Lung Cancer Brain Metastasis: Brain Microenvironmental Adaptation, Therapeutic Resistance and Translational Strategies Authors: Jun Yang, Junyi Xiong, Li Ying, Yanyang Zeng, Ziqiang Ouyang, Mei Liu, Meihua Li Published: 2026 ABSTRACT ABSTRACTLung cancer brain metastasis (LCBM) remains one of the most challenging clinical conditions limiting survival and neurological function. The development of LCBM is shaped by continuous exchange of molecular signals between tumor cells and the brain microenvironment, involving blood–brain barrier remodeling, glial and immune reprogramming, neuronal signaling, and metabolic adaptation. Together, these phenomena support brain colonization, immune escape, and therapeutic resistance. The management of LCBM has improved with the advent of stereotactic radiotherapy, targeted therapy, immunotherapy, chemoimmunotherapy, and antibody–drug conjugates. Yet durable intracranial control is still limited by intrinsic barrier heterogeneity, protected residual intracranial disease, and molecular resistance. This review summarizes recent advances in biology, preclinical modeling, therapeutic resistance, and translational treatment strategies of LCBM. By linking brain‐specific metastatic biology with evolving therapeutic approaches, this review provides a framework for understanding LCBM progression and for developing more effective strategies to overcome intracranial resistance. Abbreviations antibody–drug conjugate blood–brain barrier brain–computer interface brain metastasis (brain metastases) brain microenvironment blood–tumor barrier central nervous system cerebrospinal fluid circulating tumor DNA extracellular vesicle lung cancer brain metastasis lung adenocarcinoma multidisciplinary team non‐small cell lung cancer objective response rate oxidative phosphorylation small cell lung cancer stereotactic radiosurgery tyrosine kinase inhibitor whole‐brain radiation therapy Introduction Lung cancer brain metastasis (LCBM) is a common and clinically difficult manifestation of advanced lung cancer. Brain metastases are present in approximately 10%–20% of patients at the time of non‐small cell lung cancer (NSCLC) diagnosis and develop in 25%–50% during the disease course [1]. In small cell lung cancer (SCLC), approximately 15% of patients have brain metastases at diagnosis, and CNS relapses remain frequent [2]. Molecular subtypes further modify CNS risk and therapeutic response, especially in EGFR‐, ALK‐, KRAS‐, and MET‐altered disease [3, 4]. Genomic analyses of lung adenocarcinoma (LUAD) brain metastases have identified recurrent brain‐metastasis‐associated copy‐number alterations, including MYC, YAP1, and MMP13 amplification and CDKN2A/B deletion [5]. More recent multi‐omic studies suggest that LCBM progression reflects both shared trunk mutations and brain‐adaptive molecular programs after CNS colonization [6]. Although central nervous system (CNS)‐active systemic therapies and radiotherapy have improved intracranial management, durable control remains limited by drug‐delivery barriers, immune suppression, and acquired resistance [7, 8, 9]. LCBM cannot be explained by tumor genotype alone. Once tumor cells reach the CNS, they encounter a niche defined by the blood–brain barrier (BBB), resident glial populations, infiltrating and brain‐resident immune cells, neuronal signaling, and local metabolic restrictions [10, 11]. Multi‐omic studies suggest that, after CNS colonization, LCBM lesions acquire brain‐adapted molecular and metabolic states [6]. These changes may help metastatic cells persist in the CNS and alter immune activity and therapeutic sensitivity [12, 13]. On this basis, we view LCBM as a disease state shaped by tumor‐intrinsic programs and by vascular, immune, neuronal, glial, and metabolic pressures. This framework is useful for model development, therapeutic design, studies of drug resistance, and future clinical strategy. Brain Microenvironment LCBM progression reflects active remodeling of the CNS microenvironment. BBB remodeling enables metastatic entry and affects the level and distribution of intracranial drug exposure [14]. Along the way, glial and immune reprogramming impairs anti‐tumor surveillance and generates the niche in which disseminated cancer cells survive [15]. Signals from stromal cells, neurons, and local metabolic circuits further promote brain‐adaptive growth and therapeutic resistance. Together, these processes establish a neural–immune–vascular niche that facilitates intracranial colonization and disease progression. BBB And Vascular Niche The BBB is fundamental to LCBM biology because it links metastatic entry to early colonization. Circulating tumor cells and host factors remodel endothelial adhesion, tight‐junction integrity, and perivascular homeostasis [9]. As lesions progress, the transition from an intact BBB to a more heterogeneous blood–tumor barrier (BTB) has major therapeutic consequences, especially for drug delivery and spatial variability of exposure within intracranial disease (Figure 1). The blood–brain barrier in lung cancer brain metastasis. Created with BioRender.com. BBB Remodeling as a Gateway for CNS Entry During early brain seeding, circulating tumor cells arrest brain micro‐vessels, interact with brain microvascular endothelial cells, and undergo trans‐endothelial migration. Tumor‐derived extracellular vesicles (EVs) can prime the endothelium before overt colonization, including through lncRNA XIST/EPN2 signaling and exosomal L1CAM‐mediated permeability changes [16, 17]. Xia