The core initiating event of astrocytic ferroptosis is the lipid peroxidation cascade triggered by disrupted intracellular iron homeostasis. Significantly elevated intracellular free iron (Fe²⁺) and iron overload have been observed in astrocytes across multiple neurological disorder models, including ischemic stroke, SAH, and NMOSD (20,27,28). This iron metabolic disturbance (29) activates the Fenton reaction, catalyzing the production of abundant hydroxyl radicals, which attack cell membrane phospholipids rich in polyunsaturated fatty acids (PUFAs), triggering uncontrolled lipid peroxidation (30-32). In an NMOSD model, AQP4-IgG stimulation was shown to upregulate ACSL4 expression, promoting PUFA incorporation into membrane phospholipids and significantly increasing cellular susceptibility to ferroptosis; early growth response 1 (Egr1) small interfering RNA (siRNA) intervention reversed ACSL4 upregulation and attenuated ferroptosis (20). Additionally, in a PD model, astrocytic iron deposition was found to be closely associated with abnormal mitochondrial oxidative phosphorylation, further exacerbating lipid peroxidative injury (23). Notably, under ischemia/reperfusion conditions, elevated palmitic acid was revealed to promote palmitoylation of transferrin receptor 1 (TfR1), exacerbating astrocytic iron uptake and overload via clathrin-mediated endocytosis (27). Collectively, these findings demonstrated that iron metabolic disturbance constitutes the initial trigger of astrocytic ferroptosis by driving lipid peroxidation.

NOX4 serves as a key pro-oxidant mediator in astrocytic ferroptosis, with its sustained production of ROS significantly amplifying lipid peroxidative injury. In an AD model, scRNA-seq data revealed specifically high expression of NOX4 in astrocytes; pharmacological or genetic inhibition of NOX4 not only attenuated ferroptosis but also improved cognitive function, reduced Aβ and p-Tau levels, and alleviated mitochondrial structural and functional abnormalities (15). In the context of AD pathogenesis (33), amyloid-β (Aβ) peptides, generated through amyloidogenic processing of the amyloid precursor protein (APP), aggregate into extracellular plaques that disrupt synaptic function and activate microglia. Hyperphosphorylated Tau (p-Tau), in turn, forms intracellular neurofibrillary tangles that destabilize microtubules and impair axonal transport. Together, Aβ accumulation and p-Tau pathology synergistically drive neurodegeneration, and the observed reduction of both upon NOX4 inhibition underscores the functional link between ferroptosis-related oxidative stress and the core pathological hallmarks of AD (15,34). Similarly, in a rat model of collagenase-induced ICH, NOX4 expression was found to be upregulated in the perihematomal tissue, accompanied by activation of the ferritin/transferrin receptor system, collectively promoting oxidative stress and iron deposition; NOX4-siRNA intervention effectively alleviated lipid peroxidation and neuronal ferroptosis, and improved motor function (35). In patients with PD and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-treated mouse models, a widely used chemically-induced PD model in which the neurotoxin MPTP is metabolized to MPP⁺, selectively destroying dopaminergic neurons in the substantia nigra pars compacta and recapitulating key motor features of PD, NOX4 expression was shown to be elevated in hippocampal astrocytes, positively associated with myeloperoxidase (MPO) and osteopontin (OPN) levels. NOX4 directly induces ferroptosis by inhibiting the activity of mitochondrial electron transport chain complexes I-V and increasing lipid peroxidation end products such as 4-hydroxynonenal (4-HNE) (36). Although direct regulation of NOX4 by FoxO1a has not been fully elucidated in the current literature, multiple studies suggest that FoxO1a may participate in upstream regulation of the oxidative stress axis (37-39). As a key executor of oxidative stress, NOX4 expression may be indirectly regulated by FoxO family transcription factors. Furthermore, in a chronic kidney disease model, homeobox D10 (HOXD10) was shown to inhibit NOX4 transcription by directly binding to its promoter, attenuating ferroptosis (40), highlighting the importance of transcriptional regulation in NOX4 expression and providing indirect support for the potential regulation of NOX4 by FoxO1a (Fig. 3).

