one-year mortality rate approaching 100% and a five-year relative survival rate of only 3.6% [58]. According to GLOBOCAN 2022 data, there were 821,173 new thyroid cancer cases and 47,485 deaths globally in 2022 [2]. The disproportionate fatality of ATC, despite its low incidence, underscores the urgent need for advanced therapeutic interventions.
The formidable challenge of treating ATC is rooted in its anatomical location and microenvironmental architecture. To achieve successful clinical translation, nanomaterial development should adhere to an orthotopic TME–informed design principle. Within the thyroid parenchyma, therapeutic efficacy is obstructed by three defining orthotopic barriers: (i) dysfunctional tumor vasculature that induces hypoxia and fuels radioresistance; (ii) a dense ECM that creates a physical bottleneck for nanomedicine penetration; and (iii) a highly immunosuppressive TME that facilitates immune evasion [57, 59]. Furthermore, because ATC arises adjacent to vital structures such as the trachea and recurrent laryngeal nerve, nanoplatforms must be engineered with high precision to prevent collateral damage during intensive therapy.
A critical appraisal of preclinical methodology indicates that conventional subcutaneous tumor models are inadequate for recapitulating clinical hurdles of ATC. Subcutaneous models lack the thyroid’s specialized anatomical adjacency and fail to simulate intricate TME dynamics such as organ-specific fluid pressures and stromal interactions. While ectopic models may show promising drug accumulation in simplified environments, they can overestimate therapeutic efficacy by not accounting for the dense ECM and hypoxic barriers present in situ. In contrast, orthotopic ATC models provide a more rigorous framework to evaluate whether a nanoplatform can navigate thyroid-specific barriers and achieve targeted tumor eradication while preserving functionally indispensable adjacent organs (Table 3).
ATC is characterized by aberrantly high lactate dehydrogenase (LDH) activity, which drives excessive lactate accumulation in the TME and epigenetically enhances nucleotide excision repair (NER), thereby conferring strong resistance to DNA-damaging therapies [60–62]. To simultaneously disrupt glycolytic metabolism and DNA repair, Ge et al. [63] developed a pH-responsive, nucleus-targeted platinum nanocluster system (Pt@TAT/sPEG) that integrates tumor-selective activation with intracellular precision therapy (Fig. 3A). Pt@TAT/sPEG features a platinum nanocluster core shielded by a pH-sensitive PEG layer that remains stable under physiological conditions but dissociates in the mildly acidic TME. This pH-triggered deshielding exposes the TAT nuclear localization peptide, enabling efficient cellular uptake and subsequent nuclear accumulation via electrostatic interactions with negatively charged cellular and nuclear membranes (Fig. 3B and C). Following nuclear localization, X-ray irradiation activates platinum nanoclusters to generate ROS (Fig. 3D) and release Pt ions, resulting in localized oxidative stress and Pt–DNA adduct formation. These combined insults induce extensive DNA damage while suppressing NER by reducing assembly of the ERCC1–XPF endonuclease complex. Beyond genotoxic effects, Pt@TAT/sPEG interferes with tumor glycolysis: platinum nanoclusters interact with LDH, impair catalytic activity, and decrease substrate affinity, leading to reduced lactate production, disruption of NAD+/NADH balance (Fig. 3E), and ATP depletion. The resulting metabolic stress further weakens lactate-driven DNA repair signaling, reinforcing DNA damage accumulation and promoting caspase-dependent apoptosis. In vitro, this dual metabolic–genotoxic intervention produced markedly greater tumor cell killing than cisplatin or non–nucleus-targeted controls. In vivo, Pt@TAT/sPEG showed substantial tumor accumulation (Fig. 3F) and superior tumor growth inhibition, with efficacy more pronounced in orthotopic ATC models than in subcutaneous tumors, underscoring the importance of complex TME interactions in determining treatment outcomes (Fig. 3G). Despite these promising results, potential off-target interactions of the highly cationic TAT peptide with normal tissues and variability in Pt ion release efficiency may pose challenges for clinical translation. Further optimization of peptide exposure and manufacturing consistency will be essential to advance this nucleus-targeted metabolic nanotherapy.
PDT provides a spatially confined antitumor strategy by activating photosensitizers with NIR light, thereby generating cytotoxic ROS within the irradiated region and reducing collateral damage to surrounding structures such as the trachea and recurrent laryngeal nerves [64]. To improve tumor selectivity and therapeutic performance in ATC, Jonathan C. Irish and colleagues developed PEG-free, biomimetic ultrasmall porphyrin–lipid nanoparticles (PLPs, < 30 nm) [65]. PLPs are constructed from densely packed porphyrin–phospholipid conjugates co-assembled with 1,2-dimyristoyl-sn-glycero-3-phosphocholine, cholesteryl oleate, and R4F ApoA-1 mimetic peptides, collectively enhancing colloidal stability and tumor affinity. Their sub-30 nm size facilitates extravasation and penetration through dense thyroid tumor ECM via the EPR effect. Following systemic administration, PLPs preferentially accumulate in orthotopic ATC lesions while exhibiting limited distribution in adjacent normal tissues, including muscle and trachea. After cellular internalization, porphyrin–lipid components act as photosensitizers, and NIR irradiation triggers ROS generation, inducing oxidative damage and apoptosis. In murine orthotopic thyroid cancer and rabbit thyroid tumor models, PLPs produced markedly stronger intratumoral fluorescence signals than surrounding tissues, confirming effective tumor targeting. Correspondingly, NIR-activated PLP treatment resulted in significant tumor volume reduction, with complete regression observed in a subset of animals during longitudinal follow-up. These findings highlight the value of orthotopic models, where tumor-specific accumulation and localized light activation better reflect clinical anatomy than subcutaneous models. Nevertheless, PLP-mediated PDT is inherently constrained by limited NIR penetration depth, which may reduce effectiveness in bulky or deeply infiltrative ATC lesions frequently encountered in advanced disease.
The BRAFV600E mutation is a dominant oncogenic driver in ATC that sustains aberrant MAPK signaling and promotes rapid tumor growth, invasion, and metastatic dissemination [66]. To silence this driver, Shi et al. [67] developed an NIR fluorescent theranostic nanoplatform for systemic delivery of BRAF-targeting siRNA (siBRAF). The platform is based on a hydrophobic NIR-emissive conjugated polymer core assembled with amphiphilic cationic lipids for siRNA complexation and a PEGylated shell to prolong blood circulation, yielding nanoparticles of approximately 50 nm in diameter. After intravenous administration, PEGylated nanoparticles exhibit extended circulation and preferential tumor accumulation via the EPR effect, which is pronounced in highly vascularized ATC lesions. Intrinsic NIR fluorescence enables noninvasive monitoring of biodistribution, achieving tumor-to-background ratios exceeding 5:1 at 24 h post-injection. Following cellular uptake primarily through clathrin-mediated endocytosis and macropinocytosis, intracellular pH and ionic strength gradients destabilize lipid–siRNA interactions, promoting endosomal escape and cytosolic release of siBRAF. This delivery suppresses BRAF expression and attenuates downstream MEK and ERK phosphorylation without altering total protein levels. Functionally, BRAF silencing inhibits ATC cell proliferation, migration, and invasion in vitro. In vivo, systemic administration of siBRAF-loaded nanoparticles yields an approximately 60% reduction in BRAF protein levels in tumor tissues and significantly delays tumor growth. Notably, therapeutic efficacy is substantially enhanced in orthotopic ATC models compared with subcutaneous tumors, reflecting more realistic vascular architecture and the thyroid TME. In these orthotopic settings, siBRAF treatment produces an approximately threefold reduction in primary tumor size and a pronounced decrease in pulmonary micrometastases, while maintaining good biocompatibility and minimal systemic toxicity. Despite these promising outcomes, the platform shows moderate siRNA encapsulation efficiency (~ 50%), and reliance on EPR-mediated accumulation may contribute to interpatient variability, particularly in tumors with heterogeneous vascular permeability. These limitations highlight the need to optimize loading efficiency and develop patient stratification strategies to maximize the therapeutic potential of NIR-guided siRNA nanomedicine in ATC.

Pancreatic cancer
Pancreatic cancer, predominantly pancreatic ductal adenocarcinoma (PDAC), which accounts for approximately 95% of cases, remains one of the most lethal malignancies of the digestive system [68]. Characterized by aggressive metastasis and profound therapeutic resistance, it consistently ranks as the seventh leading cause of cancer-related mortality worldwide. According to GLOBOCAN estimates, pancreatic cancer caused approximately 466,003 deaths in 2020 and 467,409 deaths in 2022 [2, 44]. The disease burden exhibits marked geographic disparities, with Asia and Europe reporting the highest incidence rates, accounting for 47.1% and 28.3% of global cases, respectively. In the United States, 67,440 new cases and 51,980 deaths are projected for 2025, making pancreatic cancer the third leading cause of cancer-related death [31]. Similarly, China faces a substantial burden, with 118,700 new cases and 106,300 deaths recorded in 2022, corresponding to crude incidence and mortality rates of 8.41 and 7.53 per 100,000, respectively [32].
The prognosis of pancreatic cancer remains extremely poor. While the five-year relative survival rate has reached 11%–13% in high-income countries, it remains below 5% in underdeveloped regions. Pancreatic cancer is projected to become the second leading cause of cancer-related death by 2040. This dismal survival is largely attributable to the insidious onset of symptoms, which leads to late-stage diagnosis and renders only a small minority of patients eligible for curative surgery. To improve clinical outcomes, nanomedicine development should transition toward an orthotopic TME–informed design principle. Therapeutic failure in PDAC is primarily dictated by three defining orthotopic barriers: (i) dense desmoplasia (fibrous connective tissue), which elevates interstitial fluid pressure (IFP) and forms a physical bottleneck for drug delivery; (ii) hypovascularity and aberrant vessels, which foster hypoxia and limit perfusion of agents such as gemcitabine (GEM); and (iii) a profoundly immunosuppressive TME dominated by myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs) [69]. In addition, PDAC exhibits a strong propensity for perineural invasion, contributing to early dissemination and severe pain.
A fundamental mismatch exists between conventional preclinical models and the clinical reality of PDAC. Subcutaneous models, which typically display loose stroma and simplified vasculature, fail to recapitulate the dense desmoplastic response and metabolic rigidity of the pancreatic niche. Consequently, subcutaneous models often overestimate drug penetration and therapeutic efficacy because they do not reflect the physical resistance imposed by the fibrotic ECM. By neglecting organ-specific anatomical constraints and fluid dynamics, simplified models cannot predict the clinical challenge of delivering therapeutics into the poorly perfused, high-pressure core of pancreatic tumors [70, 71]. Therefore, evaluation of nanoplatforms in orthotopic settings is indispensable for assessing their ability to penetrate desmoplastic barriers and reprogram the immunosuppressive TME (Table 4).
PDAC is characterized by a pronounced desmoplastic stroma that severely restricts drug penetration and contributes to the limited clinical benefit of GEM-based chemotherapy [72], with objective response rates remaining below 10% despite combination regimens. To overcome these intertwined physical and biological barriers, Liang et al. [73] engineered a small-molecule, self-assembling nanodrug (ATRA-GEM-NO-PTP) that integrates tumor-directed delivery, microenvironment modulation, and enhanced intratumoral transport (Fig. 4A). This prodrug architecture incorporates three functional modules: GEM as the cytotoxic payload, all-trans retinoic acid (ATRA) as a stromal-modulating component, and a nitrate-based nitric oxide (NO) donor, interconnected through a glutathione (GSH)-responsive disulfide linkage. Surface presentation of the PTP peptide (KTLLPTP) promotes association with plectin-1, which is highly expressed on PDAC cells, thereby enriching tumor accumulation beyond passive EPR-driven delivery. After systemic administration, the nanodrug preferentially accumulates in tumor tissue, where elevated intracellular GSH triggers nanostructure disassembly and initiates the sequential release of ATRA, NO, and GEM (Fig. 4B and C). Mechanistically, released ATRA primarily acts on pancreatic stellate cells (PSCs), reversing their activated phenotype and reducing ECM deposition (e.g., collagen I and fibronectin). This stromal “softening” alleviates the dense fibrotic barrier typical of PDAC and improves interstitial transport. In parallel, GSH-mediated activation of the nitrate moiety sustains NO generation within the TME (Fig. 4D), which enhances penetration through two complementary effects: (i) inducing vasodilation and partial vascular normalization to improve perfusion, and (ii) activating matrix metalloproteinases (MMPs) that further degrade ECM components (Fig. 4E). Continuous NO release can additionally provide a gas-driven propulsion effect, facilitating deeper GEM transport through the remodeled stroma into poorly perfused tumor regions. Consistent with these mechanisms, in orthotopic PDAC models—where desmoplasia and aberrant vasculature more closely resemble the clinical setting—ATRA-GEM-NO-PTP produced the weakest tumor bioluminescence signals and near-complete suppression of primary tumor growth after 21 days of treatment (Fig. 4F and G). Tumor volumes were reduced to < 5% of those in saline-treated controls, markedly outperforming free GEM and nontargeted nanoformulations. By contrast, in subcutaneous models with relatively sparse stroma, the benefit of ATRA-mediated stromal remodeling was less pronounced, underscoring the model-dependent therapeutic gain and highlighting the importance of orthotopic systems for evaluating stroma-targeted nanomedicines. Nevertheless, several limitations remain. Because drug release relies on GSH-responsive kinetics, interpatient and intratumoral redox heterogeneity may introduce variability in activation and efficacy. In addition, the chemical complexity of a multicomponent, self-assembling prodrug may complicate large-scale synthesis and reproducible manufacturing. Collectively, ATRA-GEM-NO-PTP provides a representative example of rational nanodrug design that integrates tumor enrichment, microenvironment remodeling, and penetration enhancement to address the intrinsic delivery barriers of desmoplastic pancreatic cancer.