et al. reported that upregulation of mesothelin activates the JNK/MET pathway and disrupts tight junctions, promoting LCBM [18]. RBM10 deficiency can also destabilize BBB tight junctions by modulating sphingolipid metabolism, thereby promoting EGFR‐mutated LCBM [19]. Thus, BBB remodeling functions as an entry program that enables vascular arrest, adhesion, and extravasation. Neurovascular Unit Remodeling After tumor cells engage the brain vasculature, BBB dysfunction becomes a multicellular neurovascular‐unit process. Pericytes, basement‐membrane components, and astrocytic end‐feet determine whether transient permeability evolves into a stable perivascular niche. The BBB‐to‐BTB transition is accompanied by vascular dilation, basement‐membrane remodeling, altered pericyte composition, and loss of astrocytic aquaporin‐4 polarization [14, 20]. Pericytes have context‐dependent roles: Pericyte‐derived factors may restrain lung cancer cell growth at the BBB, whereas lung cancer stem cell‐derived pericyte‐like cells enhance trans‐endothelial migration through GPR124 [21, 22]. CD146+ pericyte‐like metastatic stem cells promote VEGF/VEGFR‐dependent angiogenesis [23]. Astrocytes also participate in BBB regulation; lung adenocarcinoma‐secreted acetylcholinesterase activates astrocytic cholinergic signaling, triggers excessive Ca2+ influx and apoptosis, and disrupts the BBB [24]. Perivascular Niche and Early Metastatic Colonization After BBB crossing, the perivascular niche provides a foothold for early survival. Metastatic cells often remain close to brain micro‐vessels, where endothelial adhesion, basement‐membrane contacts, and local trophic cues support outgrowth. ALCAM‐mediated endothelial interaction promotes NSCLC‐BM, while vascular co‐option and anti‐plasmin serpins support early survival in the brain [25, 26]. GPNMB‐driven endothelial remodeling links BBB disruption with CXCL12‐CXCR4‐mediated immune infiltration and T‐cell exhaustion [27]. The vascular niche therefore acts as both an entry route and an adaptive scaffold for survival, immune modulation, and treatment resistance. Remodeling of the Brain Tumor Microenvironment The LCBM microenvironment is shaped by coordinated glial, immune, and neuronal remodeling, which supports tumor survival, invasion, and immune escape (Figure 2). The tumor microenvironment in lung cancer brain metastasis. Created with BioRender.com. Astrocytes Astrocytes are among the earliest brain‐resident cells engaged after extravasation. They do more than form reactive gliosis; they organize survival, immune escape, and invasion. In SCLC, tumor‐secreted Reelin recruits astrocytes and induces brain‐development‐like programs, whereas astrocyte‐derived SERPINE1 supports metastatic growth [28]. In NSCLC, astrocyte‐induced mGluR1 stabilizes EGFR through glutamate‐dependent signaling, thereby facilitating tumor‐cell adaptation to the brain microenvironment (BME) [29]. In addition, GABA‐mediated activation of astrocytic NF‐κB is associated with brain metastatic progression [30]. Astrocytes can be reprogrammed by metabolic and vesicle‐mediated signals. Lactate‐activated astrocytes release EVs enriched in miR‐8085, which suppresses TRIM67 and stabilizes ELK1 in tumor cells, thereby promoting stemness [31]. Astrocytes also shape immune escape. In EGFR‐mutant NSCLC‐BM, astrocyte‐derived IL11 activates EGFR/IL11RA signaling, increases PD‐L1 expression, and promotes CD8+ T‐cell apoptosis [32]. STAT3‐positive astrocytes mark a pro‐metastatic reactive state, and pSTAT3+ astrocyte‐derived CHI3L1 is associated with invasive growth [33, 34]. Tumor‐derived S100A14‐enriched EVs further activate the astrocytic PIAS3‐STAT3 axis, thereby recruiting myeloid‐derived suppressor cells [35]. Microglia and Macrophages Microglia and macrophages form the major myeloid compartment in LCBM. Resident microglia and recruited macrophages occupy diverse activation states [36]. Tumor‐derived IL‐6 induces anti‐inflammatory microglial polarization through JAK2/STAT3 signaling [37], and EV‐transmitted LINC00482 promotes microglial polarization through the TGF‐β1 axis [38]. Furthermore, a TGF‐β‐related lncRNA signature in glioma is associated with macrophage infiltration and immune checkpoint expression, highlighting the role of TGF‐β signaling in CNS tumor immunity [39]. Myeloid remodeling is reinforced by glial feedback loops. MIF‐CD74 signaling has been implicated in microglial polarization and radio‐resistance [40], and LCN2 coordinates interactions among cancer cells, astrocytes, and macrophages [41]. These states often impair antitumor immunity through reduced phagocytosis, suppressive cytokines, and extracellular matrix remodeling. The HSP47‐collagen axis promotes myeloid polarization and suppresses antitumor immunity, whereas targeting this pathway restores CD8+ T‐cell activity [42]. Myeloid cells can have stage‐dependent antitumor roles: IFITM1‐high metastatic cells activate microglia through complement C3 and enhance MHC‐I‐dependent T‐cell killing during early colonization [43]. Distinguishing resident microglia from macrophages and defining time‐specific myeloid states will be essential for therapeutic targeting. T Cells and B Cells LCBM generally shows a weakened adaptive immune compartment compared with primary lung tumors. Spatial transcriptomic studies show reduced T‐cell and B‐cell abundance, and impaired cytotoxic activity [44]. CD8+ T cells may retain cytotoxic features but often localize to exhausted niches marked by PD‐1, CTLA‐4, LAG3, and TIGIT [45, 46, 47]. HSP70‐high T cells, loss of memory T cells, and stressed proliferative T cells have been associated with poor immune‐checkpoint inhibitor response [47]. However, CXCL13 + CD39+ tumor‐reactive T cells have been detected in brain metastases [48]. B cells are less studied, but single‐cell analyses identified naive, memory, germinal‐center‐like, and plasma‐cell states in selected lesions [46, 47]. Whether these states reflect productive tertiary lymphoid‐like immunity or non‐productive inflammation remains unresolved. Neutrophils Neutrophils connect local brain inflammation with systemic inflammatory remodeling. Enhanced neutrophil signatures at the tumor‐brain interface, together with reduced lymphoid activity, suggest an immunosuppressive role [44]. Tumor‐derived G‐CSF recruits ARG1 + PD‐L1+ neutrophils into the brain [49], and neutrophil extracellular trap (NET)‐releasing neutrophils accumulate around LOX+ metastasis‐initiating cells and promote metastasis through the NET‐KRT10 axis [13]. Elevated peripheral blood‐derived inflammatory indices, such as NLR and PLR, have been associated with BM risk or prognosis in NSCLC [50, 51]. Neuronal Regulation and Cancer Neuroscience in LCBM Neuronal regulation adds a brain‐specific layer to LCBM niche adaptation by linking neurotransmitter signaling, astrocyte crosstalk, BBB remodeling, and metabolic reprogramming. Neurotransmitter‐Driven Adaptation in NSCLC In NSCLC‐BM, neuronal regulation is largely reflected by neurotransmitter‐associated metabolic adaptation. GABA signaling activates the FOXA2/ABAT/GABA axis and promotes NF‐κB‐dependent metabolic reprogramming and brain colonization [30]. Astrocyte‐induced mGluR1 stabilizes EGFR through glutamate‐dependent signaling, creating a brain‐specific survival dependency [29]. In patients with a low body mass index, ghrelin‐induced neuronal NPY activates the NPY‐Y5R axis and drives tumor energy metabolic reprogramming [52]. The AChE/cholinergic‐astrocyte axis links neural signaling to astrocyte injury and BBB disruption, suggesting that neuronal pathways may regulate both entry and niche formation [24]. Neurodevelopmental Mimicry and Neuron‐Tumor Synaptic Interactions in SCLC SCLC shows more prominent neuron‐related features. SCLC cells secrete Reelin to attract astrocytes to induce brain‐development‐like programs, and astrocyte‐derived SERPINE1 supports metastatic growth [28]. This suggests that SCLC‐BM may rely on neurodevelopmental mimicry. Notably, SCLC can directly exploit neuronal activity after intracranial colonization. Glutamatergic and GABAergic cortical neuronal activity support SCLC growth mediated by paracrine and synaptic neuron‐cancer interactions [53]. Thus, SCLC brain metastasis may represent a more direct cancer‐neuroscience model. Neuroimmune Circuits Neuroimmune circuits are seldom studied in LCBM. Evidence from other brain metastasis contexts suggests that neural signaling may contribute to immunosuppression [54], but whether neuronal activity directly shapes immune remodeling in LCBM remains unclear. In glioma, enhanced neuronal connectivity is linked to regional immunosuppression and enrichment of anti‐inflammatory tumor‐associated macrophages [55]. Metabolic Adaptation The brain provides a unique metabolic niche for metastatic tumor cells. Within this environment, cancer cells face high energy demands, selective nutrient availability, and BBB‐regulated transport. To survive under these constraints, metastatic cells rewire energy production, lipid metabolism, and amino acid metabolism. These adaptations can sustain intracranial growth, reduce immune pressure, and promote therapeutic resistance. Mitochondrial Metabolism and Oxidative Phosphorylation (OXPHOS) In LCBM, multi‐omic profiling showed enrichment of mitochondrial and OXPHOS‐related programs, and gamitrinib inhibited the growth of tumor‐derived organoids [6]. HMGB3 promotes brain metastasis by recruiting SSBP1 and remodeling mitochondrial metabolism [56]. Glycolytic Metabolism LUAD‐BM is associated with high glycolysis, accumulation of lactate, and reduced CD8+ T‐cell infiltration [12]. Further evidence from brain tumors indicates that lactate released from hypoxia‐adapted tumor cells can drive the activation of SPP1+ macrophages and immune evasion [57]. SEC61G promotes metastasis by obstructing PGAM1 ubiquitination and degradation, linking phosphoglycerate metabolism to immune environment formation [58]. Lipid Metabolism Lipid remodeling in LCBM includes cholesterol signaling, sphingolipid metabolism, and fatty acid synthesis. Cholesterol causes metabolic changes and creates a favorable pre‐metastatic environment [59]. RBM10 deficiency promotes increased GALC‐dependent generation of sphingosine‐1‐phosphate and enhances BBB permeability [19]. IGF2BP3 stabilizes the FASN mRNA, promotes lipogenesis, and produces neutral lipid accumulation, emphasizing the role of de novo fatty acid synthesis as a target for therapy [60]. Amino Acid Metabolism Brain amino acid metabolism is shaped by BBB transport and neuron–glia circuits. Branched‐chain amino acid metabolism promotes BM through the BCAT1/ALKBH5 axis and m6A‐dependent regulation of epithelial–mesenchymal transition [61]. GABA‐ and glutamate‐related pathways further link neurotransmitter metabolism to niche adaptation [29, 30]. IDO1‐mediated tryptophan metabolism is associated with Treg enrichment, T‐cell dysfunction, and poor survival [62]. Cerebrospinal fluid (CSF), serum, and microbiome‐associated metabolite studies support