The GSH-GPX4 axis is the core antioxidant defense system against ferroptosis, and its dysfunction is a decisive step in astrocytic ferroptosis. GPX4 relies on GSH as a reducing equivalent to convert toxic lipid hydroperoxides into harmless alcohols, thereby blocking the lipid peroxidation chain reaction. This pathway is significantly inhibited in multiple neurological disorders. For example, after ischemic stroke, astrocytes tend to upregulate SLC7A11 and GPX4 expression more than neurons, but GPX4 inactivation still occurs under severe pathological conditions (27). In an SAH model, downregulated GPX4 expression was found to directly contribute to astrocytic ferroptosis, and the ferroptosis inhibitor Fer-1 effectively alleviated oxidative stress-induced brain injury (28). Notably, GPX4 stability is also regulated by post-translational modifications: In an ischemic stroke model, oxygen-glucose deprivation/reoxygenation (OGD/R) activated autophagy, and the E3 ubiquitin ligase Kelch like E3 ubiquitin protein ligase 8 (KLHL8) ubiquitinated GPX4, promoting its binding to the selective autophagy receptor Tax1 binding protein 1 (TAX1BP1) and degradation via the autophagy-lysosome pathway; clinical sample analysis showed significant upregulation of KLHL8 and TAX1BP1 in brain tissue of stroke patients, positively correlating with ferroptosis scores (41). Additionally, in an NMOSD model, GSH levels were found to be significantly reduced in astrocytes, accompanied by increased malondialdehyde (MDA) and lactate dehydrogenase (LDH) release, further confirming the key role of GSH metabolic disturbance in ferroptosis (20). In summary, GSH depletion and GPX4 inactivation collectively constitute the terminal step whereby astrocytes lose antioxidant capacity and undergo ferroptosis.

As a pro-inflammatory subtype of reactive astrocytes, A1 astrocytes exhibit distinct molecular markers during phenotypic transformation. Multiple studies confirm that complement component C3 is the most representative and widely recognized marker of A1 astrocytes, significantly upregulated in multiple neurological disorder models including ischemic stroke, AD, epilepsy, and TBI (42-44). Additionally, major histocompatibility complex class I molecules H2-D1 and H2-T23, and the serine protease inhibitor Serping1, have been identified as key transcriptional markers of A1 transformation (45). At the functional level, A1 astrocytes exhibit downregulated expression of neuroprotection-related genes such as S100 calcium binding protein A10 (S100A10) (26,29), and elevated expression of activation markers including GFAP and S100 calcium binding protein B (S100B) (22,46). Notably, the expression of A1 markers exhibits regional heterogeneity across different pathological contexts; for example, friend leukemia integration 1 (Fli-1) expression has been shown to be elevated and correlated with increased A1 numbers in the hippocampus of 5xFAD mice and patients with AD (42), while in an SAH model, neogenin 1 (NEO1) was revealed to drive A1 polarization via the cytosolic phospholipase A2 (cPLA2)-mitochondrial antiviral-ignaling protein (MAVS)-NF-κB pathway (47). These molecular markers not only reflect the pro-inflammatory state of A1 astrocytes but also represent potential biomarkers for targeted intervention.

The transformation of astrocytes into the A1 phenotype can be triggered by diverse stimuli beyond a single environmental factor. While this section initially focused on cadmium exposure as a specific trigger, it is important to contextualize this within the broader spectrum of known A1-inducing factors.

Environmental heavy metal exposure specifically induces the transformation of astrocytes into a pro-inflammatory A1 phenotype. Research shows that cadmium exposure triggers A1 transformation in chick striatal astrocytes, with the key mechanism involving internalization of connexin 43(48). Connexin 43 is a critical component of astrocytic gap junctions, and its internalization not only disrupts intercellular communication homeostasis but may also trigger downstream inflammatory signaling cascades. This process is accompanied by upregulation of typical A1 markers and activation of neuroinflammation, providing novel cytological evidence for cadmium-induced neurotoxicity (48). Although this study does not fully elucidate whether cadmium mediates transformation indirectly via microglia, combined with the classical mechanism of TNF-α, IL-1α, and C1q as tripartite inducers of A1 transformation in other models (42), cadmium may promote A1 polarization by directly damaging astrocytes or activating adjacent microglia to release these cytokines. This finding emphasizes the role of environmental toxins in initiating neuroinflammation and suggests that connexin 43 may be a potential target for intervening in cadmium-related neurodegenerative diseases.

Beyond cadmium exposure, multiple other factors have been demonstrated to trigger A1 astrocyte transformation. The classical induction pathway involves activated microglia releasing the triad of TNF-α, IL-1α, and C1q, which together convert resting astrocytes into the A1 phenotype (42,49). Additional triggers include ischemic stroke-induced inflammatory cascades (44), lipopolysaccharide (LPS)-mediated TLR activation (50), human immunodeficiency virus type 1 transactivator of transcription (HIV-1 Tat) protein exposure (51), and chronic neurodegenerative protein aggregates such as Aβ and α-synuclein (52). Cardiac arrest has also been shown to trigger IL-17-mediated neuroinflammation and astrocyte polarization (53). Environmental factors such as chronic stress and heavy metal exposure (including lead and mercury) have also been implicated in A1 induction (48). Cadmium is highlighted in the present review because it provides a particularly well-characterized mechanistic model, demonstrating how environmental toxicants directly modulate astrocyte-specific gap junction proteins (connexin 43) to initiate A1 transformation through a pathway distinct from the canonical cytokine-mediated mechanism. This unique mode of action offers complementary insights into how non-cytokine stimuli can drive astrocyte phenotypic conversion, broadening the therapeutic target landscape beyond classical inflammatory signaling.