Addressing the dense stromal barrier in PDAC requires delivery systems capable of both active stromal traversal and microenvironment remodeling [74]. Sun and colleagues developed a multifunctional nanoplatform, GemC18/Gal@CPLNO, to coordinate transport enhancement with therapeutic reprogramming [75]. The nanoparticles are assembled from a stearic acid–conjugated GEM prodrug (GemC18), the TGF-β pathway inhibitor galunisertib, a collagen-binding peptide (CBP), and pH-responsive amphiphilic polymers incorporating enamine N-oxide groups. Following systemic administration, CBP promotes preferential retention within collagen-rich tumor stroma, enriching local exposure at the fibrotic barrier. At outer stromal layers, enamine N-oxide moieties facilitate adsorptive-mediated transcytosis (AMT) across stromal cells, enabling the nanoparticles to traverse otherwise poorly permeable fibrotic matrices. Upon reaching hypoxic tumor regions, N-oxide groups are reduced to tertiary amines, generating reactive intermediates that consume intracellular GSH via Michael-addition reactions. This redox modulation elevates oxidative stress and sensitizes tumor cells to chemotherapy. Concurrently, protonation of amine groups under acidic conditions increases the positive surface charge, enhancing cellular uptake and supporting deeper penetration. After intracellular delivery, galunisertib suppresses TGF-β/SMAD signaling, attenuating the activation of cancer-associated fibroblasts (CAFs) and PSCs, thereby reducing collagen and hyaluronic acid (HA) secretion, alleviating fibrosis, and improving perfusion. In parallel, immunosuppressive features of the TME are partially reversed, as reflected by enhanced M1-like macrophage polarization and increased activation of cytotoxic CD8⁺ T cells. Meanwhile, GemC18 is converted intracellularly to GEM, inhibiting DNA synthesis and inducing apoptosis. Together, these biochemical, stromal, and immunological effects yield a synergistic antitumor response. In orthotopic PDAC models that more faithfully recapitulate human stromal architecture, GemC18/Gal@CPLNO achieves markedly improved intratumoral penetration and stronger tumor growth inhibition than in subcutaneous models, resulting in substantial regression and prolonged survival. Nevertheless, therapeutic performance may remain sensitive to stromal heterogeneity among PDAC subtypes, and formulation complexity (co-loading a prodrug and a small-molecule inhibitor) may challenge reproducible manufacturing and consistent release profiles during clinical translation.
US-responsive nanotherapeutic systems provide a spatiotemporal control strategy for deep-seated and highly desmoplastic tumors such as PDAC [76]. Ye et al. [77] developed an integrated theranostic nanoplatform, IR&ZnPc@LNP-NO, which combines chemotherapy (irinotecan, IR), SDT (zinc phthalocyanine, ZnPc), and NO-mediated microenvironment modulation for orthotopic PDAC treatment (Fig. 4H). After systemic administration, the lipid nanovesicles (~ 120 nm) progressively enrich within orthotopic pancreatic tumors via EPR-associated accumulation, as confirmed by in vivo fluorescence imaging (Fig. 4M). Upon tumor localization, low-intensity US induces a rapid and controllable structural transformation, shrinking the vesicles from ~ 120 nm to ~ 40 nm (Fig. 4I and J). This size contraction is critical for improving transport through the dense ECM and elevated IFP characteristic of orthotopic PDAC, thereby enabling deeper penetration and a more uniform intratumoral distribution. Simultaneously, US-triggered cleavage of surface-conjugated S-nitrosothiol (SNO) groups releases bioactive NO, which activates MMP-2, promotes ECM degradation, and partially normalizes aberrant tumor vasculature. These coordinated effects reduce IFP from 10.88 mmHg to 6.16 mmHg, enhancing intratumoral perfusion and drug transport. Following improved tumor infiltration, high-intensity US activates ZnPc to generate cytotoxic ROS (Fig. 4K), inducing apoptosis through lipid peroxidation and oxidative DNA damage, while concurrently promoting irinotecan release from the lipid core (Fig. 4L). Beyond direct cytotoxicity, NO-mediated remodeling also reshapes local immunity, increasing CD8⁺ cytotoxic T lymphocyte infiltration (12.3% → 28.7%) and reducing regulatory T cells. Moreover, the combined regimen induces immunogenic cell death (ICD), as evidenced by calreticulin exposure, ATP/HMGB1 release, and enhanced dendritic cell maturation (CD11c⁺CD80⁺CD86⁺: 11.3% → 49.8%). These events collectively establish a more pro-inflammatory antitumor microenvironment. Therapeutically, IR&ZnPc@LNP-NO combined with dual-intensity US and αPD-L1 achieved the strongest tumor suppression, as indicated by minimal bioluminescence signal (Fig. 4N), and prolonged median survival to 73 days in orthotopic PDAC models. Notably, efficacy was superior in orthotopic tumors compared with subcutaneous models, highlighting the importance of ECM remodeling and microenvironment normalization in clinically relevant PDAC settings. Nevertheless, clinical translation may be constrained by the operational complexity of dual-intensity US protocols and the short biological half-life of SNO-based NO donors, necessitating precise synchronization between administration and irradiation. Further optimization of US parameters and NO-release kinetics will be essential to maximize reproducibility and expand the therapeutic window.

