Non-small cell lung cancer (NSCLC) is the primary histological type of lung cancer, accounting for ~85% of all lung cancer cases. Its subtypes include adenocarcinoma, squamous cell carcinoma, and large-cell carcinoma (1-3). NSCLC exhibits marked molecular heterogeneity and a complex tumor microenvironment, both of which contribute to tumor stemness, metastatic potential and therapeutic resistance (4,5). Although significant progress has been made in surgery, radiotherapy, chemotherapy, targeted therapy and immunotherapy, the majority of patients are diagnosed at advanced stages, with the 5-year survival rate still <25%. Issues, such as drug resistance and recurrence urgently require innovative therapies (6,7). Neural stem cells (NSCs) are a type of progenitor cells with self-renewal capacity and multi-lineage differentiation potential, residing in the central nervous system of adult mammals (8). Under physiological conditions, NSCs undergo asymmetric division to produce neurons and glial cells (9), maintaining adult neurogenesis, supporting learning, memory and nerve injury repair, thus endowing the nervous system with high plasticity and regenerative potential (10).

In recent years, the crosstalk between NSCs and tumor cells has gradually emerged as a critical perspective for unraveling the mechanisms underlying tumor initiation and progression (11,12). Although the tumor microenvironment (TME) is commonly recognized as a sophisticated ecosystem composed of tumor cells, immune cells, fibroblasts and extracellular matrix, the pivotal roles of neural components and stem-like properties of tumor cells within this ecosystem have long been overlooked. Accumulating evidence currently demonstrates that NSCLC cells express a variety of NSC markers and activate relevant signaling pathways to exhibit 'NSC-like behaviors (13). Moreover, neurogenic factors, neurotransmitters and direct cell-cell contact can profoundly modulate the proliferation, stemness maintenance, invasion and metastasis capacities of tumor cells (14-16). This emerging research niche has been conceptualized as the 'NSC-NSCLC axis'. Within this framework, NSCLC cells augment cancer stemness and therapeutic resistance by hijacking multiple core signaling cascades of neural stem cells; nerve fibers residing in the TME remodel the immune microenvironment and facilitate angiogenesis via the secretion of neurotrophic factors. Intriguingly, tumors can reciprocally regulate peripheral neurons and potential NSC-like cell populations, thereby establishing a bidirectional regulatory loop. Nevertheless, the precise molecular mechanisms and therapeutic targets in this field remain incompletely characterized.

Therefore, analyzing the key molecular drivers of the 'NSCs-NSCLC axis', the associated microenvironment cross-talk networks and their functions in malignant progression will help uncover novel biological targets and provide innovative insights for the precise diagnosis and treatment of NSCLC. The present review aimed to summarize the latest research advances in this field and provide perspectives on potential therapeutic strategies (Fig. 1).

NSCs are the main population of endogenous adult stem cells in the central nervous system, primarily residing in the subventricular zone and the subgranular zone of the hippocampal dentate gyrus (17). Under pathological conditions, NSCs can also be derived from reactive astrocytes or neural crest-derived progenitor cells (18,19). Their core characteristics involve the long-term maintenance of a dynamic balance between 'self-renewal' and 'multi-lineage differentiation' potential (20,21). This balance is regulated by a set of highly conserved molecular markers, such as Nestin, SOX2, Musashi-1, CD133, along with core signaling pathways such as Notch, Wnt/β-catenin, Shh and BMP (22,23). Notably, these molecules and pathways are also highly active in cancer stem cells (CSCs). This 'molecular homology' is not coincidental, but rather reflects the evolutionary co-option of developmental programs by tumor cells (24). For example, the Notch signaling pathway regulates cell fate determination in normal NSCs, whereas in NSCLC CSCs, it is frequently constitutively activated, thereby promoting stemness maintenance and chemotherapeutic resistance (4). However, the majority of existing studies remain largely descriptive of this overlap (24,25). The crosstalk among distinct signaling pathways within specific TME contexts, as well as their spatiotemporal dynamics, remains poorly understood, representing a critical area for future mechanistic investigation.