A1 astrocytes lose their normal neuronal support function and instead secrete diverse neurotoxic substances, forming the core of their pathogenicity. Their secretory profile mainly includes pro-inflammatory cytokines, complement system components, and chemokines. Specifically, A1 astrocytes prominently release inflammatory mediators including IL-1β, TNF-α, and IL-6 (53-55) and amplify inflammatory responses via TLR4, receptor for advanced glycation end products (RAGE), and other pathways (22,54). Complement cascade activation is another key feature, with C3 not only serving as a marker but also being cleaved into C3a and C3b to mediate synaptic elimination and neuronal injury (42). Additionally, A1 cells have been shown to highly express the chemokine CXCL10, promoting the infiltration of peripheral immune cells across the blood-brain barrier (BBB) (42). In an ischemic stroke model, A1 astrocytes were found to also secrete vascular endothelial growth factor (VEGF), matrix metalloproteinases (MMPs), and Lcn2, directly disrupting BBB integrity and inhibiting axonal regeneration (56). Notably, some A1 cells exhibit aberrant metabolic activity, such as thioredoxin-interacting protein (TXNIP) overexpression, leading to excessive mitochondrial ROS (mtROS) production, further exacerbating oxidative stress and neuronal apoptosis (57). These secreted factors collectively construct a self-reinforcing neurotoxic microenvironment that drives the progression of neurological disorders (Fig. 4).

Pro-inflammatory cytokines released during neuroinflammation, particularly TNF-α, play a key role in driving astrocytic ferroptosis. In a cerebral ischemia-reperfusion injury model, the inflammatory response was shown to disrupt iron homeostasis and inhibit the antioxidant defense system, promoting lipid peroxidation and inducing ferroptosis (12). TNF-α binds to TNF receptor 1 (TNFR1)/TNF receptor 2 (TNFR2) on the surface of astrocytes, activating downstream signaling pathways, enhancing pro-inflammatory phenotypic transformation, and further reducing cellular resistance to oxidative stress. Studies have shown that in a PD model, the regulator of G Protein signaling 5 (RGS5) protein in astrocytes interacts with TNFR1 and TNFR2 to enhance TNF signaling, leading to chronic neuroinflammation and neuronal injury; inhibiting this interaction significantly reduces astrocytic cytokine release and improves neuroinflammation and neuronal survival (58). Additionally, in the early stage of TBI, the pro-inflammatory microenvironment is accompanied by downregulated GPX4 expression and elevated ACSL4 and arachinodate 15-lipoxygenase (ALOX15) levels, creating conditions conducive to ferroptosis (25). These findings indicate that pro-inflammatory cytokines such as TNF-α act not only as inflammatory mediators but also directly promote astrocytic ferroptosis by regulating iron metabolism and antioxidant pathways.

As an iron-dependent lipid peroxidative cell death, ferroptosis triggers the release of DAMPs, including large endogenous molecules such as HMGB1, ATP, and mtDNA, as well as small-molecule pro-inflammatory mediators including ROS and lipid peroxidation products, which further activate immune cells in the CNS and exacerbate neuroinflammation. It is important to distinguish between these two categories: DAMPs are large endogenous molecules (typically >1 kDa) released upon cell rupture that activate pattern recognition receptors, while ROS and lipid peroxides are small-molecule oxidative mediators that serve as pro-inflammatory triggers but do not fall within the classical DAMP category (14,59). Existing studies demonstrate that DAMPs and pro-inflammatory mediators whose production is triggered by ferroptosis activate microglia and astrocytes, inducing them to release more pro-inflammatory factors and form a vicious cycle (12). In particular, HMGB1 released from ferroptotic cells activates TLR4/NF-κB signaling in adjacent microglia and astrocytes, triggering a cascade of pro-inflammatory cytokine release (60,61). In the spinal cord tissue of patients with ALS and superoxide dimsutase (SOD)1 transgenic mouse models, microglia were shown to undergo sublethal ferroptotic stress, triggering an inflammatory cascade, promoting the transformation of astrocytes into a neurotoxic phenotype, and causing non-cell-autonomous neuronal death; the brain-penetrant ferroptosis inhibitor Cu(atsm) was revealed to effectively alleviate ferroptosis markers and exert neuroprotective effects (62). Furthermore, in an inflammatory pain model induced by complete Freund's adjuvant, ferroptosis in spinal glial cells was found to lead to upregulation of the nuclear receptor nuclear receptor subfamily 4 group A member 1 (NR4A1), which exacerbates lipid peroxidation by promoting mitogen-activated protein kinase 3 (MAPK3) transcription, thereby amplifying neuroinflammation (63). These findings collectively support that DAMPs and pro-inflammatory mediators released in response to ferroptosis are important drivers of sustained neuroinflammation, and that the reciprocal amplification between ferroptotic cell death and inflammatory activation constitutes the self-reinforcing core of the vicious cycle.

