Glioblastoma recurrence: An NGS-centered seed-soil framework Authors: Jun-yi Xu, Jingxian Chen, Wenyan Feng, Guo-zhong Yi Published: 2026 ABSTRACT Highlights•We propose an NGS-centered seed–soil framework for glioblastoma recurrence that unifies diverse observations.•The framework complements existing recurrence models rather than replacing them and remains to be experimentally validated.•Targeting NGS may represent a potential therapeutic strategy for preventing or delaying GBM recurrence. Introduction Despite maximal safe resection followed by radiotherapy and chemotherapy [1], recurrence of IDH-wild-type GBM remains nearly inevitable and represents the principal cause of mortality [2]. Current models of GBM recurrence largely emphasize tumor-intrinsic mechanisms, including therapeutic resistance, cellular plasticity, and clonal evolution [3,4], which explain how tumor cells survive treatment. However, they do not fully explain how residual tumor cells subsequently reconstruct an aggressive recurrent tumor ecosystem within the post-treatment brain—a process rarely observable longitudinally in patients and therefore usually inferred from primary–recurrent comparisons or treatment-induced resistance models. A seed–soil concept has been proposed for postsurgical GBM recurrence, in which residual tumor cells represent the "seed" and the surrounding tumor microenvironment (TME) the "soil" [5,6]. Yet the nature of this soil remains incompletely defined: the postsurgical environment contains diverse stromal, immune, vascular, extracellular-matrix, and neural components, each capable of influencing residual tumor-cell behavior. The central question is thus not simply which residual cells survive, but which components of the surrounding brain provide the signals enabling them to persist and reconstruct recurrent disease. Among these components, neurons are distinctive because they communicate directly and dynamically with glioma cells through NGS [7,8]. Unlike paracrine [9] or extracellular-vesicle [10] signaling, NGS provides activity-dependent synaptic input that directly influences glioma-cell behavior; glutamatergic [7,8] and cholinergic [[11], [12], [13]] NGS regulate calcium signaling, proliferation, invasion, migration, and tumor-network activity in GBM. These observations raise a broader question: could the neuron–tumor interface contribute to the evolution of recurrence rather than merely to primary tumor progression? Because recurrence begins with residual disease after surgical resection, this question must be considered within the postsurgical environment. Here, we examine surgery as the biological starting point of recurrence evolution and ask whether persistent or re-established neuron–glioma communication may interface residual tumor cells with the surrounding brain. We then discuss the neuronal features of recurrent GBM, how NGS-associated signaling may participate in recurrence evolution, and the experimental and therapeutic questions that remain unresolved. Surgery and the postsurgical environment "Seed" cells may possess the ability of forming NGS Maximal safe resection remains the cornerstone of GBM treatment, yet infiltrative tumor cells inevitably remain in the surrounding brain—and these residual cells are not necessarily a random sampling of the primary tumor. GBM exhibits extensive intratumoral heterogeneity, comprising mesenchymal-like (MES-like), astrocyte-like (AC-like), oligodendrocyte progenitor cell-like (OPC-like), neural progenitor cell-like (NPC-like), and neuron-like (NEU-like) states [14,15], as well as additional proliferative and metabolic states defined by pathway activity [16,17]. Spatial studies indicate that MES/AC-like populations are enriched toward the tumor core [18], whereas OPC/NPC/NEU-like populations localize preferentially at the invasive margin [19], exhibit invasive and migratory characteristics [12,19], and participate in neuronal interaction and NGS formation [20,21] (Fig. 1A).Fig. 1Biological basis in primary GBM for the NGS-centered seed–soil framework (A) Potential NGS-forming capacity of recurrence “seed” supported by intratumoral heterogeneity Edge regions are enriched in NEU/OPC/NPC-like cells, which exhibit invasive properties and preferentially participate in NGS formation. Intermediate regions contain GPM, MTC, GPC-like, and cilia-like populations, which are primarily associated with metabolic programs. Core regions are enriched in PPR, MES-like, and AC-like