the biomarker potential of amino acid metabolism in LCBM [63, 64, 65]. Metabolic remodeling in LCBM is both a mechanism of brain‐niche adaptation and a potential source of liquid biomarkers. Because brain metastatic tissue is difficult to obtain, serum, CSF, and microbiota‐derived metabolic profiles may provide practical tools for detecting early CNS dissemination and monitoring treatment response. Translational Models Preclinical models connect LCBM biology to therapeutic translation, but no single system reproduces the full metastatic cascade. Model selection should match the biological question (Table 1). TABLE 1Preclinical and translational models for lung cancer brain metastasis research.Model typeMetastatic step capturedMain strengthsKey limitationsMain biological questionIntracranial injection modelsIntracerebral growth and local adaptationReliable tumor take; short latencyBypasses dissemination and BBB crossingEstablished intracranial growth and therapy responseIntracardiac injection modelsVascular retention, BBB interaction, and early colonizationBetter models hematogenous dissemination than intracranial injectionVariable brain tropism; extracranial metastasesHematogenous dissemination and brain tropismCarotid artery injection modelsVascular retention, BBB interaction, and early colonizationHigher brain specificity than intracardiac injectionHigh technical requirements; potential for vascular injuryBBB crossing and brain‐enriched disseminationSyngeneic modelsRoute‐dependentPreserve the intact immune systemMouse‐specific biology; limited human subtype fidelityImmune Mechanisms and ImmunotherapyXenograft modelsRoute‐dependentPreserves human tumor geneticsLacks intact immunityTumor‐intrinsic mechanismsPDX modelsRoute‐dependentPreserves patient tumor heterogeneityLow engraftment; lacks intact immunityDrug testingHumanized modelsRoute‐dependentHuman immune modelingCostly; variable immune reconstitutionHuman‐derived immune response and immunotherapy testingSpontaneous modelsFull metastatic cascadeClosest to natural disseminationLong latency; costly; low scalabilityNatural Dissemination and Prevention StrategiesPatient‐derived organoidsBrain‐adapted tumor statesPatient‐specific and scalable drug testingLacks vascular flow, mature BBB/BTB, and systemic immunityDrug screening, metabolism, and patient‐specific vulnerabilitiesTumor‐brain organoid co‐cultureLocal invasion and neural adaptationCaptures neural niche better than 2D cultureLimited vascular and immune componentsNeural interactions and intracerebral adaptationMulti‐organ microfluidic systemsControlled inter‐organ disseminationModels organ–organ interactions and flowReductionist design; limited tissue complexityOrgan‐specific metastasis and drug resistance Preclinical and translational models for lung cancer brain metastasis research. Cell and Animal Models Cell lines and co‐culture systems remain useful for mechanistic studies and genetic perturbation. Sublines generated through repeated intracardiac or left‐ventricular selection enrich for cells capable of vascular arrest and early colonization [66], but prolonged culture and selection pressure may alter tumor states. Co‐culture systems isolate tumor‐niche interactions but lack blood flow, tissue architecture, and systemic immunity. In vivo models provide more physiologically relevant contexts. Intracranial injections are useful for established intracranial growth and treatment‐response experiments; however, they bypass hematogenous dissemination and BBB crossing [67]. Intracardiac and carotid artery injection models preserve brain seeding better, but carotid injection is technically challenging [68]. Xenograft and patient‐derived models preserve human driver mutations and allow targeted‐therapy experiments [69], whereas syngeneic models preserve immune status and answer immune remodeling and immunotherapy questions better [70]. Organoid and BBB‐On‐Chip Platforms Organoids and microfluidic models address shortcomings in cell culture and animal models. Tumor‐derived organoids preserve key tumor states and genotypes [6]. Peng and co‐workers generated a fast, personalized patient tumor organoid model suitable for primary and metastatic brain tumors. In this model, patient tumor samples are cultured in brain organoids, thus preserving the tissue cellular content and molecular profile of the tumor [71]. Organoid brain slice culture can preserve local brain tissue structures for drug testing. BBB‐on‐chip platforms can model vascular interfaces and quantify blood–brain barrier crossing and drug entry [72]. Multi‐organ chips model lung‐to‐brain dissemination and organ‐specific drug resistance [73]. The current limitations are technical complexity and the inability to fully recapitulate long‐term tumor evolution. Therefore, the best approach is to integrate vascular, immune, organoid, and in vivo systems according to the metastatic steps and the therapeutic challenge tested. Current Local and Systemic Therapies Treatment of LCBM requires balancing intracranial control, extracranial control, and preservation of neurological function. Figure 3 summarizes the current and emerging treatment landscape for LCBM. Surgery is generally reserved for selected patients with symptomatic lesions, mass effect, or diagnostic uncertainty [74]. Radiotherapy is central for local treatment. Compared with whole‐brain radiotherapy (WBRT), stereotactic radiosurgery (SRS) provides local control with less neurocognitive toxicity and is often preferred for selected patients