The positive feedback cycle between neuroinflammation and ferroptosis exhibits significant spatiotemporal dynamic properties, with heterogeneity across different disease stages and brain regions. After ischemic stroke, astrocytes and microglia rapidly undergo phenotypic transformation, initiating pro-inflammatory responses, accompanied by disrupted iron metabolism and inhibited antioxidant systems, creating conditions for ferroptosis; ferroptosis further triggers the release of DAMPs and pro-inflammatory mediators, reinforcing glial activation and forming a self-amplifying pathological loop (16,64). In an epilepsy model, A1 astrocytes were found to promote GPX4-dependent lipid peroxidation and neuronal ferroptosis by secreting CXCL10 to activate the STAT3 pathway and inhibit neuronal SLC7A11 expression; clinical samples also confirmed the significant co-localization of ferroptosis markers and A1 astrocytes in brain tissue of patients with epilepsy (19). Additionally, in a PD model, impaired fibroblast growth factor (FGF) signaling between oligodendrocytes and astrocytes was revealed to lead to abnormal mitochondrial oxidative phosphorylation and downregulation of the Nrf2/SLC7A11/GPX4 pathway in astrocytes, inducing regional ferroptosis that affects dopaminergic neurons (23). This spatiotemporal specificity indicates that the inflammation-ferroptosis cycle is not uniformly distributed but gradually spreads along specific neural circuits or susceptible brain regions (such as the hippocampus, substantia nigra, and cortex). In a postoperative cognitive dysfunction model, decreased PCBP1 expression in hippocampal astrocytes was shown to increase their susceptibility to ferroptosis, impairing glutamate clearance, interfering with neuronal activity, and inducing depression-like behaviors (65,66), further confirming the functional consequences of this cycle in specific brain regions. In summary, the positive feedback mechanism between neuroinflammation and ferroptosis exhibits a complex network characterized by dynamic evolution, regional selectivity, and cell-type dependence.

As an important member of the transcription factor family, FoxO1a plays a key role in regulating astrocytic oxidative stress responses. Research has shown that under neuroinflammatory conditions, FoxO1a influences cellular susceptibility to lipid peroxidation and ferroptosis by regulating the expression of antioxidant genes. In a letrozole-induced depression-like behavior model in female rats, rebamipide activated SIRT1, thereby upregulating the FoxO/Wnt/β-catenin signaling pathway, enhancing the expression of Nrf2, SOD, and heme oxygenase-1 (HO-1), and inhibiting NF-κB-p65, TNF-α, and NOD-like receptor family pyrin domain containing 3 (NLRP3) inflammasome activation, ultimately reducing iron accumulation and upregulating GPX4 and SLC7A11 levels to inhibit ferroptosis (67). This process could be reversed by the SIRT1 inhibitor nicotinamide, indicating the important role of SIRT1/FoxO signaling in regulating oxidative stress and ferroptosis. In summary, FoxO1a not only participates in regulating astrocytic antioxidant capacity but also interacts with other signaling pathways (such as SIRT1 and Nrf2) to form a critical defense line against ferroptosis.

The NF-κB and Nrf2 pathways exhibit a significant antagonistic relationship in the regulation of neuroinflammation and ferroptosis. Multiple studies (68,69) confirm that Nrf2 activation inhibits NF-κB signaling by enhancing antioxidant gene expression and stabilizing inhibitor of nuclear factor κB-α (IκB-α) protein, thereby blocking NF-κB nuclear translocation; conversely, NF-κB activation inhibits Nrf2 transcriptional activity and weakens cellular antioxidant defense capacity (70). In a PD mouse model, the Nrf2/SLC7A11/GPX4 pathway was significantly inhibited in astrocytes, accompanied by activated ferroptosis and loss of dopaminergic neurons (23). In a mouse cerebral ischemia and primary astrocyte OGD model, icariside II (ICS II) exerted neuroprotective effects by promoting Nrf2 nuclear translocation and effectively regulating the oxidative phosphorylation (OXPHOS)/NF-κB/ferroptosis axis, an effect significantly attenuated in Nrf2-deficient mice or siRNA-treated cells (71). Similarly, sesamin was shown to inhibit ferroptosis and improve oxidative stress and inflammatory responses by activating the Nrf2/SLC7A11/GPX4 pathway (72); Artemisia ordosica polysaccharide (AOP) was also found to reduce ROS, 8-hydroxy-2'-deoxyguanosine (8-OhdG), protein carbonyl (PC) levels, and IL-1β, IL-6 expression by activating Nrf2/Kelch-like ECH-associated protein 1 (Keap1) and inhibiting the TLR4/NF-κB pathway (73). Furthermore, in glioblastoma cells, RSL3-induced GPX4 inhibition triggered ferroptosis while activating NF-κB, and NF-κB inhibition attenuated ferroptosis, suggesting its pro-death role in ferroptosis (74). In ALS models, pharmacological activation of Nrf2 by RTA-408 restored the expression of SLC7A11 and GPX4 and elevated GSH levels, mitigating ferroptosis and conferring neuroprotection both in vitro and in vivo (75). Collectively, the dynamic balance between Nrf2 and NF-κB determines the fate of astrocytes under oxidative stress and inflammatory stimulation, representing a key target for intervening in the neuroinflammation-ferroptosis vicious cycle.