Gastric cancer
Gastric cancer is the fifth most common malignancy and the fourth leading cause of cancer-related mortality globally. In 2022, approximately 968,350 new cases and 659,853 deaths were recorded, accounting for 4.9% of all new cancer diagnoses and 6.8% of total cancer deaths [2]. The global age-standardized incidence rate (ASIR) is 12.8 per 100,000 in males and 6.0 per 100,000 in females. The cumulative risk of developing gastric cancer before age 75 is estimated at 1.53% for males and 0.67% for females. Although overall incidence has declined in recent decades due to improved food preservation and Helicobacter pylori eradication, an upward trend has been reported among younger age groups in some low-incidence populations [31].
Eastern Asia bears the highest disease burden. In China, an estimated 358,700 new cases and 260,400 deaths occurred in 2022, corresponding to crude incidence and mortality rates of 25.41 and 18.44 per 100,000, respectively [32]. Survival outcomes depend strongly on stage at diagnosis. Although five-year survival exceeds 70% in countries with robust screening programs (e.g., Japan and South Korea), the global average remains ~ 20%–30% because many patients present at advanced stages due to subtle early symptoms.
In situ gastric cancer is confined to the gastric mucosal epithelium and is governed by three defining orthotopic barriers. First, the intragastric environment is highly acidic (pH 1.5–3.5) and contains proteolytic enzymes (e.g., pepsin) that can degrade therapeutic agents [78]. Second, the thick mucus layer and dense mucosal epithelium hinder nanomaterial diffusion toward the basement membrane [79]. Third, the gastric TME is characterized by hypoxia, chronic inflammation, and stromal fibrosis. In addition, gastric lymphatic involvement provides early routes for dissemination. Conventional subcutaneous models exist in a relatively simple interstitial environment lacking gastric chemical stressors, organ-specific fluid dynamics, and lymphatic drainage; thus, they often overestimate drug stability and penetration. Therefore, evaluating nanostrategies in orthotopic models is indispensable to predict performance under hostile intragastric conditions while preserving normal gastric function (Table 5).
Metal–nucleic acid frameworks (MNFs) are highly programmable platforms that combine the structural stability of metal coordination with the sequence-specific functionality of nucleic acids [80]. In this context, Zhang et al. [81] developed an aptamer–DNAzyme MNF (IRF/H-GDz/Ca) for HER2-positive orthotopic gastric tumors (Fig. 5A). The framework is assembled from Ca2+, HER2-recognizing aptamers, GLUT-1–specific DNAzymes (GDz), and interferon regulatory factor-1 (IRF-1) protein, enabling coordinated tumor recognition and stimulus-responsive payload deployment. After systemic administration, aptamer–HER2 interactions favor tumor enrichment, while the acidic TME promotes pH-responsive framework disassembly, synchronizing the release of GDz and IRF-1 (Fig. 5B and C). Intracellularly, Ca2+ activates the DNAzyme to cleave GLUT-1 mRNA, reducing GLUT-1 expression and thereby perturbing glucose uptake and redox homeostasis. This metabolic disruption weakens the GSH/ROS buffering capacity, aggravates oxidative stress, and exacerbates mitochondrial injury and oxidative DNA damage. In parallel, IRF-1 downregulates RAD51 (Fig. 5D), impairing homologous recombination repair and limiting the resolution of therapy-induced DNA lesions (Fig. 5E). The combined pressure—ROS-amplified genotoxic stress together with compromised DNA repair—drives robust apoptotic cell death. Compared with subcutaneous tumors, orthotopic models exhibited more efficient triggered release due to the native acidic niche. Consequently, IRF/H-GDz/Ca achieved ~ 90% tumor growth inhibition after 21 days of treatment in the orthotopic model (Fig. 5F and G), without detectable systemic toxicity or body-weight loss. Despite these encouraging results, clinical translation may be limited by the scalability of long-oligonucleotide manufacturing and by interpatient heterogeneity in HER2 expression, which could affect tumor enrichment and response consistency.
Cisplatin (CDDP) remains a first-line agent for gastric cancer (GC), yet its clinical utility is constrained by poor tumor selectivity, insufficient intragastric exposure after systemic dosing, and dose-limiting systemic toxicity, particularly nephrotoxicity [82, 83]. Although oral administration is conceptually attractive for luminally accessible gastric lesions, free CDDP is rapidly compromised by acidic degradation, mucosal clearance, and gastrointestinal toxicity. To address these barriers, Chen et al. [84] developed orally administered platinum-chelated cyclic peptide nanotubes (Pt-CP NTs) for orthotopic GC therapy (Fig. 5H). Pt-CP NTs self-assemble from disulfide-cyclized peptides (CP) and cisplatin via Pt–S coordination, forming high–aspect ratio nanotubes that resist gastric acid and enzymatic degradation, thereby protecting the payload during gastrointestinal transit (Fig. 5I). Their elongated morphology and mucus-penetrating behavior facilitate traversal of the gastric mucus and epithelial barriers, enabling prolonged mucosal retention at orthotopic gastric tumors (Fig. 5J). Following mucosal access, tumor cells preferentially internalize the nanotubes via macropinocytosis, a route often enhanced under tumor metabolic stress. Inside tumor cells, elevated intracellular GSH cleaves CP disulfide bonds, triggering nanotube disassembly and on-demand cisplatin release. Released cisplatin translocates to the nucleus and forms DNA cross-links, disrupting replication and inducing apoptosis. Beyond serving as a carrier scaffold, CP also suppresses VEGF signaling, limiting tumor neovascularization and restricting nutrient supply (Fig. 5K). This dual mechanism—direct DNA cross-linking cytotoxicity plus anti-angiogenic deprivation—supports a synergistic antitumor effect. Pharmacokinetic and biodistribution analyses further showed that oral Pt-CP NTs increased gastric tumor exposure while reducing systemic accumulation; notably, the gastric-to-kidney platinum ratio increased by approximately threefold relative to intravenous cisplatin, consistent with reduced nephrotoxicity. In orthotopic GC models, oral Pt-CP NTs produced substantial tumor growth suppression and prolonged survival, achieving efficacy comparable to intravenous cisplatin but with an improved safety profile (Fig. 5L and M). The diminished performance in subcutaneous tumors underscores the importance of the native gastric mucosal interface for realizing oral, locally acting nanotherapy. Remaining translational considerations include interindividual variability in gastric pH, enzyme activity, and mucus composition, all of which may influence nanotube stability and release kinetics.
Because gastric tumors are protected by multilayer mucosal and stromal barriers, physical strategies that transiently enhance penetration can substantially improve intratumoral drug delivery [85]. US-mediated microbubble disruption is one such approach, leveraging cavitation and sonoporation to increase local permeability and promote deep tissue transport. Lin et al. [86] developed a nanoparticle–microbubble conjugate system (EGCG/DOX–HFP NP-c-MBs) for orthotopic gastric cancer therapy, integrating vascular localization, US-triggered release, and enhanced intracellular delivery. In this platform, epigallocatechin gallate (EGCG) and DOX are co-encapsulated in negatively charged nanoparticles composed of HA, fucoidan (FC), and PEGylated gelatin (PG), which are electrostatically tethered to positively charged phospholipid microbubbles filled with perfluorocarbon gas. After intravenous administration, preferential localization at tumor-associated vasculature is supported by both EPR-associated margination and ligand-mediated adhesion: FC binds P-selectin on activated endothelium, promoting microbubble retention, while HA enables subsequent CD44 engagement after nanoparticle release. Upon focused US irradiation, microbubbles oscillate and rupture, generating localized cavitation that induces transient membrane permeabilization (sonoporation) and simultaneously liberates the conjugated nanoparticles. This US-triggered event improves transvascular transport, enhances intratumoral penetration, and increases cellular uptake; quantitatively, intracellular accumulation of EGCG and DOX increased by > 2-fold compared with non-US conditions. In three-dimensional tumor spheroids, US further promoted more uniform penetration across multiple layers, consistent with improved convective transport beyond passive diffusion. After endocytosis, pH-responsive destabilization within endo/lysosomal compartments accelerates intracellular release. DOX then exerts canonical genotoxicity (DNA damage and cell-cycle arrest), whereas EGCG modulates oxidative stress and survival pathways to reinforce apoptotic signaling, together yielding synergistic cytotoxicity. In vivo, EGCG/DOX–HFP NP-c-MBs plus US markedly reduced tumor bioluminescence, suppressed tumor growth, and increased necrosis in orthotopic models, accompanied by reduced Ki-67 staining and increased M30 positivity, while sparing major organs—supporting a favorable safety profile. The consistent superiority in orthotopic tumors over subcutaneous xenografts highlights the value of evaluating US-enabled penetration under clinically relevant gastric stromal and vascular constraints. Nonetheless, translation may be limited by the short circulation half-life of microbubbles and the need to tightly control US parameters to avoid unintended mucosal injury during repeated treatments; further optimization of microbubble stability and irradiation protocols will be essential.

Liver cancer
Liver cancer represents a major global health burden and remains a critical clinical challenge. Hepatocellular carcinoma (HCC) accounts for ~ 75%–85% of primary liver cancers and ranks as the sixth most common malignancy and the third leading cause of cancer-related mortality worldwide [2]. In 2022, an estimated 865,269 new cases and 757,948 deaths were reported globally, with males exhibiting two- to threefold higher incidence and mortality than females. The global five-year survival rate remains low (~ 15%–20%), largely due to late diagnosis, recurrence, and underlying cirrhosis. China bears a particularly heavy burden, with ~ 367,700 new cases and 316,500 deaths in 2022 [32]. Although the ASIR has gradually declined from 2000 to 2018 (AAPC: − 2.3% in males; − 2.4% in females), the absolute burden remains substantial due to chronic hepatitis B virus infection, aflatoxin exposure, and population aging. Current treatments—including surgical resection, radiofrequency ablation (RFA), and transarterial chemoembolization (TACE)—provide limited survival benefits [87]. Resection is often infeasible due to metastasis and cirrhosis, and RFA may yield uneven heat distribution and suboptimal ablation [88]. Consequently, resistance and recurrence remain major obstacles.