with limited brain metastases. Chemotherapy alone is limited by the BBB, but platinum‐based chemotherapy remains important in SCLC and driver‐negative NSCLC, particularly as part of combination regimens [75, 76]. Treatment strategies for brain metastases from lung cancer. Created with BioRender.com. Systemic therapy tends to be organized according to molecular subtypes. In driver gene‐positive NSCLC, the emergence of targeted therapy has shifted the treatment strategy from conventional first‐line WBRT to systemic control (Table S1). Targeted therapy and radiotherapy are complementary, not interchangeable. Tyrosine kinase inhibitors (TKIs) can treat both intracranial and extracranial disease, but their efficacy is limited by acquired resistance, BBB heterogeneity, and the residual protected microenvironment. In driver gene‐negative NSCLC, immune checkpoint inhibitors can produce intracranial responses, especially in patients with high PD‐L1 expression. Monoclonal antibody therapy works through receptor blockade, immune checkpoint inhibition, antibody‐dependent cytotoxicity, and complement‐dependent cytotoxicity [77]. In LCBM, clinical applications should consider BBB heterogeneity, antigen expression, immunosuppression in the BME, and immune‐related adverse events. However, immunotherapy efficacy is limited by T‐cell dysfunction in the BME. Combination strategies are therefore increasingly important. C‐Brain supports the feasibility of brain radiotherapy plus camrelizumab and platinum‐doublet chemotherapy in untreated NSCLC brain metastases, while SPEED extends chemoimmunotherapy evidence to ES‐SCLC patients with untreated brain metastases [75, 76]. These approaches require careful sequencing, toxicity management, and patient selection. Mechanism‐Informed Therapeutic Strategies Targeting the Brain Metastatic Niche Mechanism‐informed therapy for LCBM seeks to translate brain‐niche biology into actionable strategies targeting CNS barriers, glial‐immune reprogramming, neuronal support, metabolic dependencies, and CNS drug delivery (Table 2). TABLE 2Niche‐targeted therapeutic strategies in lung cancer brain metastasis.CategoryStrategy/agentTarget or axisEvidence stageKey mechanism and translational implicationBBB/BTB and vascular nicheCeritinibClaudin‐10/VEGF‐A/VEGFR2PreclinicalRestores endothelial junction integrity and reduces NSCLC transendothelial invasion, supporting barrier‐preserving therapy.Anti‐CD276 antibodyCD276/B7‐H3PreclinicalTargets endothelial and mural‐cell immune regulation, suppresses brain metastatic growth and increases CD8+ T‐cell infiltration.Imprelimab; bemcentinibCD146/AXLPreclinicalDisrupts pericyte‐like metastatic stem‐cell programs and angiogenic vascular support.SB 290,157C3aRPreclinicalProtects the blood‐CSF barrier and limits leptomeningeal dissemination.Glial and immune niche reprogrammingCol003; ISO‐1HSP47‐collagen; MIF‐CD74PreclinicalReprograms suppressive microglia/macrophage states and may enhance anti‐PD‐L1 efficacy or radiosensitivity.IFITM1‐enriched oncolytic virusIFITM1/C3/MHC‐IPreclinicalActivates microglia and enhances CD8+ T‐cell cytotoxicity, with potential synergy with PD‐1 blockade.Gefitinib; gp130 inhibitionEGFR; IL‐11/gp130 axisPreclinicalBlocks astrocyte‐derived IL‐11‐associated EGFR/gp130 signaling and reduces immune escape.Silibinin; TIMP1 inhibitionSTAT3‐related astrocytes; TIMP1Clinical trial/preclinicalTargets reactive astrocyte programs and astrocyte‐mediated CD8+ T‐cell suppression.Neuron‐associated supportAS1517499; LY456236; BAY11‐7082STAT6‐GABA; mGluR1; NF‐κBPreclinicalBlocks neurotransmitter‐linked survival, EGFR stabilization, CSF niche support, and BBB penetration.Levetiracetam; riluzole; DCPGSynaptic vesicle release; glutamate release; GRM8PreclinicalDisrupts neuron–SCLC paracrine or synaptic support; represents repurposing of CNS‐active agents.Metabolic vulnerabilitiesGamitrinibMitochondrial OXPHOS/mitochondrial HSP90PreclinicalSuppresses brain metastasis organoids and may enhance anti‐PD‐1 response.TVB‐2640FASNPreclinicalInhibits de novo fatty acid synthesis and invasive growth.CNS delivery platformsHER3‐DXd; T‐DXdHER3 ADC; HER2 ADCClinical trialShows intracranial activity in selected NSCLC subtypes, but efficacy remains shaped by BTB permeability and antigen expression.Anti‐EGFR scFv exosome‐siLPCAT1EGFR/LPCAT1PreclinicalDelivers siRNA to suppress PI3K/AKT/MYC signaling, illustrating active tumor‐targeted CNS delivery.Peptide‐drug conjugates; biomimetic carriersBBB/BTB transport; gap junctions; USP7PreclinicalEnhances CNS delivery while engaging tumor–astrocyte communication or myeloid niche reprogramming. Niche‐targeted therapeutic strategies in lung cancer brain metastasis. Targeting CNS Barriers and Vascular Niches CNS barrier‐directed strategies target the vascular niche of parenchymal brain metastases and the blood–CSF barrier in leptomeningeal disease. These interfaces provide routes for metastatic entry, vascular support, and uneven drug exposure. Ceritinib reduces tumor cell trans‐endothelial invasion by restoring endothelial Claudin‐10 expression and suppressing VEGF‐A/VEGFR2 signaling [78]. Antibody‐mediated blockade of CD276, which is overexpressed on endothelial and mural cells, suppresses brain metastatic growth and increases CD8+ T‐cell infiltration [79]. Targeting CD146 and AXL disrupts pericyte‐like metastatic stem‐cell programs; imprelimab and bemcentinib showed stronger anti‐angiogenic activity than bevacizumab in preclinical models [23]. In leptomeningeal metastasis, SB 