Structural and functional coupling between mitochondria and the ER is increasingly recognized as critical for regulating astrocytic ferroptosis. These organelles mediate calcium exchange, lipid transport, and redox signaling via mitochondria-associated ER membranes (MAMs), collectively maintaining cellular homeostasis. In a PD model, astrocytes were shown to exhibit enhanced mitochondrial oxidative phosphorylation accompanied by inhibition of the Nrf2/SLC7A11/GPX4 pathway, indicating a close association between mitochondrial dysfunction and ferroptosis (23). Additionally, accumulated lipid peroxidation during ferroptosis has been demonstrated to disrupt mitochondrial membrane integrity, releasing pro-inflammatory mediators such as ROS, which activate neuroimmune pathways and promote glial release of IL-1β, TNF-α, and other inflammatory factors, forming a positive feedback cycle (3,76). It should be noted that ROS released from damaged mitochondria function as small-molecule oxidative stress signals rather than classical DAMPs, which are defined as large endogenous molecules released upon cell rupture. ER stress has also been found to trigger pathways such as C/EBP homologous protein (CHOP) and activating transcription factor 4 (ATF4), regulating xCT (SLC7A11) expression and affecting GSH synthesis and GPX4 activity (74). Notably, in an epilepsy model, CXCL10 secreted by A1 astrocytes mediated STAT3 phosphorylation via C-X-C motif chemokine receptor 3 (CXCR3), inhibiting neuronal SLC7A11 expression and leading to GPX4-dependent ferroptosis, a process potentially involving dysregulated mitochondria-ER communication (19). Although direct evidence for mitochondria-ER crosstalk in astrocytes remains limited, existing data strongly suggest that their interaction is a key platform integrating oxidative stress, iron metabolism, and inflammatory signals, representing a novel intervention target to break the vicious cycle (Fig. 5).

In recent years, zebrafish, as an important model organism for studying neural development and disease mechanisms, have provided critical evidence for elucidating astrocytic ferroptosis regulatory pathways. A foxo1a-deficient zebrafish model constructed via CRISPR/Cas9 technology showed that foxo1a gene deletion not only affects hematopoietic system development but also significantly interferes with the expression of oligodendrocyte, astrocyte, and myelin-related markers in the CNS, inducing oxidative stress and ferroptosis (77). Additionally, pharmacological modulation of astrocyte phenotype polarization using rottlerin has been shown to attenuate trimethyltin-induced neurotoxicity (43). Mechanistic studies demonstrate that foxo1a maintains intracellular GSH synthesis and antioxidant capacity by directly binding to and regulating the slc7a11 gene promoter, thereby inhibiting lipid peroxidation and ferroptosis (77). Notably, the foxo1a-slc7a11 axis is highly conserved between zebrafish and human cells, indicating its evolutionary significance in neuroprotection among jawed vertebrates (77). These findings collectively indicate that the foxo1a-deficient zebrafish model provides a powerful tool for dissecting the interactive mechanisms between astrocytic ferroptosis and neuroinflammation.

With the development of high-throughput sequencing technologies, scRNA-seq and spatial transcriptomics have become key tools for dissecting cellular heterogeneity and molecular networks in neurological disorders. In AD research, scRNA-seq data revealed significant upregulation of NOX4 in ferroptotic astrocytes. Similarly, in a PD model, single-nucleus sequencing combined with spatial transcriptomic analysis revealed impaired FGF1/FGF9-FGFR signaling between oligodendrocytes and astrocytes, leading to enhanced mitochondrial oxidative phosphorylation, downregulation of the Nrf2/SLC7A11/GPX4 pathway, and preferential induction of ferroptosis in the substantia nigra, thereby driving dopaminergic neuronal injury (23). In an NMOSD model, single-nucleus RNA sequencing identified AQP4-IgG-induced ACSL4 upregulation as a key step in astrocytic ferroptosis, which could be reversed by Egr1 siRNA (20). Furthermore, in recurrent glioblastoma, single-cell data analysis identified zinc finger protein 36 (ZFP36) as an inhibitor of astrocytic ferroptosis, which influences tumor progression by regulating MAPK and metabolic pathways (78). Recent research has also identified ferroptosis-related astrocyte subpopulations in PD through integrative scRNA-seq and bulk RNA transcriptomics, with diagnostic biomarkers including nuclear paraspeckle assembly transcript 1 (NEAT1), heat shock protein family B (small) member 1 (HSPB1), and heat shock protein family A (HSP70) member 5 (HSPA5) showing promise for non-invasive early detection (79). These studies highlight the irreplaceable value of single-cell and spatial omics technologies in accurately delineating astrocytic phenotypic transformation, ferroptosis drivers, and intercellular communication networks. Furthermore, brain organoid models are increasingly used to simulate human neuroinflammation-ferroptosis interactions, providing a platform closer to the human physiological environment for mechanistic validation and drug screening.