populations and are characterized by proliferation, hypoxia adaptation, angiogenesis, injury responses, and immune-related programs. The preferential localization of NEU/OPC/NPC-like cells at the invasive margin suggests that these residual tumor populations may serve as recurrence “seed” cells following surgical resection. (B) Neurons as special recurrence “soil” supported by neuron–GBM network Primary GBM is integrated into extensive neuronal networks composed of both short-range local Glu-AMPAR projection and long-range Ach-CHRM3 projection originating from ipsilateral and contralateral brain regions. Because neuronal circuits and their axonal projections largely persist after surgical resection, these widespread neuronal elements may constitute recurrence “soil”, providing a permissive neural environment capable of maintaining or re-establishing communication with residual tumor cells.Fig 1 dummy alt text Biological basis in primary GBM for the NGS-centered seed–soil framework (A) Potential NGS-forming capacity of recurrence “seed” supported by intratumoral heterogeneity Edge regions are enriched in NEU/OPC/NPC-like cells, which exhibit invasive properties and preferentially participate in NGS formation. Intermediate regions contain GPM, MTC, GPC-like, and cilia-like populations, which are primarily associated with metabolic programs. Core regions are enriched in PPR, MES-like, and AC-like populations and are characterized by proliferation, hypoxia adaptation, angiogenesis, injury responses, and immune-related programs. The preferential localization of NEU/OPC/NPC-like cells at the invasive margin suggests that these residual tumor populations may serve as recurrence “seed” cells following surgical resection. (B) Neurons as special recurrence “soil” supported by neuron–GBM network Primary GBM is integrated into extensive neuronal networks composed of both short-range local Glu-AMPAR projection and long-range Ach-CHRM3 projection originating from ipsilateral and contralateral brain regions. Because neuronal circuits and their axonal projections largely persist after surgical resection, these widespread neuronal elements may constitute recurrence “soil”, providing a permissive neural environment capable of maintaining or re-establishing communication with residual tumor cells. These observations provide a biological basis for considering residual invasive-margin cells as potential recurrence "seeds" that retain both invasive capacity and the ability to engage the surrounding neural environment. However, this should not imply a single cellular origin of recurrence: GBM recurrence likely arises from heterogeneous residual populations, and the contribution of individual cell states remains unresolved. The postsurgical brain constitutes a dynamic recurrence "soil" Following resection, the surrounding brain contains neurons and axonal projections together with reactive astrocytes, microglia/macrophages, vascular components, extracellular matrix, inflammatory mediators, and trophic signals, all of which may influence residual tumor-cell survival, plasticity, and recurrence evolution. Neurons are nevertheless of particular interest, as they provide direct, activity-dependent communication with glioma cells. Because primary GBM engages both local and long-range neuronal circuits, including contralateral projections [11,12,22], surgical resection cannot eliminate all connected neuronal partners. Thus, resection removes most of the tumor while leaving a neural environment capable of supporting renewed neuron–glioma interactions (Fig. 1B). At the same time, surgery profoundly remodels this environment [23,24]. Tissue injury induces inflammation, reactive gliosis, vascular and extracellular-matrix remodeling, and altered neuronal activity, with potentially opposing effects on NGS formation: injury-associated synaptogenic [25,26] and neurotrophic signals [27,28] may permit synaptic remodeling, whereas physical disruption of tumor–neuron contacts, neuronal or axonal loss, and inhibitory matrix changes [29,30] may restrict new synaptic connections. The postsurgical niche should thus be viewed as a dynamic soil rather than a fixed neuronal environment (Fig. 2).Fig. 2Postsurgical microenvironment and potential regulators of NGS remodeling Following surgical resection, residual GBM cells remain within a dynamically remodeled microenvironment. Reactive astrocytes, altered neuronal activity, neurotrophic factors, and tumor-driven synaptic