290,157 blocks C3aR‐mediated blood–CSF barrier disruption and reduces dissemination [80]. These strategies remain largely preclinical but illustrate how barrier biology can be therapeutically exploited. Beyond direct modulation of barrier biology, advances in nanomedicine in other metastatic cancers provide additional frameworks for improving therapeutic delivery to metastatic lesions. Multifunctional platforms, including lipid and polymeric nanoparticles, extracellular vesicles, biomimetic carriers, and stimuli‐responsive systems, can integrate active targeting, controlled release, and therapeutic co‐delivery [81]. Reprogramming Brain‐Resident Cells Therapeutic reprogramming of glial and immune cells aims to reverse tumor‐supportive niche states. The HSP47 inhibitor Col003 reduces immunosuppressive microglial polarization and enhances anti‐PD‐L1 efficacy [42]. ISO‐1 blocks the MIF‐CD74 axis, promotes antitumor microglial states, and increases radiosensitivity [40]. IFITM1‐enriched oncolytic virus activates microglia through C3 secretion and synergizes with PD‐1 blockade [43]. Astrocyte‐directed strategies are also emerging. Combined targeting of EGFR and gp130 suppresses IL‐11‐mediated immune escape and brain metastatic growth [32]. CHI3L1 released by pSTAT3‐positive astrocytes is associated with invasive growth [34], and silibinin is being evaluated in the SILMET trial to prevent intracranial recurrence after resection (NCT05689619). TIMP1 inhibition combined with immunotherapy may help reverse astrocyte‐mediated CD8+ T‐cell suppression [82]. Targeting Neuron‐Associated Metastatic Support Therapeutic strategies targeting neuron‐associated mechanisms mainly focus on brain‐specific signaling pathways. One approach is to interrupt neurotransmitter‐derived metabolic support. Therapeutic inhibition of STAT6 with AS1517499 reduces GABA‐supported tumor cell invasion and prolongs survival [83]. The non‐competitive mGluR1 inhibitor LY456236 blocks glutamate–mGluR1‐mediated EGFR stabilization and suppresses brain metastasis [29]. Thus, neurotransmitter‐directed therapy may be most effective in molecularly defined or spatially restricted brain‐adaptive niches. Another strategy is to disrupt direct communication between neurons and tumor cells. In SCLC, the antiepileptic drug levetiracetam reduces synaptic vesicle release and suppresses tumor growth [53]. In addition, the glutamate release inhibitor riluzole and the GRM8 agonist DCPG delay SCLC progression and prolong survival, and riluzole further enhances the efficacy of cisplatin‐etoposide therapy [84]. Overall, these approaches are attractive because some agents are CNS‐active or already have a clinical‐use background. However, the current evidence is mainly preclinical, so they should be regarded as experimental strategies for neural‐adapted brain metastases. Targeting Brain‐Adaptive Metabolic Vulnerabilities Metabolism‐targeted therapy aims to disrupt dependencies that support tumor survival, immune escape, and brain adaptation. Gamitrinib inhibits mitochondrial OXPHOS and suppresses brain metastasis organoids; in preclinical models, it prolongs survival when combined with anti‐PD‐1 therapy [6]. The FASN inhibitor TVB‐2640 inhibits palmitic acid synthesis and suppresses NSCLC brain metastatic growth in preclinical models [13]. Engineering Niche‐Informed CNS Delivery Platforms CNS drug delivery remains a major barrier. ADCs can achieve intracranial activity in selected molecular subtypes but remain constrained by BTB permeability, antigen expression, and micro‐metastatic disease. HER3‐DXd produced a CNS objective response rate (ORR) of 33.3% in patients with nonirradiated measurable brain metastases in HERTHENA‐Lung01 and an intracranial ORR of 30.0% in patients with active brain metastases in TUXEDO‐3 [85, 86]. In HER2‐mutant NSCLC, T‐DXd achieved a 50% confirmed intracranial ORR [87]. Emerging platforms are dedicated to improving drug delivery in the CNS through vector optimization, targeted modification, and alternative pathways. Anti‐EGFR scFv‐functionalized exosomes deliver siLPCAT1 and inhibit metastasis by suppressing the PI3K signaling pathway [88]. Peptide‐drug conjugates have enhanced drug penetration into the brain and tumors in preclinical models [89]. Other approaches combine drug delivery with microenvironmental remodeling, such as SR‐B1‐targeted microparticles that deliver USP7 inhibitors to reprogram M2‐like macrophages [90]. Treatment Resistance Despite advances in targeted therapy and immunotherapy, treatment resistance remains a major barrier to durable benefit. Resistance in LCBM arises from tumor‐intrinsic molecular alterations, microenvironmental remodeling, and BBB heterogeneity. Intrinsic Tumor Molecular Resistance Acquired resistance reflects both tumor‐intrinsic evolution and CNS‐specific pharmacologic and microenvironmental selection. Intracranial progression on the background of osimertinib treatment can involve EGFR‐dependent changes, MET or HER2 amplification, bypass pathway activation, or transformation into a less‐responsive histology. FLAURA2 showed that first‐line osimertinib plus platinum–pemetrexed improved CNS disease control compared with osimertinib alone, although intracranial progression was not eliminated [91]. Early ALK inhibitors have poor CNS penetration; later failure can therefore reflect single or even multiple ALK mutations. The next‐generation ALK inhibitors, for instance, NVL‐655, have CNS‐penetrating activity and are developed to address CNS exposure and