Given that ferroptosis is an iron-dependent, lipid peroxidation-driven form of regulated cell death, the detection of ferroptosis-specific biomarkers is critical for disease diagnosis and therapeutic efficacy evaluation. Currently recognized ferroptosis markers include GSH depletion, decreased GPX4 activity or expression, accumulation of lipid peroxidation products (such as MDA and 4-HNE), and elevated iron levels (20,80). In an epilepsy model, ferroptosis was confirmed by detecting an inactivated GPX4-GSH system and downregulated SLC7A11 expression in neurons, consistent with pathological changes observed in brain tissue of clinical patients (8). In an NMOSD model, ferroptosis was confirmed by multiple indicators including elevated iron concentration, increased MDA and LDH release, decreased GSH levels, and upregulated ACSL4 expression (20). Additionally, mitochondrial shrinkage and increased membrane density observed by transmission electron microscopy are typical morphological features of ferroptosis, used for auxiliary validation in multiple models (81). In recent years, the protective effects of ferroptosis inhibitors such as Fer-1 and liproxstatin-1 have also been used as functional biomarkers, significantly alleviating neural injury and improving behavioral outcomes in animal models, indirectly reflecting ferroptosis activity (80,82). Emerging neuroimaging techniques, including quantitative susceptibility mapping and susceptibility-weighted imaging, enable non-invasive detection of brain iron accumulation in specific regions such as the substantia nigra in PD and the hippocampus in AD, although these methods cannot yet distinguish ferroptosis-specific iron from other forms of deposition (83). Biofluid biomarkers, including cerebrospinal fluid (CSF) lipid peroxide levels, serum GPX4 activity, and the ACSL4/GPX4 ratio, are being validated across multiple disease contexts and hold promise for clinical translation (84). Although fully specific and noninvasively detectable clinical biomarkers remain lacking, multidimensional detection strategies combining oxidative stress indicators, iron metabolism proteins, and lipid peroxidation end products are gradually improving the capacity for monitoring and guiding interventions of ferroptosis in neurological disorders.

The combined strategy of iron chelators and antioxidants aims to simultaneously block the core drivers of ferroptosis, iron overload and lipid peroxidation. This combined intervention has shown significant neuroprotective effects in multiple neurological disorder models. For example, in a mouse model of ischemic stroke, the iron chelator deferoxamine (DFO) combined with antioxidants effectively alleviated astrocytic iron accumulation and neuronal injury, with mechanisms involving inhibition of TfR1 palmitoylation-mediated iron endocytosis and restoration of GSH-dependent antioxidant defense systems (27). Additionally, in an in vitro model of rotenone-induced neurotoxicity, DFO and hydroxytyrosol (HT) co-delivered via a nanocarrier system exhibited synergistic neuroprotective effects, significantly reducing iron deposition and lipid peroxidation (7,85). In PD models, the oral iron chelator deferiprone, which can cross the BBB, has demonstrated safety and efficacy in Phase II clinical trials, reducing substantia nigra iron deposition and slowing motor dysfunction progression in patients with early-onset PD (86). Iron chelation combined with antioxidant supplementation thus addresses both the initiating trigger (iron overload) and the amplifying mechanism (lipid peroxidation) of the ferroptotic cascade, representing a dual-target strategy for breaking the vicious cycle.

Given the central role of NOX4 in driving astrocytic ferroptosis and neuroinflammation, NOX4 has emerged as a promising therapeutic target. Pharmacological inhibition of NOX4 using GKT137831 or specific siRNA approaches was shown to effectively reduce lipid peroxidation and neuroinflammatory cytokine release while restoring mitochondrial function (36). NOX4 inhibition was also found to alleviate autophagy-lysosome pathway impairment and reduce the degradation of GPX4, thereby restoring the cellular antioxidant defense system (15,87). Notably, neuron-specific knockdown of the NOX4 gene has been shown to alleviate cognitive decline, indicating NOX4 as a key pathogenic target for Tau-related neurodegenerative diseases (87). Therefore, developing NOX4-specific inhibitors with BBB permeability is a current research focus. Although no clinically approved NOX4-selective inhibitors for neurological disorders exist, such drugs hold therapeutic potential for established disease stages based on their key position at the intersection of ferroptosis and neuroinflammation (15,61,82). Additionally, NOX4 inhibition has been proposed as part of glioblastoma therapy to exploit tumor cell sensitivity to iron-dependent death (88).