remodeling may promote NGS formation, whereas physical disconnection, neuronal/axonal loss, reactive gliosis, and inhibitory extracellular-matrix remodeling may restrict NGS. The net effect of the postsurgical niche on NGS formation and function remains not that clear. NGS, neuron–glioma synapse; C1QL1, complement C1q-like 1; BAI3, brain-specific angiogenesis inhibitor 3; RAC1, Ras-related C3 botulinum toxin substrate 1; CSPGs, chondroitin sulfate proteoglycans; ECM, extracellular matrix.Fig 2 dummy alt text Postsurgical microenvironment and potential regulators of NGS remodeling Following surgical resection, residual GBM cells remain within a dynamically remodeled microenvironment. Reactive astrocytes, altered neuronal activity, neurotrophic factors, and tumor-driven synaptic remodeling may promote NGS formation, whereas physical disconnection, neuronal/axonal loss, reactive gliosis, and inhibitory extracellular-matrix remodeling may restrict NGS. The net effect of the postsurgical niche on NGS formation and function remains not that clear. NGS, neuron–glioma synapse; C1QL1, complement C1q-like 1; BAI3, brain-specific angiogenesis inhibitor 3; RAC1, Ras-related C3 botulinum toxin substrate 1; CSPGs, chondroitin sulfate proteoglycans; ECM, extracellular matrix. An NGS-centered seed–soil framework emerges from the postsurgical setting The coexistence of potential recurrence seeds and a persistent but remodeled neural environment provides the context for an NGS-centered interpretation of recurrence. We propose that residual tumor cells represent the potential "seed", the postsurgical brain the broader "soil", and neurons a specialized neural component of that soil, with NGS serving as a potential interface between the two. This framework implies neither that neurons are the sole determinants of recurrence nor that postsurgical NGS formation is necessarily enhanced. Rather, it proposes the neuron–tumor interface as one potentially important interaction through which the postsurgical environment may influence residual-cell evolution. The central question is whether NGS persists or becomes re-established after surgery and, if so, whether it contributes to the survival, adaptation, organization, or dissemination of residual tumor cells. Several testable predictions follow. Recurrent GBM would be expected to exhibit enhanced neuronal and NGS-associated features; treatment may remodel neuron–glioma communication during the transition from primary disease to recurrence; and NGS-associated processes may contribute to reconstruction of local tumor systems or dissemination to distant sites. These propositions should be distinguished from established observations; the current evidence and remaining uncertainties are summarized in Table 1.Table 1Evidence, interpretations, and untested predictions of the NGS-centered seed–soil framework.Table 1 dummy alt textEstablished observationsInterpretive propositionsUntested predictionsIntratumoral heterogeneityResidual NEU/OPC/NPC-like cells as “seed”NGS-dependent recurrence initiationNeuron–GBM networkNeurons as a special component of “soil”Neuron-dependent post-surgical nicheNGS formation and signalingNGS as the seed–soil interfaceNGS-dependent recurrenceIncreased NGS and neuronal signaling in recurrent GBMNGS-associated recurrence evolutionNGS precedes recurrenceTherapy-associated NGS remodelingNGS-associated treatment adaptationNGS-mediated therapeutic resistanceGBM network and TM signalingNGS–TM complementary rolesStage-specific NGS–TM interactionNeuronal-like migrationNGS-associated distant disseminationNGS-driven distant recurrence Table 1Evidence, interpretations, and untested predictions of the NGS-centered seed–soil framework.Table 1 dummy alt textEstablished observationsInterpretive propositionsUntested predictionsIntratumoral heterogeneityResidual NEU/OPC/NPC-like cells as “seed”NGS-dependent recurrence initiationNeuron–GBM networkNeurons as a special component of “soil”Neuron-dependent post-surgical nicheNGS formation and signalingNGS as the seed–soil interfaceNGS-dependent recurrenceIncreased NGS and neuronal signaling in recurrent GBMNGS-associated recurrence evolutionNGS precedes recurrenceTherapy-associated NGS remodelingNGS-associated treatment adaptationNGS-mediated therapeutic resistanceGBM network and TM signalingNGS–TM complementary rolesStage-specific NGS–TM interactionNeuronal-like migrationNGS-associated distant disseminationNGS-driven distant recurrence