In the TME, NSCs can be actively recruited. Chemokines and growth factors, including SDF-1α/CXCL12, SCF, IL-6 and TGF-β secreted by tumor cells and tumor-associated macrophages bind to surface receptors, such as CXCR4 and c-Kit on NSCs, subsequently activating the PI3K/Akt, MAPK and JAK/STAT3 signaling pathways to drive the directional migration of NSCs toward tumor lesions (26-28). In addition, the hypoxic microenvironment upregulates VEGF expression via HIF-1α, which cooperates with exosome-carried miR-21 to further facilitate the homing and survival of NSCs (29). These mechanisms lay a cellular foundation for the subsequent functional aberrance of NSCs within the TME. After entering the TME, the intrinsic normal differentiation program of NSCs is hijacked. Tumor cell-derived Notch ligands (Jagged1/DLL4) persistently activate intracellular Notch signaling in NSCs to repress cell differentiation (30); miR-21 and miR-210 encapsulated in tumor-derived exosomes remodel the survival and metabolic phenotypes of NSCs by targeting PTEN and inhibiting mitochondrial metabolism, respectively (31); furthermore, hypoxia upregulates the levels of pluripotency factors, including Oct4 and Nanog in a HIF-1α-dependent manner, arresting the differentiation of NSCs into mature neural cells (32). Collectively, these regulatory mechanisms endow NSCs with a pro-tumorigenic phenotype

NSCLC CSCs achieve the functional hijacking of the NSC programs by aberrantly activating core molecular signatures and signaling networks that highly overlap with those of NSCs, exhibiting significant 'NSC-like behavior (33). This 'co-option of homologous programs' not only endows NSCLC CSCs with extreme stemness, therapeutic resistance and metastatic potential, but also provides a molecular basis for the tumor cells to efficiently recruit and utilize authentic NSCs within the TME.

A summary of the functional divergence of certain shared markers and their clinical significance is provided in Table I (34-73). Taking SOX2 as an example, in normal NSCs, it forms a positive feedback loop with other factors (such as Nestin and TLX) to maintain stemness homeostasis (34). However, in NSCLC CSCs, SOX2 often collaborates with ASCL1 to induce neuroendocrine differentiation (35), and interacts with various factors such as miRNAs, lncRNAs, STAT3 and Wnt/β-catenin to drive tumor progression and drug resistance (36,37), becoming an independent predictor of a poor prognosis (40). Notably, discrepancies exist among studies regarding the reported functions of these markers, which may arise from tumor heterogeneity, differences in model systems (such as cell lines, organoids and in vivo models),and the complexity of microenvironmental signals. For example, CD133 expression is associated with multiple pro-tumor characteristics: It can maintain stemness by activating the Src/PI3K/mTOR pathway (49); under hypoxic conditions, CD133⁺ cells enhance adaptability by stabilizing HIF1α/2α (50). Furthermore, this subpopulation exhibits a strong potential for initiating brain metastasis (53,54). However, the expression and function of CD133 are not fixed and may be dynamically regulated by microenvironmental factors such as hypoxia and inflammation. Therefore, future research should not be limited to descriptive correlations of markers, but should also delve into their upstream regulatory mechanisms and downstream effect networks in specific clinical subtypes or microenvironmental contexts.

Notably, the hijacking of neural programs is not confined to stemness and therapeutic resistance, but extends directly to metastasis. Several of the shared markers and pathways co-opted by NSCLC CSCs are mechanistically linked to invasive and metastatic behavior. CD133⁺ subpopulations represent the strongest initiators of brain metastasis (53,54), and Nestin serves as an independent risk factor for brain and bone metastasis in lung adenocarcinoma by forming complexes with vimentin/Snail/Slug to drive epithelial-mesenchymal transition (58-60). Moreover, as detailed below, functionally reprogrammed NSCs further promote metastasis by secreting VEGF to induce angiogenesis and releasing MMP-2/9 to degrade the extracellular matrix, thereby paving the way for the invasion and metastatic dissemination of CSCs. This metastatic dimension is further reinforced at the level of neural signaling and of the brain-metastatic niche, as discussed in the corresponding sections. In this sense, the same neural-program hijacking that sustains stemness and resistance also drives the metastatic potential of NSCLC.