A1 reactive astrocytes are considered important drivers of the neuroinflammation-ferroptosis vicious cycle due to their secretion of neurotoxic factors. Research shows that A1 astrocytes induce neuronal ferroptosis by inhibiting neuronal SLC7A11 expression via the CXCL10/CXCR3 axis, thereby weakening GPX4-dependent antioxidant capacity (19). Thus, blocking A1 transformation is a key strategy for intervening in this vicious cycle. Although no gene editing therapies directly targeting A1 markers have entered the clinic, basic research provides feasible approaches. For example, in an NMOSD model, inhibiting ACSL4 expression attenuated AQP4-IgG-induced astrocytic ferroptosis and reactive hyperplasia, suggesting that targeting key lipid metabolism enzymes may indirectly inhibit the A1 phenotype (20). Furthermore, single-cell sequencing data have revealed multiple intervenable transcriptional regulatory nodes during A1 transformation, such as sustained activation of the NF-κB pathway. In the future, CRISPR/Cas9 or RNA interference technology can be used to specifically knockdown pro-inflammatory signaling molecules (such as C3, Serping1, and A1 markers) or upstream regulatory factors (such as STAT3), enabling precise blockade of A1 transformation. Combining such strategies with astrocyte-specific promoters can further enhance targeting and reduce off-target effects, providing novel gene therapy approaches for epilepsy, AD, and other diseases (19).

The BBB severely limits the central delivery efficiency of macromolecular or polar drugs such as iron chelators, antioxidants, and NOX4 inhibitors. Accordingly, diverse BBB-targeted delivery systems have been developed to improve the central bioavailability of therapeutic drugs. For example, molecularly imprinted nanoparticles templated with ferrous sulfate adsorb iron ions with high affinity and penetrate the BBB via circulating neutrophils as ‘Trojan horses’, achieving targeted iron clearance in ischemic brain regions, significantly inhibiting ferroptosis, and regulating inflammatory responses (89). Another study constructed a nanocarrier co-assembled with Pluronic F68 and dequalinium, successfully co-delivering DFO and HT to the hCMEC/D3-SH-SY5Y co-culture system, with particle size <170 nm and encapsulation efficiency of 97%, significantly enhancing neuroprotective effects (85). Additionally, peptidomimetic chelators integrating copper/zinc bimetallic chelating structures not only instantly inhibit ROS production induced by amyloid-β peptide-Cu complexes but also initially exhibit favorable BBB penetration potential (90). The common advantages of these delivery platforms are targeted delivery, high drug loading efficiency, and biocompatibility, providing technical support for dual intervention of ferroptosis and neuroinflammation. In the future, combining receptor-mediated transport (such as transferrin receptor targeting) or cell membrane biomimetic strategies may further optimize the spatiotemporal precision of delivery systems and promote the clinical translation of related therapeutic strategies (88,91).

Recent research has revealed that activation of receptor tyrosine kinases, particularly Erb-B2 receptor tyrosine kinase 4 (ErbB4), represents a novel approach to suppressing ferroptosis and alleviating neuroinflammation. Dao et al (13) demonstrated that targeted ErbB4 receptor activation using a small-molecule agonist (E4A) effectively prevents D-galactose-induced neuronal senescence by inhibiting the ferroptosis pathway (13). In this study, E4A administration significantly reversed the downregulation of GPX4 and SLC7A11, reduced iron accumulation and lipid peroxidation, and ameliorated cognitive dysfunction in D-galactose-treated mice. Mechanistically, ErbB4 activation attenuates ferroptosis by upregulating the Nrf2/SLC7A11/GPX4 antioxidant axis and suppressing NF-κB-mediated neuroinflammation, thereby simultaneously addressing both arms of the vicious cycle. Notably, E4A also markedly ameliorated erastin-induced ferroptosis in mouse hippocampal neuronal cells, confirming the direct anti-ferroptotic effect of ErbB4 signaling (13). These findings suggest that receptor tyrosine kinase activation may serve as a complementary therapeutic strategy to iron chelation and antioxidant approaches, offering a receptor-level intervention point that can modulate both ferroptotic and inflammatory pathways simultaneously. The combination of ErbB4 agonists with conventional ferroptosis inhibitors or nano-delivery systems may provide synergistic neuroprotection, although further studies are needed to evaluate their efficacy in astrocyte-specific contexts and their potential for clinical translation.