Treatment resistance in in situ HCC is strongly shaped by the orthotopic TME. Three defining barriers characterize the hepatic setting: (i) the hepatic sinusoidal barrier, in which fenestrated endothelium and the Space of Disse regulate extravasation and restrict nanoparticle diffusion [89]; (ii) dense fibrosis/cirrhosis, which elevates IFP and forms a mechanical barrier to penetration [90]; and (iii) a highly immunosuppressive milieu sustained by Kupffer cells, tumor-associated macrophages (TAMs), and MDSCs, which collectively dampen antitumor immunity [91]. In addition, the dual hepatic blood supply (hepatic artery and portal vein) creates complex hemodynamics that restrict uniform drug distribution—features absent in ectopic models.
A major limitation in preclinical HCC research is overreliance on subcutaneous xenografts that do not reproduce liver physiology. Subcutaneous tumors form in loose connective tissue with simplified vasculature, lacking the sinusoidal network, biliary system, and fibrotic stroma intrinsic to the liver. As a result, ectopic models tend to overestimate delivery and efficacy, contributing to poor translation. Accurately modeling in situ liver cancer is therefore essential to capture the metabolic, immune, and hemodynamic constraints that define orthotopic HCC (Table 6).
Given the liver’s central role in iron storage and metabolism, CDT that exploits its iron-rich milieu represents a rational strategy for HCC treatment [92]. Bu et al. [93] developed polyethylene glycol–modified zinc peroxide nanoparticles (ZnO2@PEG NPs) to potentiate Fenton-based CDT by leveraging key physicochemical features of the hepatic TME (Fig. 6A). Following intravenous administration, ZnO2@PEG NPs accumulates preferentially in hepatic tumors through passive enrichment and hepatic retention. In the mildly acidic and CAT-deficient tumor milieu (pH ≈ 6.5), the nanoparticles rapidly disintegrate to release Zn2+ and H2O2 (Fig. 6B and C), initiating a self-sustained redox cascade. The liberated H2O2 reacts with intrahepatic Fe2+ to drive Fenton/Fenton-like reactions, producing cytotoxic •OH that inflict oxidative damage on lipids, proteins, and DNA (Fig. 6D–F). Concurrently, Zn2+ perturbs mitochondrial integrity and homeostasis, amplifying oxidative stress and promoting apoptotic and necrotic cell death. Release kinetics were strongly pH-dependent, with 93% Zn2+ release at pH 5.4 versus 13% at pH 7.4, confirming acidity-triggered disassembly. Consistently, intracellular ROS levels in HCC-LM3 cells increased up to 15-fold relative to controls, as indicated by DCFH-DA fluorescence. Cell viability decreased to 10.5 ± 2.1% at pH 5.4 while remaining above 95% under physiological pH, supporting tumor-selective activation. Flow cytometry revealed pronounced early and late apoptosis under acidic conditions, consistent with the combined effects of ROS and zinc overload. In contrast, normal hepatocytes (AML12) and macrophages (RAW 264.7) exhibited negligible cytotoxicity, attributable to intact CAT-mediated ROS clearance. In orthotopic HCC models, ZnO2@PEG NPs induced substantial regression, with MRI/PET showing reduced tumor burden and metabolic activity. Histology confirmed iron enrichment, ROS generation, reduced proliferation, and increased apoptosis, with minimal off-target toxicity. Therapeutic benefit was more pronounced in orthotopic than subcutaneous models, consistent with organ-specific “chemical factory” conditions (iron-rich, mildly acidic, low CAT) that enhance CDT. While ZnO2@PEG NPs elicited moderate tumor inhibition in subcutaneous models, responses were markedly amplified in orthotopic tumors, consistent with the iron-enriched, mildly acidic, and low-CAT microenvironment of hepatic lesions that favors ROS production and Fenton efficiency. This contrast underscores the organ-specific dependency of CDT and the necessity of orthotopic validation for translational assessment. Nevertheless, interpatient variability in hepatic iron stores and antioxidant capacity may influence outcomes, and long-term biosafety evaluations as well as optimization of dosage, circulation time, and surface modification remain essential for clinical translation.
PDT requires nanocarriers that can navigate the hepatic milieu while limiting off-target activation. Tang et al. [94] engineered dual-sensitive supramolecular nanocarriers (BM-Ce6 NPs) for orthotopic HCC ablation (Fig. 6G). These nanoparticles are built on a poly(ethylene glycol)-block-polylysine (PEG-b-PLys) scaffold conjugated with benzimidazole (BM), enabling supramolecular loading of (2-hydroxypropyl)-β-cyclodextrin (β-CD-HP) and the photosensitizer chlorin e6 (Ce6). After tumor accumulation, acidic microenvironmental conditions protonate BM moieties, inducing charge reversal and dissociation of β-CD-HP, which enhances cellular uptake by HCC cells (Fig. 6H). Following internalization, elevated intracellular GSH acts as a second trigger by cleaving disulfide linkages between Ce6 and the β-CD carrier (Fig. 6I and J), enabling timely liberation of photoactive Ce6. Upon 660 nm laser irradiation, Ce6 undergoes photoexcitation and transfers energy to molecular oxygen, generating high levels of ROS (Fig. 6K) that damage membranes, proteins, and DNA, culminating in apoptosis and/or necrotic death. In orthotopic tumors, more efficient GSH-triggered release also coincides with partial GSH depletion, which further weakens antioxidant defenses; in an iron-rich hepatic context, this redox imbalance can amplify oxidative stress and may contribute to ferroptosis-like lipid peroxidation cascades [91, 103]. Comparative evaluation in orthotopic versus subcutaneous models showed that the more faithful hepatic microenvironment supported more efficient cargo activation and stronger ROS production, resulting in markedly greater tumor suppression in the orthotopic setting (Fig. 6L). Nonetheless, PDT efficacy remains constrained by the limited penetration depth of 660 nm light, restricting applicability to superficial lesions or tumors accessible by intraoperative or interventional light delivery. Moreover, heterogeneous acidity within bulky HCC masses may cause incomplete activation in necrotic cores, motivating further optimization to improve intratumoral distribution and activation uniformity.
The hepatic TME is profoundly immunosuppressive and often exhibits limited natural killer (NK) cell infiltration. To overcome this barrier, Shi et al. [95] developed a magnetothermal-regulated nanoplatform, MNPs@PEI-FA/pDNA (MPFD), to activate NK immunity in situ (Fig. 7A). The system comprises superparamagnetic ZnCoFe2O4@ZnMnFe2O4 nanoparticles (MNPs) functionalized with polyethyleneimine (PEI) and folic acid (FA), carrying a plasmid encoding an HSP70-promoter–driven IL-2 construct (HSP70–IL-2–EGFP). After systemic administration, FA–folate receptor interactions favor tumor cell association and receptor-mediated endocytosis. PEI then promotes endosomal escape through proton-sponge–associated osmotic swelling, releasing pDNA into the cytoplasm (Fig. 7B). Upon exposure to an alternating magnetic field (AMF), the MNP core generates mild hyperthermia (~ 40 ℃), which activates the HSP70 promoter and drives localized IL-2 expression (Fig. 7C). Secreted IL-2 engages IL-2 receptors on resident and infiltrating NK cells, triggering downstream signaling (e.g., JAK/STAT and stress-activated kinase pathways) that supports NK proliferation and enhances cytotoxic effector function, reflected by increased NKG2D expression and elevated granzyme B, perforin, and TNF-α production (Fig. 7D). In orthotopic HCC models, MPFD plus AMF markedly increased intratumoral IL-2/EGFP transcripts within 48 h (Fig. 7E) and reduced tumor burden by day 14 (Fig. 7F). The orthotopic setting is mechanistically important because it more faithfully reproduces hepatic immune suppression and reticuloendothelial sequestration, whereas subcutaneous models can provide an artificially permissive immune context that overestimates NK activation. Translation will require improved control of local AMF exposure and a clearer understanding of the long-term fate of magnetic nanoparticles in the liver’s reticuloendothelial system.
Advanced HCC frequently demands multimodal intervention to address physical barriers and immunosuppression. Dong et al. [96] developed a multifunctional nano–ultrasound contrast agent, ATO/PFH NPs@Au-cRGD, for synergistic chemo-PTT and enhanced immunotherapy (Fig. 7G). The platform consists of a lipid core encapsulating arsenic trioxide (ATO) and perfluorohexane (PFH), coated with an Au nanoparticle shell functionalized with cRGD peptides. In vivo, cRGD–integrin interactions promote enrichment at αvβ3-expressing tumor cells and neovasculature (Fig. 7H). Upon internalization and NIR irradiation, the Au shell converts light into heat, generating localized hyperthermia (up to ~ 49 ℃) that directly injures tumor cells and promotes ICD, thereby releasing tumor-associated antigens to support immune priming. In parallel, US-targeted microbubble destruction (UTMD) induces acoustic cavitation that facilitates payload release and triggers PFH phase transition, strengthening US contrast for real-time monitoring (Fig. 7K). Released ATO perturbs redox homeostasis by depleting intracellular GSH and inhibiting glutathione peroxidase 4 (GPX4), leading to toxic accumulation of lipid peroxides and ferroptotic cell death (Fig. 7I and J). Orthotopic validation confirmed peak tumor accumulation at ~ 48 h (Fig. 7L), providing a rational window for synchronized external triggering. Combined with anti–PD-L1 therapy, the platform achieved ~ 82.5% tumor suppression and inhibited both primary growth and metastasis. A key limitation is operational: reliance on dual external triggers (NIR and US) complicates clinical workflow and may restrict efficacy in deep-seated lesions unless interventional delivery of light and precise acoustic focusing are available.
Transarterial embolization/chemoembolization (TAE/TACE) is a first-line therapy for intermediate-stage HCC, but recurrence and drug resistance remain common [97, 98]. Cheng et al. [99] developed Mn-GMSs sono-microspheres, integrating gelatin microspheres with oxygen-deficient manganese tungstate (MnWOX) nanodots to combine embolization, SDT, and metalloimmunotherapy. Mechanistically, gelatin microspheres lodge within tumor-feeding arteries to induce ischemia and aggravate hypoxia, thereby starving tumors of oxygen and nutrients. Under US irradiation, MnWOX nanodots act as sonosensitizers and amplify cavitation-associated ROS generation, abruptly disrupting redox homeostasis to trigger apoptosis and ICD. ICD is supported by calreticulin exposure and HMGB1 release, which function as immunostimulatory danger-associated signals that facilitate antigen uptake and presentation. Beyond ROS-mediated cytotoxicity, Mn2+ release activates the cGAS–STING axis within the TME, increasing STING/TBK1/IRF3 signaling and promoting type I interferon production, thereby enhancing dendritic cell maturation (e.g., CD80/CD86 upregulation). When combined with anti–PD-L1 antibodies, this immune activation restores CD8 + T-cell cytotoxicity against residual tumor cells and suppresses distant metastasis. Importantly, orthotopic liver tumor models are essential to evaluate embolic performance because subcutaneous platforms lack hepatic arterial architecture and cannot reproduce the hemodynamic constraints required for transarterial occlusion. Translationally, precise control of microsphere size distribution will be necessary to avoid off-target embolization of healthy vessels, and interindividual variability in gelatin degradation kinetics warrants systematic assessment.

Lung carcinoma
Lung carcinoma, primarily originating from bronchial mucosal epithelium, remains the most lethal malignancy globally. According to GLOBOCAN 2022, lung cancer accounted for ~ 2.48 million new cases (12.4% of all malignancies) and ~ 1.82 million deaths (18.7% of all cancer deaths), making it the leading cause of cancer-related mortality worldwide [2]. The burden shows gender differences: males account for ~ 1.57 million new cases (ASIR 32.1/100,000) and ~ 1.23 million deaths (ASMR 24.8/100,000), whereas females account for ~ 0.91 million new cases and ~ 0.58 million deaths [100]. In China, lung cancer remains the most prevalent malignancy, with 1.06 million new cases and 0.73 million deaths in 2022 (crude incidence and mortality: 75.13 and 51.94 per 100,000) [32]. Notably, in the United States, incidence among women younger than 65 surpassed that of men in 2021 (15.7 vs. 15.4 per 100,000) [31]. Despite advances in radiotherapy and chemotherapy, the global five-year relative survival rate remains ~ 27%, driven by late diagnosis and rapid emergence of resistance/recurrence. Systemic drug delivery is further limited by suboptimal pulmonary distribution and dose-limiting toxicity.
Improving outcomes requires an orthotopic TME–informed design principle. Many nanomedicines fail because they cannot overcome key pulmonary barriers: (i) viscous mucus that traps and clears particles [101]; (ii) the air–blood barrier (ABB) that restricts transport of systemically administered agents into alveolar spaces [21]; and (iii) pulmonary microbiota that may enzymatically degrade payloads [102]. The lung TME also features a specialized immunosuppressive architecture and a propensity for brain metastasis, which are poorly captured in non-pulmonary sites. Subcutaneous xenografts lack the mucus layer and ABB and often overestimate penetration/efficacy while failing to model metastasis. Therefore, orthotopic evaluation is indispensable (Table 7).
To address drug resistance and brain metastasis risk in non-small cell lung cancer (NSCLC), Lin et al. [103] developed an inhalable nanoliposomal system (MPDOLs) to concurrently treat primary lung tumors and brain metastases (Fig. 8A). MPDOLs are engineered to address two orthotopic barriers in NSCLC therapy—efficient deposition within lung lesions after inhalation and effective control of the lung–brain metastatic axis. Structurally, a plasmid encoding IGF2BP3 siRNA together with a brain-targeting tag (RVG) is condensed within a cationic poly(β-amino ester) (PβAE) core, while the lipid shell encapsulates osimertinib (OST). The entire construct is further camouflaged with mesenchymal stem cell (MSC) membranes expressing pulmonary surfactant protein B (SP-B) to enhance pulmonary retention and reduce immune clearance. Mechanistically, SP-B facilitates mucus/tissue penetration and improves deposition within lung parenchyma, promoting enrichment at orthotopic tumor sites following inhalation (Fig. 8B). After cellular internalization, OST inhibits EGFR phosphorylation and downstream signaling, thereby suppressing tumor proliferation, while the IGF2BP3 siRNA cassette silences IGF2BP3 expression to sensitize tumor cells to OST (Fig. 8C). Notably, the RVG-encoding module further enables an “in situ exosome factory” effect: transfected tumor cells generate RVG-decorated exosomes (RVG-EXOs) loaded with IGF2BP3 siRNA, which can cross the BBB and deliver gene-silencing cargo to brain metastases (Fig. 8D). Beyond direct tumor inhibition, MPDOLs remodel the immune microenvironment by increasing CD8 + T-cell and NK-cell infiltration (Fig. 8E) and elevating systemic IFN-γ and TNF-α levels, thereby amplifying antitumor immunity. In vivo inhalation therapy produced marked tumor growth inhibition, with bioluminescence signals substantially reduced by day 28 (Fig. 8F), and MPDOLs-treated mice achieved an 80% survival rate at day 70 post-treatment (Fig. 8G). Despite these promising outcomes, translation may be constrained by the scalability and batch-to-batch reproducibility of MSC membrane coating, as well as variability in exosome production efficiency across heterogeneous patient tumors.
Cancer stem cells (CSCs) and TAMs establish an immunosuppressive niche that promotes lung tumor progression. Shi et al. [104] developed iron-based nanoparticles (Dex@Fe-HMSNs@Ct, DFHC) to induce CSC ferroptosis by leveraging TAM-directed delivery (Fig. 8H). DFHC are iron-doped mesoporous silica nanoparticles loaded with citrate (Ct) and surface-modified with dextran to promote CD206-mediated uptake by TAMs (Fig. 8I). After lysosomal trafficking, acidic conditions accelerate nanoparticle degradation, releasing Fe3+/Fe2+ ions and citrate (Fig. 8J). Citrate chelation stabilizes iron species and facilitates their redistribution from lysosomes to the cytoplasm and surrounding interstitial space, increasing local iron availability in CSC-adjacent regions. Given the high CD44 expression on CSCs, elevated iron exposure promotes CSC iron overload, which drives excessive ROS production via Fenton-type redox cycling (Fig. 8K). In parallel, DFHC suppresses the pentose phosphate pathway (PPP), as reflected by reduced glucose-6-phosphate (G6P) levels (Fig. 8L), thereby depleting NADPH and GSH pools and weakening antioxidant defenses. Consistently, DFHC downregulates GPX4 and induces mitochondrial abnormalities (reduced size, increased membrane density, and outer membrane rupture) (Fig. 8M), culminating in lipid peroxidation–driven ferroptosis. To enhance pulmonary delivery, nebulized administration was employed and significantly increased lung accumulation within 4 h post-nebulization (Fig. 8N). Therapeutically, nebulized DFHC produced substantial tumor suppression and prolonged survival, increasing median survival from 30 days (control) to 63 days, while attenuating body-weight loss. Remaining questions for clinical translation include long-term safety in normal lung tissue, particularly the consequences of repeated iron loading and potential off-target oxidative injury.
To tackle chemoresistance, Li et al. [105] developed an “all-in-one” formulation (PASA/CDDP/MK-2206) for orthotopic lung cancer, integrating cytotoxic chemotherapy with pathway blockade and inflammation control. In this design, PASA is a 5-aminosalicylic acid (5-ASA)–derived nanocarrier that co-loads CDDP and the AKT inhibitor MK-2206, while 5-ASA provides COX-2/PGE2 axis suppression as a functional component of the carrier. Upon exposure to the mildly acidic TME, protonation of PASA carboxyl groups disrupts ionic interactions and accelerates drug release, enabling synchronized delivery of CDDP and MK-2206. Released CDDP damages tumor DNA to inhibit replication and transcription, whereas MK-2206 suppresses AKT1 signaling, reducing survival advantages and limiting CSC enrichment. Concurrently, 5-ASA dampens COX-2/PGE2-driven inflammation, which is associated with immune evasion and CSC persistence, thereby contributing to immunosuppressive-TME remodeling and improved immune surveillance. In the FVBW-17-GFP-luc orthotopic lung cancer model, intravenous administration of PASA/CDDP/MK-2206 significantly reduced tumor bioluminescence, inhibited tumor growth, and extended survival, with the highest survival rate among treatment groups. Nevertheless, pharmaceutical translation will require precise control of multi-agent loading and distinct release kinetics, as well as robust manufacturing standardization to ensure batch consistency.