Evidence, interpretations, and untested predictions of the NGS-centered seed–soil framework. Recurrent GBM exhibits enhanced neuronal characteristics One notable feature of recurrent GBM is the enrichment of neuronal and synaptic characteristics, including increased presynaptic contacts, a higher number of NGS per tumor cell, and axonal elongation [31]. Consistently, neuronal-associated markers, including MAP2, SYN1, and DAMM1, are upregulated in recurrent tumors [31]. At the cellular level, neurons [32,33], oligodendrocytes [32,33], and NEU-like tumor cells [31] all increase during recurrence, creating a cellular context that may be permissive for enhanced neuron–glioma communication. Transcriptomic analyses further support increased neuron–tumor interaction in recurrent disease. The invasive margin of recurrent GBM shows greater enrichment of synapse-associated pathways than primary GBM [32]. Moreover, recurrent NEU-like tumor cells are predicted to express NRXN proteins capable of interacting with neuronal NLGNs, with stronger predicted interaction than in primary tumors [33]. Because NRXN–NLGN interactions contribute to synaptic organization, these findings are consistent with enhanced NGS-associated remodeling during recurrence. A potential challenge to a neuron-centered interpretation of recurrence is that recurrent GBM also frequently exhibits mesenchymal characteristics [32], including enrichment of mesenchymal tumor populations, increased myeloid infiltration, and broader remodeling of the immune microenvironment [34,35]. Neuronal and mesenchymal phenotypes, however, are not mutually exclusive and can coexist within recurrent tumors [32] (Fig. 3). One possibility is a spatial model, in which neuronal programs are preferentially enriched at invasive or neuron-rich regions, whereas mesenchymal programs are more prominent in hypoxic, immune-enriched, or otherwise remodeled tumor regions. A temporal model, in which neuronal characteristics precede later mesenchymal enrichment during recurrence evolution, is also conceivable. However, this sequential interpretation remains untested and should not be considered an established trajectory. Distinguishing between spatial coexistence and temporal evolution will require longitudinal sampling of matched primary–recurrent tumors and, ideally, lineage-tracing approaches.Fig. 3Microecosystem difference between primary GBM and recurrent GBM The left panel depicts the microecosystem of primary GBM, and the right panel illustrates that of recurrent GBM. At recurrence, the invasive margin is dominated by a neuronal phenotype, characterized by increased NGS and accompanied by marked expansion of neurons, oligodendrocytes, and NEU-like tumor cells. In contrast, the tumor core predominantly exhibits a mesenchymal phenotype, defined by substantial increases in myeloid cells and T cells. Notably, these two phenotypes can coexist within the same recurrent GBM.Fig 3 dummy alt text Microecosystem difference between primary GBM and recurrent GBM The left panel depicts the microecosystem of primary GBM, and the right panel illustrates that of recurrent GBM. At recurrence, the invasive margin is dominated by a neuronal phenotype, characterized by increased NGS and accompanied by marked expansion of neurons, oligodendrocytes, and NEU-like tumor cells. In contrast, the tumor core predominantly exhibits a mesenchymal phenotype, defined by substantial increases in myeloid cells and T cells. Notably, these two phenotypes can coexist within the same recurrent GBM. Notably, although mesenchymal states have long dominated models of GBM recurrence [36,37] and have traditionally been linked to aggressiveness and poor prognosis [38], neuronal signatures likewise associate with invasive behavior and unfavorable outcomes [39]. NGS-associated processes during recurrence evolution How might the postsurgical setting and the neuronal features of recurrent GBM be connected over time? We propose that neuron–glioma communication may be involved at several stages of recurrence evolution, from therapy-associated NGS remodeling to local network reconstruction and distant dissemination. Importantly, the following model integrates established experimental observations with hypothesis-driven interpretations; it does not imply that NGS has been proven to cause clinical recurrence. Therapy-associated remodeling of NGS Standard therapy represents a major transition point between primary disease and recurrence. In