After entering the TME, the normal differentiation program of NSCs is hijacked, which in turn acts reciprocally on NSCLC CSCs to form a malignant amplification loop. Tumor-cell-secreted Notch ligands (Jagged1/DLL4) can persistently activate intracellular Notch signaling in NSCs and inhibit their differentiation (74); miR-21 and miR-210 derived from tumor exosomes remodel the survival and metabolic phenotypes of NSCs by targeting PTEN and suppressing mitochondrial metabolism, respectively (31,75); the hypoxic microenvironment upregulates pluripotency factors, such as Oct4 and Nanog via HIF-1α, thereby blocking the differentiation of NSCs into mature neural cells (76). Following functional aberrance, these NSCs secrete neurotrophic factors, including brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF) and glial cell line-derived neurotrophic factor (GDNF) in a paracrine manner to activate PI3K/Akt,MAPK/ERK and other signaling cascades in NSCLC CSCs, elevating the expression of stemness markers and conferring chemoresistance (77,78). Furthermore, NSCs produce VEGF to facilitate angiogenesis and release MMP-2/9 to degrade the extracellular matrix, paving the way for the invasion and metastasis of CSCs (79). A subset of NSCs can further differentiate into neuron-like cells that form synapse-like junctions with tumor cells to supply sustained neurotrophic support (80). Collectively, the aforementioned mechanisms constitute the molecular basis underlying the bidirectional crosstalk between NSCs and NSCLC CSCs.

NSCLC is not a disease process driven solely by genetic mutations. An increasing body of evidence suggests that, under sustained neural input, tumor cells can undergo alterations in receptor expression profiles and reprogram downstream signal transduction, thereby converting external neural stimuli into a broad spectrum of biological effects, including enhanced proliferation and survival, invasion and metastasis, angiogenesis, immune suppression and therapeutic resistance (81,82). These neural signals include neurotransmitters, neurotrophic factors and neuropeptides. It should be emphasized, however, that substantial heterogeneity exists among studies with respect to experimental models, exposure paradigms and endpoint definitions. As a result, the magnitude and even the direction of the effects attributed to a given 'neural signal' are not always consistent across different investigations. Accordingly, this section is structured based on the relevance of each signaling axis to NSCLC and the relative contribution of direct versus indirect effects, thereby clarifying which findings are approaching consensus and which should be interpreted as preliminary or suggestive (Fig. 2).

In NSCLC, the 'neural-stem cell crosstalk' is not merely regulated via paracrine signaling mediated by neurotransmitters, neurotrophic factors or neuropeptides; instead, it is predominantly manifested as niche remodeling driven by nerve fiber infiltration and glial cell recruitment (114). Neural signals deliver not only transient stimulatory cues, but may also reshape the TME into a neutralized microenvironment that favors the survival, migration and therapeutic escape of CSCs by altering local tissue architecture and cellular composition (115,116).

Notably, intratumoral nerve fibers are heterogeneously distributed and preferentially enriched at invasive fronts, perivascular regions and sites with prominent stromal remodeling, implying their synergistic interplay with invasive pathways (117). Accordingly, perineural invasion (PNI) enables tumor cells to spread along nerve tracts and gain survival privileges, which is associated with elevated risks of local recurrence and distant metastasis. Moreover, PNI provides pathological evidence for spatially focused neural input: Nerve tracts function as anatomically structured migratory conduits on the one hand, and confer localized survival benefits via continuous, short-range signaling and nutritional supply on the other. In other words, PNI should no longer be regarded solely as a passive consequence of advanced tumor progression; it serves as a pathologically accessible research model to verify whether neural-associated signaling pathways actively facilitate tumor invasion and metastasis.