Breast cancer
Breast cancer is the most frequently diagnosed malignancy among women worldwide. In 2022, approximately 2.30 million new cases were recorded globally (ASIR 46.82 per 100,000), accounting for 11.5% of all new cancer diagnoses and ~ 0.67 million deaths (ASMR 12.65 per 100,000) [2, 100]. Incidence varies geographically, with highest rates in Australia/New Zealand (ASIR 100.3), Northern America (95.1), and Northern Europe (90.8). Eastern Asia remains high-burden (ASIR 37.5). In China, breast cancer is the second most common female malignancy (357,200 new cases in 2022; ASIR 33.04) [32]. In the United States, 316,950 new cases and 42,170 deaths are projected for 2025, with breast cancer comprising ~ 32% of new female cancer diagnoses [31].
While five-year survival has reached ~ 91% globally, regional and socioeconomic disparities persist [100]. In the United States, five-year survival is ~ 93% in White females but ~ 84% in Black females [31], reflecting inequities in access and care. Incidence continues to rise (AAPC: 3.1% in China, 2000–2018; 1.0% in the United States, 2012–2021) [32]. Despite early detection, ~ 20%–30% of patients relapse due to metastatic progression and therapy resistance. Subcutaneous xenografts fail to reproduce breast-specific ECM, hormone-related vascular heterogeneity, and dynamic immune infiltration patterns [106], which strongly influence penetration and immune evasion. Orthotopic models (mammary fat pad implantation) more faithfully mimic the breast TME and metastatic behavior, providing a superior platform for translational nanomedicine evaluation (Table 8).
MiR-200c-3p is a tumor suppressor that restrains invasiveness by inhibiting ZEB1/ZEB2, master drivers of epithelial–mesenchymal transition (EMT) [107]. Clinically, miR-200c is often downregulated in breast cancer stem cells (BCSCs), particularly in resistant/metastatic disease. Eroles et al. [108] developed a mesoporous silica nanoparticle platform (MSN-PEI-miR-200c-HA) for targeted delivery of miR-200c-3p to BCSCs (Fig. 9A). This nanoplatform is designed to overcome two delivery bottlenecks: selective entry into BCSCs and endo/lysosomal sequestration. HA promotes CD44-mediated internalization by BCSCs (Fig. 9B and C), while PEI facilitates endosomal escape through proton-sponge–associated membrane destabilization. After cellular uptake, nanoparticles traffic to lysosomes (Fig. 9D), then release miR-200c-3p into the cytoplasm, leading to ZEB1/2 downregulation (Fig. 9E and F). This gene regulation reverses EMT-associated markers (e.g., N-cadherin, vimentin, fibronectin, and β-catenin), suppresses migration and invasion, and induces G2 cell-cycle arrest. In vivo, intravenous administration of MSN-PEI-miR-200c-HA significantly inhibited tumor growth compared with controls (Fig. 9G). Key translational considerations include PEI-associated systemic toxicity risks and interpatient variability in CD44 expression that may affect targeting consistency.
Triple-negative breast cancer (TNBC) lacks ER/PR/HER2 expression, rendering endocrine and HER2-targeted therapies ineffective [109–111]. Hong et al. [112] developed DOX-loaded vanadium carbide (V2C) MXene nanosheets (V2C-DOX) for combined PTT and chemotherapy (Fig. 9H). Owing to strong NIR absorption, V2C converts NIR irradiation into localized heat, directly injuring tumor cells via hyperthermia (Fig. 9I and J). In parallel, photothermal heating and the acidic, H2O2-rich TME accelerate nanosheet degradation, enabling triggered DOX release and reinforcing apoptosis. Consistent with a synergistic mechanism, V2C-DOX plus NIR irradiation increased pro-apoptotic markers (caspase-3, Bax, and PTEN) and reduced Bcl-2 expression (Fig. 9K). Multi-omics analyses further suggested disruption of amino acid biosynthesis and iron metabolism pathways, potentially increasing vulnerability to oxidative stress. In vivo, V2C-DOX followed by NIR irradiation achieved 100% tumor inhibition without recurrence (Fig. 9L and M). Limitations include restricted NIR penetration for deep lesions and the need for long-term evaluation of vanadium-containing degradation products.
HER2-positive breast cancer (15%–20%) is aggressive and prone to metastasis [113, 114]. Although trastuzumab improves outcomes, ~ 30% of patients develop resistance [115]. One resistance mechanism involves chromosome 17q23 amplification, which can suppress ATM–p53 signaling via WIP1 and activate PI3K/AKT signaling via miR-21, collectively enabling survival under HER2 blockade. To address both axes, Zhang et al. [116] engineered a CO2-generating nanoplatform (in-MW@NP) composed of PLGA, Poloxamer F127, guanidyl-modified chitosan (CG-CO2), and DPPC to co-deliver the WIP1 inhibitor GSK2830371 and anti-miR-21 oligonucleotides (antagomiR21). In acidic endo/lysosomes, CG-CO2 generates CO2 gas, disrupting membranes and facilitating cytosolic delivery. AntagomiR21 relieves miR-21-mediated PTEN suppression to inhibit PI3K/AKT signaling and enhance trastuzumab sensitivity; GSK2830371 inhibits WIP1 to restore the ATM–p53 DNA damage response. Orthotopic models are essential to validate CO2-triggered escape, which depends on acidic and dense stromal conditions often underestimated in subcutaneous sites. Translation may be limited by the need for precise stoichiometric control and consistent loading/release of multiple payloads.

To sensitize tumors to immune checkpoint therapy, Florindo et al. [117] developed biodegradable polylactic acid (PLA) nanovaccines co-delivering tumor-associated antigens (TAAs; α-lactalbumin and KRAS peptides), TLR ligands (CpG-ODN and Poly(I: C)), and TGF-β1–targeting siRNA (siTGF-β1) for orthotopic breast cancer. The formulation incorporates a PLA core and surface elements that support tumor localization and intracellular siRNA delivery, including GlutCs/siRNA complexes and HA for CD44-mediated targeting. After uptake, TME-associated acidity promotes nanoparticle destabilization and coordinated cargo release: siTGF-β1 suppresses TGF-β1 in tumor cells and dendritic cells via RNA interference, while TLR ligands drive dendritic cell maturation and enhance antigen presentation. Antigen-primed CD4 + and CD8 + T cells expand and infiltrate tumors, and combination with a costimulatory agonist antibody (anti-OX40) further amplifies T-cell proliferation, survival, and cytokine production. The resulting cytokine milieu (e.g., IFN-γ, TNF-α, and IL-2) supports broader immune recruitment, including NK cells and macrophages, and helps remodel the suppressive TME. Translational challenges include ensuring siRNA stability, minimizing off-target immune activation, and standardizing complex multi-component manufacturing.

Colon cancer
Colon cancer is a prevalent gastrointestinal malignancy and remains the third most commonly diagnosed cancer and the second leading cause of cancer-related mortality worldwide. According to GLOBOCAN 2022, there were ~ 1.14 million new cases (ASIR 10.69/100,000) and 538,167 deaths (ASMR 4.71/100,000) globally [100]. Males exhibit higher incidence and mortality than females. In China, colon cancer is among the top five malignancies, with 382,430 new cases (crude incidence 26.97/100,000) and 186,810 deaths in 2022 [32]. In the United States, 107,320 new cases and ~ 52,970 deaths are projected for 2025 [31]. Notably, incidence is rising in younger populations; in the United States (2012–2021), AAPC was 0.4% for ages 50–64 and 2.4% for < 50 [31]. Despite multimodal therapy, the global five-year relative survival rate remains ~ 63%, largely due to late diagnosis and early lymphatic spread leading to resistance and recurrence.
Developing effective nanomedicines for colon cancer requires an orthotopic TME–informed design principle. Therapeutic efficacy in situ is constrained by three defining barriers: (i) a thick intestinal mucus layer that limits nanoparticle adhesion and penetration; (ii) intestinal fluid dynamics and peristalsis that shorten local residence time; and (iii) dense ECM that elevates IFP and restricts deep penetration [118, 119]. In addition, the colon TME exhibits a specialized immunosuppressive profile (e.g., M2 macrophages; reduced IFN-β), facilitating immune evasion and metastasis to mesenteric lymph nodes. Subcutaneous xenografts lack the mucus barrier, microbiome–immune crosstalk, and lymphatic hierarchy of the intestinal wall, leading to systematic overestimation of perfusion and failure to model mesenteric spread. Orthotopic validation is therefore indispensable for nanoplatforms targeting mucosal barriers and local immunosuppression (Table 9).
To bypass the limited tissue penetration of light in PDT, Shi et al. [120] developed a piezoelectric nanoparticle system (BaTiO3@Glu NPs) as an in-situ ROS spatiotemporal immune-adjuvant platform for orthotopic colon cancer. BaTiO3@Glu NPs are constructed by covalently linking BaTiO3 to glucan (via Ba–S bonding), enabling both immune-cell targeting and mucosal trafficking. The glucan component promotes uptake by Dectin-1–expressing macrophages, improving delivery into tumor-associated immune compartments, while also facilitating transport across gut-associated lymphoid interfaces (e.g., via M-cell–associated pathways). In the intestinal environment, mechanical forces generated by peristalsis activate the piezoelectric BaTiO3 core, producing local ROS within the TME. The resulting oxidative stress damages biomembranes, proteins, and nucleic acids, inducing tumor cell apoptosis and provoking ICD–associated signals (e.g., CRT exposure and ATP/HMGB1 release). In parallel, glucan functions as an immune adjuvant that promotes dendritic cell maturation and antigen presentation, leading to expansion of tumor-infiltrating CD8 + and CD4 + T cells with enhanced effector function. Macrophage activation further elevates pro-inflammatory cytokines (e.g., TNF-α and IL-12), reinforcing immune recruitment and antitumor activity. Future work should clarify safety under chronic peristalsis-driven activation and standardize activation intensity across variable intestinal motility states.
Oral immunization provides a noninvasive route to activate gut-associated lymphoid tissue (GALT) [121], but conventional oral vaccines are degraded by gastric acid and limited by epithelial barriers [122]. Zhang et al. [123] developed an orally administered engineered nanovaccine, E. coli (AH1-CDA-Co1)@iPDA, for orthotopic colorectal cancer (Fig. 10A). The system employs engineered E. coli strains producing OMVs encoding the tumor antigen AH1, a STING agonist (CDA), and an M-cell–targeting peptide (Co1), encapsulated within a polydopamine (iPDA) shell that protects against gastric acid, prolongs intestinal retention, and improves formulation stability. Upon US irradiation, the engineered bacteria are stimulated to continuously secrete OMVs (Fig. 10B and C), enhancing OMV delivery to intestinal lymphoid structures (Fig. 10D and E). Co1-mediated M-cell targeting promotes epithelial transcytosis, enabling OMVs to access immune cells; CDA activates STING signaling (e.g., TBK1, NF-κB, and IRF3 phosphorylation) and drives IFN-β production (Fig. 10F). Meanwhile, OMVs provide adjuvant cues (e.g., TLR4 activation), promoting dendritic cell maturation and antigen presentation and ultimately expanding AH1-specific CD8 + cytotoxic T lymphocytes. In orthotopic models, oral E. coli (AH1-CDA-Co1)@iPDA plus US markedly inhibited tumor growth, with bioluminescence weakening from day 7 and near-complete suppression by day 21 (Fig. 10G and H). Translational limitations include biosafety control for live engineered bacteria, interindividual variability in gut flora and mucosal barriers, and the need to standardize US exposure for reproducible activation.