patient-derived xenograft models, prolonged TMZ exposure is accompanied by enrichment of synapse-associated transcriptional programs and increased neuronal characteristics [31]. Single-cell analyses further show an increased proportion of NPC-like tumor cells following TMZ treatment [40]. Because NPC-like populations overlap with residual tumor states proposed to function as recurrence “seeds”, these observations suggest that TMZ may enrich tumor populations with greater potential for neuron–glioma interaction. Radiotherapy appears to produce a more direct remodeling of the neuron–glioma interface. It could increase the number of neuronal connections per tumor cell and enhance NGS connectivity [12]. Radiation-associated neuronal hyperexcitability and action-potential bursts further indicate reciprocal changes in neuronal and tumor-cell activity. Pharmacological inhibition of AMPAR signaling reduces the radiation-associated increase in NGS and attenuates tumor progression in experimental co-culture and in vivo models [12]. Although these findings provide functional evidence that radiotherapy can remodel neuron–glioma connectivity, they derive from a single preliminary study and require independent validation. Within the proposed framework, therapy-induced NGS remodeling may represent a transition between treatment and recurrence: residual cells must survive therapeutic stress, and NGS-associated Ca²⁺ signaling provides biologically plausible survival inputs [41] (Fig. 4A). However, direct evidence that NGS itself confers therapeutic resistance remains limited; NGS-associated calcium signaling should therefore be regarded as a plausible mechanism of treatment adaptation rather than an established mechanism of resistance.Fig. 4Molecular and cellular basis of neuron–glioma synapses and tumor microtube networks (A) NGS-mediated neuronal input through glutamatergic, GABAergic, and cholinergic signaling regulates Ca²⁺ dynamics, membrane excitability, and tumor-cell behaviors, including invasion, migration, and proliferation. GABAergic NGS are shown based on evidence from diffuse midline glioma (DMG) [42]. (B) Tumor microtube (TM)-associated networks connect multiple GBM cell states and astrocytes through gap junctions, enabling exchange of Ca²⁺, IP3, ATP, and metabolites and promoting synchronized cellular responses, stress adaptation, collective survival, and therapy resistance. NGS, neuron–glioma synapse; TM, tumor microtube; AMPAR, α-amino-3‑hydroxy-5-methyl-4-isoxazolepropionic acid receptor; GABAAR, γ-aminobutyric acid type A receptor; NKCC1, Na⁺–K⁺–2Cl⁻ cotransporter 1; CHRM3, cholinergic muscarinic receptor 3.Fig 4 dummy alt text Molecular and cellular basis of neuron–glioma synapses and tumor microtube networks (A) NGS-mediated neuronal input through glutamatergic, GABAergic, and cholinergic signaling regulates Ca²⁺ dynamics, membrane excitability, and tumor-cell behaviors, including invasion, migration, and proliferation. GABAergic NGS are shown based on evidence from diffuse midline glioma (DMG) [42]. (B) Tumor microtube (TM)-associated networks connect multiple GBM cell states and astrocytes through gap junctions, enabling exchange of Ca²⁺, IP3, ATP, and metabolites and promoting synchronized cellular responses, stress adaptation, collective survival, and therapy resistance. NGS, neuron–glioma synapse; TM, tumor microtube; AMPAR, α-amino-3‑hydroxy-5-methyl-4-isoxazolepropionic acid receptor; GABAAR, γ-aminobutyric acid type A receptor; NKCC1, Na⁺–K⁺–2Cl⁻ cotransporter 1; CHRM3, cholinergic muscarinic receptor 3. Reconstruction of a new local tumor system The predominant clinical pattern of GBM recurrence is local regrowth near the original lesion or resection cavity. Within the proposed framework, local recurrence may begin when residual tumor “seeds” remain within the postsurgical niche and progressively reconstruct a new tumor system. Recurrent GBM contains multiple tumor-cell states and re-establishes an interconnected GBM network. NGS may contribute to this process at several levels. First, neuronal signaling may help restore intratumoral heterogeneity during recurrence [43]. At the network level, tumor cells with strong neuronal synaptic connectivity extend elongated processes corresponding to tumor microtubes (TM) [44]. Glu–AMPAR signaling promotes Ca²⁺ transients and TM formation [7], whereas Ach–CHRM3 signaling induces Ca²⁺ oscillations and enhances coordinated tumor-cell activity [11]. These observations provide