Furthermore, the core feature of neural invasion/PNI lies not in the simple upregulation of soluble transmitters, but in the transformation of neural input from intermittent to persistent stimulation. Direct juxtaposition between nerve fibers and tumor cells or NSC-like cells establishes stable local concentration gradients, which endows CSC subsets with high receptor expression with persistent selective advantages and shifts intratumoral heterogeneity toward the emergence of neural-dependent clones (118).

In addition to neuronal axons, SCs represent another key cellular component of the peripheral nervous system whose role in the neural microenvironment of NSCLC may have been substantially underestimated. Under physiological conditions, SCs exhibit a pronounced injury-response and repair capacity: Following nerve injury, they can dedifferentiate into a 'repair-like' state characterized by enhanced migratory ability, the secretion of multiple growth factors and chemokines, and active remodeling of the extracellular matrix to support axonal regeneration (119,120). As the TME and chronic inflammation share substantial signaling similarities with 'tissue injury', it is plausible that tumors co-opt the SC repair program, positioning SCs as a potential hub linking neural input to the maintenance of tumor phenotypes, including stemness-associated states (121,122).

It should be emphasized that, to date, the NSCLC field lacks direct experimental studies centered on 'SC-tumor interactions'. Accordingly, the following discussion is primarily intended to propose transferable and testable mechanistic frameworks based on evidence from other solid tumors, rather than to draw definitive conclusions for NSCLC. Existing studies nonetheless provide at least two lines of evidence that are highly relevant to NSCLC (93,94). First, in melanoma models, tumors have been shown to reprogram surrounding SCs into a repair-like state and activate 12/15-LOX (Alox15) and COX2-associated eicosanoid pathways, thereby increasing the production of immunosuppressive lipid mediators, such as PGE2. These mediators subsequently suppress T-cell activation via EP4 signaling on T-cells. By analogy, if similar repair-like SCs are present in the peritumoral regions of NSCLC, they may contribute to the formation of an immunosuppressive niche through a '12/15-LOX/COX2-PGE2-EP4' axis (93). Second, in pancreatic ductal adenocarcinoma, cancer cells can induce the c-Jun-dependent injury-repair-like reprogramming of non-myelinating SCs, leading to the formation of dynamic 'tumor-activated Schwann cell tracks' (TASTs) that provide both physical paths and mechanical support for tumor cell migration and local invasion (94). Correspondingly, in NSCLC subpopulations exhibiting neural infiltration or PNI phenotypes, whether SCs display c-Jun activation and form TAST-like structures, and whether such structures spatially colocalize with the invasive front, represents a question of high priority for experimental validation.

Beyond their roles in immunity and invasion, SCs are also key sources of neurotrophic signals. SCs can produce neurotrophic factors, such as NGF, BDNF and GDNF, which may enhance tumor cell survival and stress tolerance on the one hand (123), while promoting further growth and the branching of nerve fibers on the other hand (124). Together, these effects may reinforce intratumoral neural input. Moreover, dedifferentiated repair-like SCs can participate in matrix remodeling and alterations in adhesion molecule networks, facilitating tumor cell migration along neural or stromal tracks and thereby providing structural conditions for PNI and the formation of invasive fronts (125,126). When tumor cells are chronically exposed to a combination of neurotrophic factors, inflammatory mediators and hypoxia within SC-enriched regions, they are more likely to sustain the functional activity of proteins, such as SOX2 and Nestin, and to acquire enhanced resistance under therapeutic pressure (127,128).