Despite significant progress in understanding astrocytic ferroptosis and its role in neuroinflammation, achieving precise cell-type-specific intervention remains a major technical obstacle. Existing ferroptosis inhibitors (such as Fer-1) or antioxidant strategies are broadly active, unable to distinguish between glial subpopulations or neuronal subtypes, leading to off-target effects and systemic side effects (92). For example, in an epilepsy model, A1 astrocytes specifically induced neuronal ferroptosis via the CXCL10/CXCR3 axis, but no delivery system currently exists that targets only this astrocytic subpopulation without affecting other neuroprotective glial cells (19). Furthermore, single-cell transcriptomic studies have revealed the high heterogeneity of reactive astrocytes; for example, astrocytes in sepsis models simultaneously express neurotoxic A1 and neuroprotective A2 gene signatures (93), dynamic changes in the Lcn2CreERT2⁺ subpopulation (18), and a specific subpopulation with low GFAP expression but high Aβ clearance capacity (94), further increasing the complexity of targeted intervention. Although gene editing tools (such as CRISPR/Cas9) and RNA interference technology have successfully inhibited ACSL4 or Egr1 to attenuate astrocytic ferroptosis in in vitro models (20), achieving astrocyte-specific delivery in vivo without affecting microglia or other non-target cells remains an urgent bottleneck. Nanocarriers show brain-targeting potential, such as the RFP nanoplatform that co-delivers Pt-TiC nanozymes and Fer-1 to ischemic regions (95), but their selectivity for specific glial subtypes requires optimization.

The vicious cycle formed by neuroinflammation and astrocytic ferroptosis is highly dynamic and region-specific; however, no technical means currently exist for real-time, noninvasive, in vivo monitoring of this interaction. Core ferroptosis markers such as lipid peroxides, GSH depletion, and decreased GPX4 activity mostly rely on post-fixation immunohistochemistry or biochemical detection, failing to reflect instantaneous changes at key nodes during disease progression (20,96). Although single-cell and single-nucleus RNA sequencing technologies have revealed the dynamic polarization and enhanced communication between astrocytes and microglia in sepsis, AD, and other models (93,94), these data are endpoint analyses and cannot capture the continuous process of cycle initiation, amplification, or regression. For example, in the hippocampal CA3 region following high-salt diet induction, the reactivity of disease-associated astocyte-like cells depends on the Lcn2/24p3R axis, but the time window of this transformation and its coupling relationship with local iron fluctuations remain unclear (97). Similarly, after TBI, activation of the cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway has been shown to drive neuroinflammation and promote ferroptosis, but no dynamic imaging tools exist to define when the positive feedback between inflammatory factor release and ferroptosis-related pro-inflammatory mediators (such as ROS and MDA) reaches a critical threshold (98,99). Developing fluorescent probes or positron emission tomography tracers that penetrate the BBB and respond to lipid peroxidation or iron concentration changes will be a key direction to break through spatiotemporal monitoring bottlenecks.

Translating intervention strategies targeting the neuroinflammation-ferroptosis vicious cycle from preclinical models to human applications requires a systematic translational roadmap. First, high-potential targets should be screened based on disease-specific mechanisms. For example, in multiple sclerosis, ACSL4 expression changes were shown to be associated with early ferroptosis in experimental autoimmune encephalomyelitis, and ACSL4 targeting improved behavioral phenotypes and alleviated neuroinflammation (100); in NMOSD, AQP4-IgG-induced astrocytic ferroptosis could be reversed by inhibiting ACSL4 or Egr1(20), indicating ACSL4 as a biomarker and therapeutic target for specific diseases. Second, combined intervention strategies show greater promise, such as iron chelators combined with antioxidants, NOX4 inhibitors combined with anti-inflammatory drugs, or the RFP nanoplatform that simultaneously regulates oxidative stress and ferroptosis (95). Third, a standardized biomarker system should be established to support clinical trial design. Combined detection of ferroptosis-specific markers (such as plasma MDA, and GSH/GSSG ratio) and neuroinflammatory indicators (such as IL-1β and TNF-α) can help identify patient subgroups eligible for enrollment (3,12). Finally, candidate drugs with initially verified safety should be prioritized for advancement, such as Fer-1 liposomes showing favorable biocompatibility in a corneal injury model (92), or rebamipide that simultaneously inhibits neuroinflammation and ferroptosis by activating SIRT1(67), with drug repurposing strategies considered to accelerate clinical translation. Future multicenter, longitudinal cohort studies should be conducted, combining imaging, liquid biopsy, and multi-omics analysis to construct a ‘mechanism-target-efficacy’ closed-loop validation system and promote the implementation of precision neuroprotective therapies.