To overcome washout and stromal barriers in colorectal cancer, Xiao et al. [124] developed MnOx-based nanomotors (CS-ID@NMs) for chemo–sonodynamic–immunotherapy (Fig. 10I). The platform consists of MnOx cores loaded with ICG–derived mitochondrial sonosensitizers (IDs), sequentially coated with silk fibroin (RSF) and chondroitin sulfate (CS), and formulated within a chitosan/alginate hydrogel. CS promotes CD44-associated tumor enrichment (Fig. 10J), while the RSF layer undergoes microenvironment- and US-responsive destabilization to accelerate ID release. Released IDs accumulate in mitochondria (Fig. 10K), sensitizing cells to US-triggered ROS. Concurrently, MnOx degradation releases Mn2+ to catalyze Fenton-like redox cycling, generating cytotoxic ·OH for CDT and producing O2 that both alleviates hypoxia and enhances sonodynamic 1O2 generation under US irradiation. O2 microbubble formation also contributes to propulsion, improving penetration (Fig. 10L and M). These combined CDT/SDT effects can trigger ferroptosis and ICD, releasing antigens and danger-associated signals that promote dendritic cell maturation and T-cell priming; combination with anti–PD-L1 further sustains T-cell activity and amplifies systemic immunity. After oral administration, CS-ID@NM-hydrogel exhibited the strongest colonic fluorescence at 24 h (Fig. 10N), and the combined regimen (hydrogel + US + anti–PD-L1) significantly suppressed orthotopic tumor growth without inducing body-weight loss (Fig. 10O). Key translational considerations include controlling motor activation intensity, ensuring hydrogel batch consistency, and validating long-term mucosal safety under repeated US exposure.

Bladder cancer
Bladder cancer remains a major global health challenge, notable for frequent recurrence and pronounced sex disparity. In 2022, approximately 614,298 new cases were diagnosed worldwide (ASIR: 5.58 per 100,000), with 220,596 deaths (ASMR: 1.82 per 100,000) [2, 100]. The burden is substantially higher in males (471,293 cases; ASIR: 9.31) than in females (143,005 cases; ASIR: 2.38). In China, 82,260 new cases and 32,980 deaths were reported in 2022 [32]. In the United States, 84,870 new cases and 17,420 deaths are projected for 2025 [31]. Although the global 5-year relative survival remains comparatively favorable (~ 78%), recurrence is still the dominant clinical bottleneck, with ~ 50% of patients relapsing after initial therapy, underscoring the limited durability of current intravesical and systemic regimens.
From a delivery perspective, the bladder is an archetypal organ where anatomy and fluid dynamics dictate efficacy, making an “orthotopic-TME-informed design principle” particularly necessary. Three defining orthotopic barriers collectively compress the therapeutic window: periodic micturition generates hydrodynamic washout that rapidly dilutes and expels instilled agents [125]; the negatively charged glycosaminoglycan (GAG) layer limits stable adhesion and intimate contact between formulations and the urothelium; and the highly impermeable urothelium restricts transurothelial transport into deeper layers, reducing drug access to invasive lesions [126–128]. These physical constraints are further compounded by local immunosuppressive features of the in-situ microenvironment, which can blunt chemo-immunotherapy synergy.
This barrier triad also explains why subcutaneous xenografts often mislead translation. Subcutaneous tumors reside in a relatively static interstitium, lacking urine dilution, filling–emptying cycles, and the GAG/urothelial barrier; drug exposure is therefore artificially prolonged and penetration is overestimated. As a result, systems optimized in ectopic models may appear effective yet fail orthotopically because they cannot sustain residence, resist washout, or maintain adequate effective concentration under continuous hydrodynamic stress (Table 10) [126].
Against this background, strategies that “work” in the bladder typically succeed by either prolonging mucosal residence or actively counteracting washout, thereby increasing the probability of drug–tumor interaction. Yu et al. [129] used Sporopollenin Exine Capsules (SECs) derived from sunflower pollen to carry pirarubicin intravesically. SECs feature a hollow core–shell architecture and a rigid echinate (spiny) exine, which promotes mechanical interlocking with the mucosal surface. This physics-driven anchoring converts transient luminal contact into sustained attachment, allowing the carrier to better withstand micturition-associated flushing and to maintain prolonged exposure at the urothelial interface. Extended residence enables sustained local release of pirarubicin and supports its diffusion into tumor tissue, where the anthracycline can intercalate DNA and interfere with replication/transcription, culminating in apoptosis. Notably, SECs were also reported to exert adjuvant-like immunomodulation, enhancing infiltration of T cells and macrophages within the TME, which may reinforce local immune surveillance. Importantly, these advantages can only be credibly quantified in orthotopic settings that preserve urine flow and mucosal barriers. Potential limitations include mild mucosal irritation from the rigid echinate surface, and the need for rigorous biosafety evaluation of immunogenicity and long-term metabolic fate of pollen-derived sporopollenin.

While intravesical retention addresses one axis of the problem, systemically delivered platforms must additionally contend with hypoxia, which frequently constrains oxidative therapies. Qian et al. [130] engineered a GSH-responsive PTX-SS-HPPH/Pt@RGD-NP (Fig. 11A), combining chemotherapy with PDT. At the tissue and cellular level, the RGD modification promotes preferential association with integrin-rich bladder cancer cells, facilitating internalization (Fig. 11B). Once inside the reductive cytosol, elevated GSH cleaves the disulfide (S–S) linkage, releasing paclitaxel (PTX) and the photosensitizer HPPH in an on-site manner. Upon 660 nm irradiation, HPPH is excited to a higher-energy state and transfers energy/electrons to surrounding molecular oxygen, generating cytotoxic ROS—predominantly 1O2 and related radicals—thereby inducing oxidative damage to membranes, proteins, and nucleic acids (Fig. 11C). Because ROS production in PDT is oxygen-dependent, the co-loaded Pt nanozyme plays a crucial amplification role: it catalyzes endogenous H2O2 decomposition to generate O2, partially relieving hypoxia and sustaining oxygen availability, which in turn boosts ROS yield and PDT cytotoxicity (Fig. 11D). Following intravenous administration in tumor-bearing mice, nanoparticles accumulated at the tumor site with a maximal signal at 24 h (Fig. 11E), and irradiation at this time point produced marked tumor inhibition through coordinated chemotherapy and PDT (Fig. 11F). Taken together, ligand-directed delivery, stimulus-triggered drug release, and hypoxia modulation constitute a coherent strategy to enhance intratumoral efficacy, although clinical translation may be influenced by the limited tissue penetration of 660 nm light and the need to standardize irradiation parameters for reproducibility.
When washout dominates as the governing constraint, an effective solution is to convert passive diffusion into active transport. Nanomotors offer a promising means to counteract urine-driven clearance in intravesical therapy by improving residence time and facilitating penetration across the urothelium and tumor parenchyma [131]. Building on this concept, Wu et al. [132] developed an intravesical self-propelled nanodrug system (DMCU), composed of polydopamine, Mn2+, cyclic GMP–AMP (cGAMP), and urease for bladder cancer immunotherapy (Fig. 11G). After instillation, urease on the particle surface hydrolyzes urea to carbon dioxide and ammonia, generating propulsion that increases mucosal contact and prolongs intravesical retention (Fig. 11H). Following cellular uptake, Mn2+ promotes genotoxic stress and facilitates the accumulation of cytosolic double-stranded DNA, thereby activating cyclic GMP–AMP synthase (cGAS). The resulting cGAMP engages stimulator of interferon genes (STING) and drives phosphorylation of STING, TANK-binding kinase 1 (TBK1), and interferon regulatory factor 3 (IRF3), culminating in downstream innate immune activation (Fig. 11I and J). This cascade promotes dendritic cell maturation and primes cytotoxic T-cell responses, while cGAMP acts synergistically with Mn2+-associated stress to strengthen pathway engagement. In orthotopic models, DMCU achieved the highest fluorescence intensity and the longest bladder retention among comparator groups (Fig. 11K). A 21-day treatment course produced complete survival, the smallest tumor burden, and pronounced suppression of tumor growth when laser irradiation was applied at the optimal retention time window (Fig. 11L and M), consistent with durable intravesical residence and robust antitumor immunity. Despite these encouraging outcomes, translational feasibility will depend on mitigating potential mucosal irritation associated with urease-generated ammonia and establishing safe, standardized irradiation/instillation protocols.