Therefore, the significance of discussing SCs in NSCLC extends beyond merely 'identifying an additional cell type'. Rather, SCs provide a more explanatory and potentially actionable upstream framework for understanding why neural-related signals can form localized, high-intensity and sustained inputs within tumors, and how such persistent inputs establish more coherent causal links with key phenotypes, such as immunosuppression, invasion and metastasis, and therapeutic resistance. On this basis, validation studies focusing on SC spatial localization, repair-like markers and key signaling axes may help advance 'neural-related phenomena' from correlative observations toward mechanistically tractable targets.

Across the NSCs-NSCLC axis, inflammation emerges as an integral intermediary rather than an incidental feature. Neural signaling and inflammatory responses are tightly intertwined at multiple nodes discussed throughout the present review. NGF signaling is closely associated with the initiation and progression of tumor-associated inflammation (97); catecholamine/β2-adrenergic signaling stimulates the secretion of inflammatory cytokines, such as IL-6, thereby remodeling the immune microenvironment and attenuating cytotoxic T-cell activity (90); the activation of the substance P receptor is frequently accompanied by upregulation of inflammation-related responses (110) and repair-like Schwann cells generate immunosuppressive lipid mediators, such as PGE2 via the 12/15-LOX/COX2-EP4 axis (93). Inflammatory mediators act as key intermediaries linking neural input to immune remodeling, stemness maintenance, and therapeutic resistance.

An improved understanding of the NSCs-NSCLC axis is reshaping contemporary paradigms of cancer therapy. Accumulating evidence suggests that neural signals not only directly promote tumor growth, but may also function as an upstream regulatory layer driving multiple malignant phenotypes, including immunosuppression, the maintenance of stemness, metastatic colonization and the evolution of therapeutic resistance. For example, Cao (129) proposed the concept of 'neural stemness' as a basal state underlying cellular tumorigenesis and differentiation potential, providing a theoretical framework for tumor stemness maintenance. Cerv antes-Villagrana et al (130) demonstrated that tumor-induced neurogenesis and immune evasion may represent actionable targets for innovative anti-cancer therapies. As regards metastatic colonization, Ouyang et al (131) revealed that the neuropeptide precursor, neural growth factor-inducible gene (VGF), can remodel the microenvironment through paracrine loops to promote liver metastasis of uveal melanoma, implying that analogous mechanisms may also operate in NSCLC. By contrast, Rotow and Bivona (132) focused directly on NSCLC and comprehensively reviewed multiple resistance mechanisms, including those mediated by neural signaling. Collectively, these findings indicate that interventions targeting neural inputs possess multifaceted clinical value. Such strategies may not only directly suppress tumor cell proliferation and invasion, but may also enhance the efficacy of immune checkpoint inhibitors (ICIs) by relieving neural-associated immunosuppression, while additionally contributing to cancer pain control and an improved quality of life. Current translational efforts mainly advance along five directions: Receptor antagonism, intervention against PNI, targeting neuroendocrine programs, neural adaptation in brain metastases, and strategies for delivery and biomarker development (Fig. 3).

Brain metastasis is a major cause of mortality in NSCLC. Its therapeutic intractability is attributable not only to the blood-brain barrier (BBB), which restricts the penetration of numerous pharmacologic agents, but also to the adaptive support provided by immune networks formed by neural and glial cells within the brain (148). At the mechanistic level, preclinical studies have shown that brain metastases can engage in metabolic exchange and signaling crosstalk with astrocytes through Cx43-mediated gap junctions, thereby establishing a tumor-favoring protective microenvironment and enhancing resistance to chemotherapy (149). Consequently, Cx43 channel blockers, such as tonabersat, represent a potential strategy to disrupt astrocyte-mediated support (150,151).

In parallel, translational efforts are addressing BBB-related delivery constraints, with encouraging progress in receptor-mediated BBB nanocarriers, immune cell- or exosome-based 'Trojan horse' delivery and focused ultrasound (FUS)-mediated BBB opening (152-155). Notably, these technologies are not independent of neuro-oncological biology. If drugs targeting neural-dependent pathways, such as inhibitors of the glutamatergic signaling or glial interaction axes, can achieve effective concentrations brain metastatic lesions (156-158), their efficacy signals may emerge more clearly in brain metastases than in primary tumors, thereby in turn accelerating drug development and biomarker maturation in this field.

