Reactive oxygen species (ROS) in cancer: from mechanism to therapeutic implications.

Introduction
Reactive oxygen species (ROS) are an array of byproducts produced in cells through aerobic cellular metabolism that play a stimulatory role in multiple crucial signaling pathways in cells via alterations in intra- and extracellular environmental conditions.1 The ROS include nitric oxide (NO•), hydrogen peroxide (H2O2), superoxide anion (O2•−), hydroxyl radical (OH•), and organic peroxides that act as secondary messengers in many signaling mechanisms and are responsible for cell proliferation and differentiation.1 An imbalance between ROS production and antioxidant scavenging disrupts redox homeostasis, contributing to cancer onset and progression by causing genetic mutations, DNA damage, genomic instability, and altered cellular metabolism.2 Elevated ROS levels result in overactivation of key signaling pathways such as extracellular-regulated kinase 1/2 (ERK1/2), mitogen-activated protein kinase (MAPK), and phosphatidyl inositol 3 kinases (PI3Ks), which are crucial for regulating cell survival and propagation.3
On the other hand, an excessive amount of ROS may promote cellular damage, leading to cell death by regulating certain signaling cascades.4 This cell death signaling process involves the activation of different signaling cascades such as apoptosis signal-regulating kinase 1 (ASK1), poly (ADP-ribose) polymerase (PARP), and autophagy-related (ATG) 4 (ATG4) by ROS.5 Thus, the two-edged trait of ROS allows for the exertion of survival or death of cancer cells based on the intracellular levels of ROS.6
Manipulating ROS levels represents a promising therapeutic strategy involving the use of antioxidants, pro-oxidants, targeted ROS signaling inhibitors, combination therapies, and personalized approaches guided by ROS signatures.7 However, challenges remain, including achieving selectivity, determining optimal ROS thresholds, distinguishing ROS levels in healthy versus cancer cells, and bridging preclinical and clinical studies. Therefore, as an emerging research frontier, advancing redox-based cancer therapies requires a deep understanding of how aberrant redox states modulate signaling pathways that can both promote and inhibit tumor growth.
While many studies have focused on the mechanisms of ROS in cancer pathogenesis and ROS-modulating therapies,6,7 this review stands out by comprehensively examining the complex roles of ROS in cancer biology, exploring diverse signaling pathways and how ROS modulate them in cell survival and cell death, and offering detailed insights into therapeutic interventions guided by ROS pathogenesis and addressing challenges in cancer treatment.

Insights into ROS
ROS are a collective term for a group of molecules derived from molecular oxygen, which are formed through redox reactions or electronic excitation that play a role in cellular signaling; however, when produced imbalanced, they can cause oxidative stress that damages cellular structures like lipids, proteins, and DNA.7 They can be classified into nonradical and free-radical species, with the latter having at least one unpaired electron.8 Among nonradical ROS, H2O2 is produced mainly by NADPH oxidases (NOX) and other enzymes and functions as a two-electron oxidant in redox signaling, with selective reactivity toward certain protein cysteines.9 Organic hydroperoxides (ROOHs) result from both enzymatic and nonenzymatic lipid peroxidation, contributing to cell signaling and cell death via ferroptosis.10 Singlet oxygen (1O2) is an electronically excited form of O2 generated through photoexcitation or enzyme reactions, especially in light-exposed tissues.11 O2•− is a prominent ROS that is dismutated into H2O2, participating in various redox reactions, notably influencing lipid peroxidation and nitric oxide interactions.12 OH•, recognized as the most reactive ROS, is generated through Fenton chemistry and plays a critical role in initiating lipid peroxidation.12 Peroxyl radicals (ROO•) are significant in propagating lipid peroxidation chain reactions, whereas alkoxyl radicals (RO•) act as intermediates in the degradation of lipids.13 Each of these radicals contributes to cellular damage and oxidative stress through their highly reactive nature and involvement in complex biochemical processes, but in cancer, the O2•−, H2O2, OH•, and ROO• play major roles.14 O2•− is involved in the early stages of oxidative stress and can generate more reactive species, such as H2O2, which contributes to cancer by inducing genetic mutations and chronic inflammation.12 OH• causes significant damage to DNA, lipids, and proteins, playing a major role in initiating lipid peroxidation and cancer development.6 H2O2 diffuses through membranes, disrupting signaling pathways, and promoting oxidative damage, with elevated H2O2 levels linked to increased cancer risk.6 ROO• propagates lipid peroxidation, leading to cellular damage and contributing to cancer.13