Cervical cancer
Cervical cancer remains a major cause of cancer mortality among women worldwide. In 2022, ~ 662,301 new cases (ASIR: 14.12 per 100,000) and 348,874 deaths (ASMR: 7.08 per 100,000) were reported [2, 100]. China recorded 110,700 new cases and 55,700 deaths in 2022 [32], while the United States is projected to have 13,360 new cases and 4,320 deaths in 2025 [31]. Despite improvements in screening and prevention, local relapse still occurs in a substantial fraction of patients (~ 21.7%) after standard surgery and chemoradiotherapy, often driven by radioresistant clones and protective TME niches [133].
Mechanistically, in situ cervical therapy is constrained by a set of orthotopic barriers that are poorly represented by subcutaneous models. The cervicovaginal mucus (CVM) layer restricts diffusion and retention of locally administered agents; HSP90-mediated thermotolerance dampens hyperthermia/HIFU/PTT efficacy by stabilizing proteins under heat stress; and hypoxic, immunosuppressive stroma supports resistance to chemo-radiotherapy while limiting immune activation [133, 134]. Moreover, pelvic lymphatic drainage provides clinically relevant dissemination routes that ectopic models do not capture. Together, these features explain why subcutaneous tumors—superficial and optically accessible—often overestimate the performance of light-triggered systems and underestimate depth-dependent transport constraints (Table 11).
Given that hypoxia is a central driver of failure by promoting radioresistance and efflux programs (e.g., MDR1) [135, 136], platforms that simultaneously deliver cytotoxics and relieve hypoxia have particular rationale. Lee and coworkers [137] designed hypoxia-relieving hemoglobin nanoclusters for combined chemotherapy and PDT in cervical cancer. Hemoglobin was conjugated with chlorin e6 and biotinylated polyethylene glycol to generate an oxygen-carrying photosensitizer (HPBC), which was subsequently loaded with DOX to afford DOX@HPBC (Fig. 12A). In tumor cells, DOX@HPBC released DOX in a sustained, pH-dependent manner, as increased drug solubility under acidic conditions weakened hydrophobic interactions with the protein scaffold (Fig. 12B). Upon 660 nm irradiation, Ce6 is photoexcited and undergoes intersystem crossing to a long-lived triplet state, which transfers energy/electrons to molecular oxygen to generate ROS (e.g., 1O2 and •OH), thereby producing oxidative injury. Hb contributes a second, synergistic layer by releasing O2 to mitigate hypoxia, which sustains oxygen availability and amplifies ROS generation (Fig. 12C and D). This oxygenation also downregulates HIF-1α and MDR1, reducing drug efflux and re-sensitizing tumor cells to DOX (Fig. 12E). In tumor-bearing mice, DOX@HPBC markedly inhibited tumor growth via chemo–PDT synergy, achieving near-complete suppression in heterotopic models and robust control in orthotopic disease with minimal systemic toxicity (Fig. 12F and G). Collectively, oxygen delivery, biotin receptor–mediated uptake, and stimulus-responsive drug release provide a coherent approach to overcome hypoxia-driven recalcitrance in cervical cancer, although efficacy may depend on intra-tumoral oxygen heterogeneity and light-delivery geometry in clinical settings.
Beyond cytotoxicity, photoimmunotherapy seeks to translate local tumor destruction into systemic immune priming [138, 139]. Chen et al. [140] developed donor–acceptor–donor (A–D–A)-structured nanoaggregates (DNPs@CM) consisting of DTPC-N2F nanoparticles (DNPs) coated with U14 cervical cancer cell membranes (CM) for photoimmunotherapy. The membrane coating provides tumor-associated antigens and promotes homologous cellular recognition, thereby enhancing internalization into cervical cancer cells. Upon entry, DTPC-N2F exhibits strong intramolecular charge transfer; under 980 nm irradiation, absorbed energy is converted into heat via vibrational relaxation to generate PTT. In parallel, favorable HOMO–LUMO distribution facilitates intersystem crossing, converting ground-state oxygen (3O2) into ROS including 1O2, superoxide (·O2−), and ·OH, thereby enabling PDT. This dual PTT/PDT assault induces oxidative stress and apoptotic signaling, as evidenced by increased Bax and cleaved caspase-3 expression and decreased Bcl-2. Importantly, combined phototoxicity triggers ICD, leading to the release of tumor antigens and danger-associated molecular patterns (DAMPs). Surface-exposed CRT and extracellular HMGB1 promote dendritic cell recruitment and maturation, characterized by upregulated CD80 and CD86. Mature dendritic cells present antigens to naïve T cells in lymphoid tissues, driving CD4 + helper T-cell responses and cytokine secretion (e.g., IL-2 and IFN-γ), which in turn support CD8 + T-cell expansion, macrophage activation, and NK-cell cytotoxicity, collectively amplifying systemic antitumor immunity. Notwithstanding its promise, clinical translation may be limited by the complexity of membrane-coating manufacture and the need for standardized, safe intralesional light delivery.

For deep pelvic tumors where optical access is intrinsically limited, energy delivery can be shifted from photons to acoustics. HIFU induces thermal coagulative necrosis but can provoke compensatory HSP90 upregulation, generating thermotolerance and recurrence risk [141]. Wang et al. [142] developed gambogic acid–based coordination polymer nanoparticles (GAZn-PEG NPs), in which the HSP90 inhibitor gambogic acid is coordinated with Zn2+ and encapsulated within a lipid bilayer (DOPA/DPPC/cholesterol/DSPE-mPEG) for combination therapy with HIFU (Fig. 12H). Under HIFU-induced hyperthermia, GAZn-PEG NPs binds to and suppress HSP90, thereby sensitizing tumors to thermal ablation (Fig. 12I). In parallel, Zn2 + cooperates with gambogic acid to potentiate antitumor immunity: Zn2 + supports immune-cell activation, while gambogic acid–induced apoptosis increases tumor antigen availability. Transcriptomic and immunoblot analyses revealed activation of the cGAS–STING axis with increased expression and phosphorylation of pathway components (Fig. 12J), enhanced dendritic cell maturation and antigen presentation (Fig. 12K), and increased infiltration of T cells and NK cells. In orthotopic tumor models, intravenous GAZn-PEG NPs combined with HIFU produced sustained suppression of tumor burden, evidenced by progressive bioluminescence decline (Fig. 12L) and minimal tumor volume at late time points (Fig. 12M and N). Notably, rechallenge experiments suggested immunological memory, supported by expanded central and effector memory T-cell compartments in blood and spleen, indicating durable systemic immunity against recurrence and metastasis. Future translational work should focus on optimizing acoustic parameters, dosing schedules, and safety margins to ensure reproducible synergy between HIFU ablation and immune reprogramming.

Prostate cancer
Prostate cancer is the second most commonly diagnosed malignancy and the fifth leading cause of cancer death in men worldwide. In 2022, ~ 1.47 million new cases (ASIR: 29.42 per 100,000) and 397,430 deaths (ASMR: 7.27 per 100,000) were reported [2, 100]. China recorded 125,600 new cases and 55,700 deaths in 2022, with rapidly rising incidence (AAPC: 7.0% from 2000 to 2018) [32]. In the United States, 313,780 new cases and 35,770 deaths are projected for 2025 [31]. While localized disease has excellent outcomes, metastatic disease remains difficult to control, emphasizing the need for strategies that simultaneously address primary and disseminated niches.
From an orthotopic perspective, three features jointly constrain nanotherapeutic efficacy in the prostate: the prostatic capsule as a rigid fibromuscular barrier that limits intratissue dispersion; AR-driven and CXCR4-linked immunosuppression that sustains a “cold” microenvironment and supports drug efflux programs; and prostate-specific lymphovascular drainage with strong osteotropic metastatic propensity [143, 144]. These constraints are largely absent in subcutaneous xenografts, which fail to reproduce capsular mechanics, glandular secretions, and endocrine–immune coupling, and therefore frequently overpredict therapeutic performance (Table 12).

Antibody-targeted therapy offers high specificity with a favorable safety profile, and the insulin-like growth factor 1 receptor (IGF-1R) is a rational target in prostate cancer because ERG overexpression driven by the TMPRSS2–ERG fusion upregulates IGF-1R and sustains proliferative signaling. Although the anti–IGF-1R antibody AVE1642 showed acceptable tolerability in early clinical studies, antitumor activity was limited and off-target effects curtailed development, motivating strategies that refine receptor engagement and intratumoral pharmacology. María J. Vicent and colleagues [145] addressed these issues by engineering a poly(L-glutamic acid)–conjugated AVE1642 (PGA-AVE1642) via bioresponsive disulfide linkages, thereby tuning the antibody’s bio–nano interface without compromising recognition of the IGF-1R epitope. In ERG-positive prostate cancer models, PGA-AVE1642 displayed higher apparent affinity for IGF-1R, restrained receptor internalization with prolonged membrane residence, and redirected trafficking away from clathrin-dependent uptake. This altered receptor interaction translated into concomitant suppression of Shc-linked MAPK and IRS-1–linked PI3K signaling, in contrast to the narrower pathway modulation observed with native AVE1642. In an orthotopic VCaP model, PGA-AVE1642 improved tumor accumulation and achieved superior control of tumor growth, accompanied by reduced proliferation and attenuated angiogenic remodeling. Collectively, polymer conjugation provides a mechanistically coherent route to enhance IGF-1R blockade by modulating receptor trafficking and downstream signaling while preserving target selectivity.
For prostate tumors that are immunologically “cold,” an effective design objective is to convert tumor cell death into an immunogenic event while simultaneously preventing checkpoint-mediated exhaustion. Zhang et al. [146] reported a microenvironment-dual-responsive nanodrug (TSD@LSN-D) that couple’s epigenetic modulation with PDT and PD-L1 blockade for self-synergistic immunotherapy of prostate cancer (Fig. 13A). The prodrug core (TPRA-SS-DAC) covalently links an aggregation-induced emission photosensitizer to decitabine via a disulfide bond and is encapsulated within pH-responsive nanoparticles composed of poly(ε-caprolactone), poly(β-amino ester), and polyethylene glycol bearing a PD-L1–blocking peptide. In the acidic tumor milieu, poly(β-amino ester) undergoes a phase transition that enhances tumor association and accumulation. After uptake, the reductive cytosol cleaves the disulfide bond to release decitabine (Fig. 13B), demethylating the DFNA5 promoter and upregulating GSDME (Fig. 13C). Upon light irradiation, TPRA generates ROS that activate caspase-3 (Fig. 13D), producing the GSDME N-terminal fragment that perforates the plasma membrane and triggers immunogenic pyroptosis (Fig. 13E). Resultant DAMPs—including surface-exposed CRT, HMGB1, and ATP (Fig. 13F)—promote dendritic cell maturation and antigen presentation, while the peptide ligand blocks PD-1/PD-L1 interactions to sustain cytotoxic T-cell activity. In orthotopic models, intravenous TSD@LSN-D significantly restrained tumor growth (Fig. 13G), demonstrating synergy between pyroptosis-driven immune activation and checkpoint inhibition. Nonetheless, translation will require scalable manufacturing and stringent control of multi-component loading and release kinetics.
Physical-field guidance provides another route to overcome capsular constraints while enabling multi-modal oxidative killing. Jiang et al. [147] developed ZnFe2O4@Pt@PEG-GOx nanoparticles (ZFPG NPs) for prostate cancer, comprising a superparamagnetic ZnFe2O4@Pt core, a PEG intermediate layer, and a glucose oxidase (GOx)-functionalized surface (Fig. 13H). Under an external magnetic field, ZFPG NPs achieve magnetic targeting and enhanced tumor accumulation. After internalization, GOx catalyzes intratumoral glucose oxidation to gluconic acid and H2O2 (Fig. 13I), inducing a starvation effect while elevating intracellular peroxide. The ZnFe2O4 core exhibits POD-like activity, catalyzing Fenton-like reactions with H2O2 to generate·OH (Fig. 13J), thereby damaging lipids, proteins, and DNA and promoting ferroptosis. Concurrent GSH depletion—further exacerbated by intrinsic glutathione oxidase (GSHOx)-like activity—weakens antioxidant defenses and amplifies lipid peroxidation [148]. Under US irradiation, the ZnFe2O4@Pt Schottky heterojunction facilitates rapid electron transfer and improves charge separation, boosting 1O2 generation (Fig. 13K and L). The combined ·OH/1O2 oxidative assault enhances tumor cell killing, while US can improve intratumoral distribution to support more uniform efficacy. Ferroptosis and oxidative injury release DAMPs (e.g., CRT, HMGB1, HSP90, and ATP), which promote antigen-presenting cell activation, dendritic cell maturation, and cytotoxic T-lymphocyte infiltration; the nanoplatform also reduces MDSC abundance to alleviate immunosuppression. In orthotopic prostate cancer models, ZFPG NPs + US produced the weakest tumor fluorescence signal by day 14 compared with controls (Fig. 13M), consistent with potent tumor suppression and extensive necrosis/apoptosis. Key translational considerations include controlling systemic glucose depletion risk, minimizing off-target oxidative injury, and standardizing magnetic/US parameters for reproducible clinical performance.