Single-cell sequencing and spatial omics technologies are driving the field from mechanistic narratives toward testable and stratifiable frameworks (159). At the single-cell level, these approaches enable the identification of tumor cell subpopulations with neural-like or stemness-associated features, as well as key cellular states such as repair-like SCs, thereby illuminating potential drug targets and trajectories of resistance evolution (160). In parallel, spatial transcriptomics preserves tissue architecture, while elucidating spatial associations among nerve fibers, immune cells and tumor cells, rendering phenomena such as the co-localization of increased neural density and immune exclusion quantitatively assessable (161,162). Based on these spatial and single-cell insights, neural signal blockade may not only suppress tumor growth, but may also enhance immunotherapeutic responses by alleviating local immunosuppression, providing a more robust rational for combining local neural modulation with systemic therapies.

In summary, translational pathways targeting the NSCs-NSCLC axis are becoming increasingly defined, from rapid clinical entry via β-blockers and glutamatergic pathway inhibitors (163), to structural interventions targeting neurotrophic signaling and perineural invasion with added analgesic benefits (140), to the precision immunotherapeutic targeting of neuroendocrine differentiation-associated resistance branches (164), and further breakthroughs achieved by integrating glial interaction targets and advanced delivery technologies in the brain metastatic niche (80). Future research priorities may focus on three major areas: First, precision stratification: Not all NSCLCs are neural-dependent, underscoring the urgent need for scalable biomarkers based on neural density, neurotransmitter levels and receptor expression profiles to identify responsive populations. Second, brain metastasis intervention: Dissection of the 'neuron-astrocyte-tumor' triadic interaction network to develop combination therapies capable of penetrating the BBB and disrupting key interaction nodes. Third, multidisciplinary integration: Advancing oncology-neuroscience convergence under the framework of cancer neuroscience to open new therapeutic avenues for refractory lung cancer.

The 'NSCs-NSCLC axis' framework proposed in the present review provides an integrated analytical perspective for understanding the interplay between neural stem cells and NSCLC. At present, studies on the interaction between NSCs and NSCLC remain limited, particularly as regards the regulatory mechanisms of the NSC-NSCLC axis within the TME. Translating discoveries in this field into clinical therapies presents both challenges and opportunities. One major obstacle is the BBB, which severely restricts the delivery of agents targeting key receptors, thereby rendering the highly active NSC-tumor interactions within brain metastatic lesions difficult to modulate effectively. Accordingly, the development of novel drug-delivery systems capable of traversing both the BBB and the blood-tumor barrier is of critical importance. Moreover, the same molecular signal may exert opposing effects depending on microenvironmental context or concentration, rendering single-target interventions prone to compensatory resistance or off-target toxicity. A more in-depth mechanistic analysis may facilitate the design of more precise multi-target combination strategies. Emerging technologies, including single-cell sequencing and spatial omics, provide powerful tools to overcome these limitations. These approaches enable the characterization of NSC heterogeneity within the TME and the delineation of their spatial distribution at the tumor-immune interface, thereby identifying the specific NSC subpopulations that truly drive therapeutic resistance and metastasis and furnishing a solid basis for precision targeting. Looking ahead, the integration of spatial multi-omics with artificial intelligence-based analysis may enable the identification of patient-specific 'neural-immune-stem cell' molecular landscapes, allowing for individualized stratification and precise modulation of the neural microenvironment. Combined with the dynamic monitoring of peripheral blood biomarkers, such approaches could support the real-time adjustment of therapeutic regimens. Collectively, this paradigm may shift lung cancer treatment from direct tumor eradication toward reprogramming of the tumor microecosystem, transforming neural stem cells from 'drivers' of tumor progression into controllable therapeutic targets.