experimental support for a link between neuronal input and reconstruction of tumor-cell connectivity. The resulting TM-associated GBM network provides an internal communication system among multiple tumor cells and supports coordinated calcium and electrical signaling through gap junctions and adhesion junctions [[45], [46], [47]]. Once established, this network contributes to adaptive responses following chemotherapy and radiotherapy [48] (Fig. 4B). An important implication is that NGS and TM networks may represent complementary rather than strictly linear mechanisms. In the early postsurgical phase, when much of the pre-existing tumor network has been disrupted by resection, NGS may provide external neural input to residual cells. As the tumor expands and the recurrent ecosystem is reconstructed, TM networks may provide additional internal communication and network-level coordination. This temporal relationship remains a proposed model requiring longitudinal validation. Neuronal-like migration and distant recurrence Distant recurrence represents a distinct spatial trajectory in which residual tumor cells disseminate beyond the immediate vicinity of the original lesion before establishing recurrent growth. In GBM, invasive cells are enriched for NEU/OPC/NPC-like states and exhibit branching migration, locomotion, and translocation patterns resembling developmental neuronal migration [19]. Emerging evidence suggests that neuronal activity and NGS-associated signaling can enhance these migratory behaviors. Neuronal inputs from distant brain regions promote glioma invasion [49], while tumor cells with greater synaptic connectivity exhibit increased invasion in experimental spheroid models [44]. Glu–AMPAR signaling promotes Ca²⁺-dependent TM dynamics and migration [19], whereas Ach–CHRM3 signaling induces a motile cellular state associated with increased invasion [11]. These findings provide a biological basis for proposing that NGS-associated signaling may facilitate neuronal-like migration of residual tumor cells. However, migration should not be equated with distant recurrence. Dissemination is only one component of a multistep process that also requires survival during transit, adaptation to the new microenvironment, proliferation, and reconstruction of a supportive tumor ecosystem at the ectopic site. Thus, neuronal-like migration should be considered a potential contributor to distant dissemination rather than a demonstrated mechanism of distant recurrence. An integrated model of recurrence evolution Taken together, local and distant recurrence may represent different spatial outcomes of a common postsurgical evolutionary process: cells retained near the resection cavity progressively reconstruct a local tumor system, whereas a subset of residual cells undergoes neuronal-like dissemination to distant sites (Fig. 5). The model thus does not require a single obligatory NGS-to-recurrence pathway; rather, NGS-associated processes may influence several stages of recurrence evolution, providing a conceptual link between established neuron–glioma biology and the spatial patterns of recurrent GBM. The causal contribution of NGS to clinical recurrence remains an open question.Fig. 5Potential NGS-associated processes underlying local and distant recurrence Following surgical resection, residual NEU/OPC/NPC-like tumor cells persist at the invasive margin and retain the capacity to establish NGS with surrounding neurons. Under the influence of neuronal activity and therapy-enhanced neuron–glioma communication, these residual cells may follow two spatial trajectories. In local recurrence, residual cells expand in situ and progressively reconstruct a new local tumor system characterized by renewed intratumoral heterogeneity, GBM network formation, and therapeutic resistance. In distant recurrence, a subset of residual cells may first disseminate to anatomically distant brain regions through neuronal-like migration and subsequently establish new tumor systems at ectopic sites.Fig 5 dummy alt text Potential NGS-associated processes underlying local and distant recurrence Following surgical resection, residual NEU/OPC/NPC-like tumor cells persist at the invasive margin and retain the capacity to establish NGS with surrounding neurons. Under the influence of neuronal activity and therapy-enhanced neuron–glioma communication, these residual cells may follow two spatial trajectories. In local recurrence, residual cells expand in