complex mutant resistance [92]. Thus, CNS‐active targeted therapy can reshape relapse patterns but does not eliminate persistent clonal evolution within intracranial disease. Brain‐Niche‐Mediated Therapeutic Escape Interactions between tumor cells and microenvironmental cells can weaken antitumor immunity by shaping an immunosuppressive niche through cytokine signaling and metabolic reprogramming. Although TKIs can suppress tumor cells, they may also increase CTLA‐4 expression on T cells, promoting T‐cell exhaustion and therapeutic resistance [93]. These microenvironmental pressures provide a framework for understanding shared and treatment‐specific mechanisms of LCBM failure. Radiotherapy resistance may involve residual clonogenic cells, hypoxia, and local immunosuppression; immunotherapy resistance is associated with T‐cell dysfunction, myeloid suppression, and impaired antigen presentation; ADC efficacy can be limited by heterogeneous BTB permeability and antigen heterogeneity; and metabolic reprogramming may reduce sensitivity to cytotoxic therapy [47, 62, 94, 95]. Therefore, durable control of brain metastases requires integrated consideration of tumor genotype, CNS drug exposure, and the brain microenvironmental state. Future Perspectives Durable control of LCBM remains challenging because intracranial progression is shaped by both tumor genotype and the dynamic CNS ecosystem. Future work should move beyond local treatment of visible radiographic lesions and place greater focus on early detection, mechanism‐guided treatment, and sustained intracranial disease control. From Visible Lesions to CNS Minimal Residual Disease At present, clinical assessment and response monitoring of LCBM still rely heavily on contrast‐enhanced MRI. However, micro‐metastatic deposits, perivascular residual tumor cells, dormant cells, and even early leptomeningeal dissemination might already exist before a radiographic change is evident. Future monitoring strategies should therefore integrate higher‐resolution imaging with cerebrospinal fluid ctDNA, extracellular vesicles, metabolites, protein biomarkers, and long‐term follow‐up. Such an approach may aid early detection of dissemination and residual disease in the CNS. From Single Mechanisms to Actionable Brain Metastatic Niche States Various brain‐resident cell types and metabolic programs contribute to LCBM progression, but most have been studied individually. For this reason, this field still lacks a unified picture able to predict treatment responses or recurrence risk. Current spatial omics studies have also remained mostly descriptive and static, without any insight into microenvironmental evolution over time before and after therapy. Neuronal regulation is especially ripe for future work. First, how neurotransmitters and synaptic activity affect immune cells in the metastatic niche remains unclear. Second, whether lung cancer cells with different genetic backgrounds remodel neural circuits in different ways remains unknown. Third, the mechanisms by which brain metastases cause neurological injury and cognitive decline also need to be better understood, because even small brain metastatic lesions can cause cognitive impairment. Future studies should integrate functional neuroimaging, electrophysiology, spatial immune‐metabolic profiling, and cognitive evaluation. Doing so will help clarify how metastatic lesions disturb brain function, guide LCBM research from individual mechanism descriptions to a clinically relevant definition of brain‐metastatic ecosystem states. Preclinical research also requires models that more realistically reflect the natural development of metastasis. Existing models each have distinct advantages, but none represents the full multistep cascade of brain metastasis. Models of spontaneous brain metastasis may help solve this limitation. An orthotopic SCLC PDX model can generate recurrent and symptomatic brain metastases from primary lung tumors and shows specific adaptive changes [96]. Such models are appropriate for interrogating early dissemination, brain organotropism, neural adaptation, and anti‐metastatic strategies. However, they are limited by long experimental timelines, cost, and scalability. Future efforts should focus on reproducible, image‐trackable spontaneous models of LCBM that, as much as possible, reproduce immune and brain environment properties. Such models will better connect mechanistic science to therapeutic testing relevant to cancer in patients. From Short‐Term Radiographic Response to Durable Intracranial Control Existing clinical trials often exclude patients with active brain metastases or leptomeningeal disease. This is why the CNS objective response rate alone cannot define meaningful long‐term benefit. Future trials will need to place greater emphasis on endpoints such as intracranial progression‐free survival, time to intracranial progression, duration of intracranial response, leptomeningeal control, neurocognitive function, and quality of life. For high‐risk populations, risk‐stratified preventive strategies are also needed to identify which patients may benefit from earlier CNS‐directed intervention. In addition, trial designs should clearly report CNS‐specific eligibility criteria, prior radiotherapy exposure, and lesion measurability. Only when these background factors are explicitly defined can intracranial benefit from systemic therapy be interpreted appropriately and distinguished from the effects of patient selection or previous local treatment. From