Conclusions and prospects
In recent years, nanomedicines designed to accommodate tumor heterogeneity have catalyzed notable advances in precision oncology. This review systematically evaluates nanotherapeutic strategies across 12 major malignancies by consolidating nanomaterial design principles, animal model fidelity, and therapeutic outcomes into disease-specific summaries (Tables 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 and 12). By integrating enabling technologies such as targeted delivery and intelligent stimulus responsiveness, nanosystems have increasingly enabled more precise intervention across these cancer types [149]. To date, dozens of nanomedicines have entered clinical or late-stage evaluation, and clinically established nanoformulations—exemplified by liposomal DOX and albumin-bound paclitaxel—have improved safety profiles by reducing off-target toxicities and altering systemic pharmacokinetics. Nevertheless, a persistent translational “valley of death” remains between compelling preclinical efficacy and clinical benefit.
A central conclusion emerging from this review is that preclinical model choice is not a peripheral technicality but a primary determinant of translational reliability. Across the cancers surveyed, subcutaneous xenografts often provide an overly permissive microenvironment that inflates penetration, underestimates clearance, and misrepresents immune constraints. In contrast, orthotopic models expose nanotherapeutics to the organ-native barriers that ultimately govern delivery, activation, and efficacy in patients. Accordingly, the major challenges and future directions of nanomedicine should be reframed through a single organizing principle: orthotopic model validation as the anchor for credible mechanism-of-action claims and clinically relevant performance metrics.

Core challenges: the orthotopic–subcutaneous disconnect
The high clinical attrition of nanomedicines—often reported to exceed 75%—is attributable in part to historical dependence on subcutaneous xenografts that fail to recapitulate the native organ’s anatomical, biochemical, and immunological architecture. To align with the theme of this manuscript, we summarize the core translational barriers explicitly through an orthotopic validation lens and propose a framework that explains why subcutaneous models systematically misestimate nanodrug performance.
(1) A taxonomy of orthotopic barriers that subcutaneous models cannot reproduce.
Therapeutic performance in native organs is governed by three interdependent classes of orthotopic barriers that are absent or profoundly attenuated in ectopic sites:
Physical barriers: Organ-native structures (e.g., the rigid prostatic capsule, dense desmoplastic stroma in pancreatic cancer, multilayer mucus in the cervicovaginal tract, and urothelial architecture in the bladder) generate elevated IFP, restricted convection, and physical exclusion zones. These constraints fundamentally reshape how nanoparticle size, shape, stiffness, and motility translate into intratumoral distribution.
Biochemical barriers: Orthotopic tissues exhibit compartmentalized enzymatic activities and steep microenvironmental gradients (e.g., hepatic metabolism and clearance pathways; acidic, enzyme-rich vaginal or bladder milieus), which can trigger premature prodrug cleavage, degrade payloads, or alter “responsive” chemistry outside the intended compartment. Subcutaneous sites rarely reproduce these organ-specific triggers with comparable spatial and temporal heterogeneity.
Immune barriers: Native TMEs contain niche-specific immunosuppressive programs (e.g., MDSC-dominant suppression in prostate cancer, Kupffer-cell capture in the liver, or organ-resident macrophage phenotypes), which can dominate nano-immunotherapy outcomes and dictate clearance, antigen presentation, and therapeutic synergy. Subcutaneous tumors, shaped by skin-associated immunity, frequently misrepresent these constraints.
Importantly, these barrier classes are not independent: physical fibrosis can create hypoxia and acidity (biochemical), which in turn drives immunosuppression (immune). Therefore, orthotopic validation is essential not only for efficacy estimation but also for mechanistic attribution.
(2) Pharmacokinetic and distribution “false positives” in ectopic models.
Subcutaneous tumors commonly overestimate penetration and retention because they lack organ-specific stromal density, fluid dynamics, and clearance routes. For example, GEM accumulation can differ markedly between pancreatic orthotopic and subcutaneous models due to the absence of dense fibrosis in the latter, producing misleadingly optimistic exposure estimates [75]. Similar distortions arise when evaluating non-intravenous routes: oral, inhaled, intravesical, or intravaginal delivery must be tested in orthotopic settings to capture physiological “washout,” mucus turnover, mucosal absorption, and luminal clearance—features that subcutaneous models inherently lack.
(3) Methodological gaps that obscure the true “nanocarrier advantage”.
A recurring weakness across orthotopic nanomedicine studies is incomplete rigor in comparative design. Frequent omissions include: (a) inadequate free-drug controls matched for dose and schedule, which prevents quantifying the true carrier-mediated benefit; (b) insufficient reporting of spatiotemporal distribution metrics (e.g., radial penetration depth, microregional exposure in hypoxic cores vs. invasive margins); and (c) limited assessment of long-term, organ-specific accumulation and toxicity, particularly for materials that may persist in reticuloendothelial organs. Without these elements, orthotopic studies can still yield mechanistic ambiguity and underpowered safety conclusions, weakening translation even when efficacy signals appear strong.
Collectively, these issues demonstrate that the “orthotopic–subcutaneous disconnect” is not simply a difference in tumor location; it is a systemic mismatch in barrier architecture that can invert conclusions about penetration, activation, immunomodulation, and safety. Orthotopic validation should therefore be treated as a non-negotiable evidentiary standard for claims of delivery optimization and translational readiness.

Future directions: a strategic roadmap to clinical translation
To move beyond generic calls for “better delivery” and directly reflect the orthotopic theme of this manuscript, future nanomedicine research should adopt a barrier-to-material mapping strategy supported by rigorous, stage-gated validation models.
(1) “Non-negotiable” validation models and a stage-gated pipeline.
Future development should transition from subcutaneous xenografts as the default endpoint to a tiered validation pipeline:
Early screening (optional subcutaneous): Rapid formulation optimization and initial safety profiling.
Mandatory orthotopic validation: Demonstrate penetration, activation, and efficacy under native organ constraints with standardized metrics (penetration depth, intratumoral exposure heterogeneity, barrier remodeling indices, and immune infiltration signatures).
PDOX as a translational gatekeeper: Patient-Derived Orthotopic Xenografts (PDOX) should be prioritized to preserve tumor architecture, stromal composition, and metastatic behavior, thereby reducing false positives driven by clonal cell-line artifacts.
Organ-on-chip for mechanistic deconvolution: Organ-on-chip platforms can recapitulate human-relevant fluid dynamics, endothelial transport, and barrier permeability, enabling controlled testing of size/charge/shape effects and orthotopic barrier traversal in a high-throughput and human-relevant manner.
Within this pipeline, PDOX and organ-on-chip systems should be treated as non-negotiable for late-stage preclinical claims, consistent with the reviewer’s recommendation.
(2) A barrier–property mapping framework to guide rational nanodesign.
Rather than proposing organ-specific delivery strategies in general terms, we recommend an explicit framework: classify orthotopic barriers (physical, biochemical, immune) and then map nanomaterial properties against each class:
Physical-barrier solutions: Size-transformable systems (shrink-on-demand), shape/softness tuning, ECM-interactive motifs (collagen-binding peptides), active propulsion or externally driven penetration (US-triggered contraction, microbubble-assisted transport), and barrier-remodeling modules (MMP activation, stromal deactivation).
Biochemical-barrier solutions: Trigger designs matched to compartmental chemistry (acidic lumen vs. intracellular reductive cytosol), enzyme-selective linkers, protection against premature degradation, and multi-stage release to separate “transport” from “cytotoxicity” temporally.
Immune-barrier solutions: Biomimetic camouflage (cell membranes, “don’t-eat-me” signaling), organ-specific RES evasion strategies (e.g., Kupffer cell avoidance), and immunotherapy pairing that is validated specifically in orthotopic immune niches (including humanized immune models where feasible).
This mapping approach converts “orthotopic complexity” from an obstacle into a design blueprint, enabling formulations to be justified by the barrier they are engineered to overcome.
(3) Multi-omics integration for context-specific targeting and functionalization.
To add novelty and align with the manuscript’s orthotopic focus, future studies should integrate multi-omics (genomics, proteomics, metabolomics) derived directly from orthotopic tumors. Such datasets can identify context-specific targets that do not emerge in subcutaneous settings—e.g., proteins enriched in hypoxic orthotopic cores, invasion fronts, or organ-specific lymphatic niches—thereby enabling rational functionalization of nanocarriers. Importantly, multi-omics should be coupled to spatial profiling to connect targets with barrier microregions (hypoxia, acidity, immune deserts), transforming targeting from “universal biomarkers” toward niche-informed precision delivery.
(4) Biomimetic yet scalable manufacturing and standardization.
While biomimetic strategies (e.g., cell membrane coating) may be necessary to navigate organ-specific clearance and immune barriers, they also introduce manufacturing complexity. Bridging bench-scale synthesis to industrial production will require prioritizing scalable technologies (e.g., microfluidic assembly, continuous manufacturing) and establishing standardized characterization criteria for clinical-grade nanoparticles (size distribution, surface charge, ligand density, stability, and release kinetics). Internationally harmonized standards (e.g., ISO-aligned parameter definitions) will be critical to improve batch-to-batch reproducibility and regulatory clarity.
(5) From TME-responsive carriers to orthotopic-driven discovery and closed-loop adaptive systems.
Finally, prospects such as “intelligent nanorobots” should be integrated into a coherent roadmap rather than introduced as an isolated vision. A compelling trajectory is: current TME-responsive nanocarriers → orthotopic model–driven discovery and validation → closed-loop adaptive systems.
In this end-stage paradigm, nanosystems would not only respond to static triggers but sense and adapt to evolving orthotopic conditions in real time—monitoring hypoxia, pH, redox state, or perfusion and dynamically adjusting release kinetics or activation thresholds accordingly. Coupling theranostic readouts with feedback control (potentially supported by artificial intelligence) could enable “closed-loop” precision dosing that is tuned to the native organ microenvironment, thereby maximizing efficacy while minimizing off-target toxicity.
In conclusion, nanomedicine for precision oncology has matured from proof-of-concept delivery vehicles to multifunctional platforms capable of barrier remodeling and immune reprogramming. Yet durable translation will require a decisive paradigm shift: anchoring mechanistic claims and performance benchmarks in orthotopic-informed validation, enforcing PDOX and organ-on-chip testing as late-stage standards, and adopting a barrier–property mapping framework to guide rational design. With these orthotopic-centered principles, the field can more credibly progress toward adaptive, closed-loop systems that deliver clinically meaningful and durable benefits for patients.