situ and progressively reconstruct a new local tumor system characterized by renewed intratumoral heterogeneity, GBM network formation, and therapeutic resistance. In distant recurrence, a subset of residual cells may first disseminate to anatomically distant brain regions through neuronal-like migration and subsequently establish new tumor systems at ectopic sites. Discussion and future perspective The necessity and sufficiency of NGS remain unresolved The most important challenge is establishing causality—and a related limitation must be acknowledged: most evidence for neuron–glioma communication derives from intact-brain xenograft models, whereas clinical recurrence arises in the post-surgical resection cavity, a markedly different microenvironment (Fig. 2). Encouragingly, this gap is beginning to close. The recent identification of a C1QL1–BAI3–RAC1 axis was established in a post-resection recurrence model [50]: GBM-secreted C1QL1 promotes tumor-microtube expansion, synaptic pruning, and malignant synapse formation after surgical removal of the tumor bulk, while pharmacological RAC1 inhibition impedes recurrence in this setting. This study provides the first direct evidence that neuron–tumor network remodeling remains operative—and therapeutically targetable—within the post-surgical niche. Nevertheless, RAC1 is a broadly acting regulator rather than an NGS-specific target; its inhibition therefore does not establish that the antitumor effect arises specifically from NGS disruption. More direct tests are needed: demonstrating necessity requires selective disruption of neuron–glioma synaptic communication followed by delayed recurrence, whereas sufficiency requires that experimental enhancement of NGS accelerate recurrence under otherwise comparable conditions. Critically, such experiments must be performed in faithful post-resection models that reproduce the wound-healing niche—models that are currently scarce and whose development should be a priority for the field. Therapeutic implications: targeting the tumor–brain interface in the perioperative window The framework broadens the therapeutic landscape beyond tumor-intrinsic targets. Pharmacological strategies against Glu/Ach signaling or synaptic receptors are one approach, although systemic disruption of neuronal signaling risks interfering with physiological brain function, making selective targeting of tumor-associated synaptic interactions particularly important. Neuromodulation provides a complementary direction: a recent study of an eye–brain neuroimmune stimulation axis prolonged survival in a murine GBM model with reduced pathological neuron–glioma connectivity [51] —proof-of-concept, though not yet a clinical strategy, that modulating neural activity can influence GBM progression. The perioperative period may be a particularly informative window for such interventions, as the recurrent ecosystem is then still being established. The ongoing PerSurge (NOA-30) phase II trial [52] exemplifies this logic, evaluating the AMPAR antagonist perampanel around re-resection in progressive GBM with tumor network connectivity as a translational endpoint. Future perioperative studies should move beyond tumor-growth measurements to directly assess NGS abundance, synaptic connectivity, residual-cell states, and longitudinal recurrence, thereby providing the temporal evidence currently missing between surgery, NGS remodeling, and recurrence. Author Statement All authors read and approved the final manuscript. Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Funding This study was supported by the National Natural Science Foundation of China (82272636), Clinical Research Fund of Nanfang Hospital, Southern Medical University (2024CR008), Research Fund of Nanfang Hospital, Southern Medical University (2024B019), National College Student Innovation and Entrepreneurship Training Program (S202412121162). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Availability of data and materials Not applicable. CRediT authorship contribution statement Jun-yi Xu: Writing – original draft, Visualization, Investigation, Data curation, Conceptualization. Jingxian Chen: Writing – review & editing, Visualization, Investigation, Conceptualization. Wenyan Feng: Writing – review & editing, Supervision, Funding acquisition. Guo-zhong Yi: Writing – review & editing, Supervision, Funding acquisition. Declaration of competing interest The authors declare that they have no competing interests.