Empirical Combination Therapy to Mechanism‐Guided Sequencing Future strategies should move toward mechanism‐guided combinations and sequencing. For driver‐positive disease, genotype, CNS drug exposure, and residual local risk should be considered together. For immunotherapy, brain‐infiltrating T‐cell states, myeloid suppression, and corticosteroid exposure should be evaluated. For ADCs and delivery platforms, antigen expression, BTB permeability, and payload sensitivity should be incorporated. Precision therapy for LCBM should integrate tumor mutations with CNS pharmacology and brain‐niche state. From Mechanism‐Guided Therapy to Integrated Clinical Care Brain‐computer interfaces (BCIs) and related neuro‐bioelectronic platforms are still exploratory in LCBM. Their current relevance is closer to intraoperative functional mapping, perioperative neuro‐monitoring, neurorehabilitation, and longitudinal assessment of neurological function than to direct antitumor therapy [97]. Moreover, experimental bioelectronic systems integrating remotely controlled photothermal nanomedicine have shown proof‐of‐concept therapeutic potential in brain tumor models [98]. A clinical trial is evaluating graphene‐based cortical electrodes for real‐time signal capture during brain tumor surgery (NCT06368310). These studies suggest that neural interfaces may eventually contribute to intraoperative mapping, perioperative monitoring, or device‐assisted local therapy. Future work should therefore define clinically relevant signals, safety standards, integration with neurosurgical and radiotherapy workflows, and trial endpoints before BCIs can be considered part of multidisciplinary LCBM management. The translation of mechanism‐guided LCBM therapy will require better drugs and biomarkers and structured clinical coordination. The multidisciplinary team (MDT) should be defined as a practical decision‐making framework rather than a general statement of multidisciplinary care. It involves coordinated input from neurosurgery, radiation oncology, medical oncology, radiology, rehabilitation, and supportive care (Figure 4). Initial management should integrate neurological symptoms, lesion number and size, edema, corticosteroid requirement, molecular subtype, extracranial disease burden, prior radiotherapy, and the expected CNS activity of systemic agents. This framework may help clinical teams select treatment strategies in a more systematic manner. During follow‐up, MDT involvement remains important for distinguishing intracranial progression from radiation necrosis, guiding corticosteroid tapering, monitoring neurocognitive function, and making salvage treatment decisions after oligo‐progression or diffuse intracranial recurrence. A statewide cohort study found that greater hospital connectedness was associated with lower inpatient mortality, whereas the association with hospital length of stay was non‐linear among patients with brain metastases [99]. Standardized MDT workflows may therefore shorten decision‐making time and improve the sequencing of local and systemic therapies. Multidisciplinary collaborative diagnosis and treatment model and process for lung cancer brain metastases. The role of multidisciplinary teams in lung cancer brain metastases (blue section, top). Diagnosis and treatment process of patients with lung cancer brain metastases (green part, below). Created with BioRender.com. Conclusion LCBM is driven by dynamic interactions between tumor cells and the CNS microenvironment, including BBB remodeling, glial and immune‐cell reprogramming, neuronal signaling, and metabolic adaptation. These interactions promote intracranial colonization, immune escape, heterogeneous therapeutic exposure, and treatment resistance. Current local and systemic therapies have improved intracranial management, but durable control remains limited by barrier heterogeneity, molecular evolution, and brain microenvironment‐driven therapeutic escape. Emerging mechanism‐informed approaches offer new opportunities, although most remain at the preclinical or translational stage. Future progress will require earlier detection of CNS residual disease, clinically relevant models, CNS‐specific trial endpoints, and mechanism‐guided treatment sequencing. Translating these advances into practice will depend on multidisciplinary decision‐making that integrates tumor genotype, brain microenvironmental state, and neurological function. Author Contributions Jun Yang: writing – original draft, writing – review and editing, conceptualization. Junyi Xiong: writing – original draft, writing – review and editing, formal analysis, methodology. Li Ying: conceptualization, validation. Yanyang Zeng: software, validation. Ziqiang Ouyang: validation, data curation, investigation. Mei Liu: conceptualization, validation, visualization. Meihua Li: writing – review and editing, data curation, resources, project administration, supervision. Funding This work was supported by the National Natural Science Foundation of China (82201596, 82260248, and 82560268), the Natural Science Foundation of Jiangxi Province (20232BAB216041, 20242BAB26132, and 20252BAC240017), the Double Thousand Talents Plan of Jiangxi, Jxsq2023101024, and the Key Research and Development Program of Jiangxi Province (20261BCG330025). Ethics Statement The authors have nothing to report. Consent The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Table S1: Clinical evidence for systemic therapy and combined‐modality treatment in lung cancer brain metastasis.