ROS from a historical perspective
Oxygen is highly reactive and can form several types of ROS through both endogenous and exogenous factors. Among these, H2O2 was first discovered by Louis Jacques Thénard in 1818 in redox chemistry, although its role in biology was not fully understood until 1954.15
In 1900, catalase was recognized as a key enzyme in breaking down H2O2, marking an early discovery in antioxidant research.16 Selenium, identified as a toxic catalyst in 1817, was later found to be an essential component of the glutathione peroxidase (GPX) family, with GPX1 becoming widely accepted as a novel peroxidase in 1957.17 Thioredoxin reductase (TRXR), discovered in 1964, and peroxiredoxin (PRDX), discovered in 1968, were also noted as key components of antioxidant systems regulating cell responses to stress.18
That the enzyme NOX was linked to H2O2 production in phagocytes was disclosed in 1964,17 following earlier discoveries of superoxide radicals by Sbarra and Karnowski in 1959.19 Studies by Babior in 1973 confirmed O2•− as the primary product of respiratory bursts, with H2O2 originating from its metabolism.17 McCord and Fridovich’s 1969 discovery of superoxide dismutase (SOD) showed that O2•− could be converted to H2O2.20 The mitochondrial electron transport chain (ETC) was identified as a source of H2O2 in 1967, with Loschen in 1974 confirming that O2•− is the precursor in mitochondria.17
Early studies on ROS highlighted their harmful effects, with key contributions from Gershman and Harman. In 1954, Gershman proposed that excessive oxidants cause tissue injury, linking oxidative stress to cellular damage.21 In 1956, Harman introduced the “free radical theory of aging,” suggesting that accumulated free radicals damage DNA, proteins, and lipids, driving aging and disease.22 In 1971, Loschen first demonstrated that ROS are generated during cellular respiration, a finding supported by Nohl and Hegner in 1978, emphasizing the role of ROS in biological processes.23 In 1977, Mittal and Murad reported that OH• activates guanylate cyclase, which leads to the production of cyclic guanosine monophosphate (GMP), a key signaling molecule.24 Halliwell and Gutteridge further refined the understanding of ROS in 1989, defining them as both free radicals and nonradical oxygen derivatives.25
As scientific understanding evolved in the late 20th century, research revealed that ROS, especially H2O2, are not only destructive agents but also essential regulators of cellular signaling.17 This shift in understanding demonstrated that ROS, particularly H2O2, play dual roles: as signaling molecules at low levels (1–100 nM) regulating growth, differentiation, and apoptosis and as damaging agents at higher levels (>100 nM), causing oxidative damage and cell death.17 The balance between ROS and antioxidant defense relies on adaptive pathways such as the nuclear factor erythroid 2–related factor 2 (Nrf2)/Keap1 and NF-κB pathways. Nrf2, which can induce the expression of phase II detoxifying enzyme-encoding genes through antioxidant response elements (AREs), was defined as a master regulator of antioxidant systems in 1997.17
Methionine was discovered in 1921 by John Howard Mueller,26 but its critical role in oxidative modification emerged decades later. In 1999, Van Patten et al. revealed the involvement of methionine in the oxidative modification of antithrombin by H2O2.27 Earlier, in 1981, Brot et al. identified methionine sulfoxide reductase A (MsrA) as the first enzyme to repair oxidized methionine.28 MsrB, which is specific for methionine-R-sulfoxide (Met-RO), was discovered in 1999 by Hans J. Forman’s team.29 In 2007, free methionine-R-sulfoxide reductase (fMRSR), which is capable of reducing free Met-RO but not protein-bound Met-RO, was identified.30 Finally, in 2013, Drazic and colleagues highlighted methionine oxidation as a novel mechanism for redox regulation of protein function.31
The role of ROS in cell death signaling emerged in the late 1990s. In 1998, David G. Green and colleagues identified ROS as key regulators of programmed cell death.32 In 2007, Scherz-Shouval et al. reported that ROS critically induce autophagy.33 In 2009, Cho et al. demonstrated that mitochondrial ROS are essential for necroptosis execution.34 In 2008, Marcus Conrad’s team reported that redox gene modulation could trigger nonapoptotic death, and Yang and Stockwell reported that this form was iron dependent.35 In 2012, Brent Stockwell’s group formally defined this distinct, iron-regulated, lipid peroxidation-driven cell death as “ferroptosis”.36
Research on redox detection has progressed significantly since the early 20th century, evolving from basic electrochemical methods to advanced computational and machine learning (ML)-driven approaches. In 2020, Barton and colleagues developed a biocompatible nanoscale tool using quantum sensing with nitrogen-vacancy (NV) centers in nanodiamonds (NDs) coupled to nitroxide radicals, enabling highly sensitive detection of paramagnetic species (~10 spins per ND) and real-time monitoring of redox processes such as ascorbic acid oxidation.37 In 2023, Oliveira and colleagues developed a method (MD + CB) to calculate redox potential changes by integrating fluctuation relations with molecular dynamics (MD) simulations, efficiently estimating redox potentials using Bayesian inference.38 Tested on heme proteins, it showed reliable shifts (0.85 correlation) and matched experimental data, aiding in the understating on ROS biology by advancing the prediction and engineering of redox-active proteins and understanding redox signaling in cellular processes and diseases.38 In 2024, Jinnouchi and colleagues developed a method combining first-principles calculations and machine learning to predict redox potentials with high precision.39 The method accurately modeled Fe3+/Fe2+, Cu2+/Cu+, and Ag2+/Ag+ redox potentials, closely aligning with experimental values, advancing ROS biology by enhancing redox reaction modeling and electron transfer studies39 (Fig. 1a).

Source and mechanism of ROS generation
ROS are generated from both endogenous (internal) and exogenous (external) sources. Endogenous sources include cellular processes like mitochondrial respiration and enzyme activity, whereas exogenous sources encompass environmental factors and lifestyle choices. These sources contribute to the overall cellular redox balance, and their dysregulation can lead to oxidative stress and cellular damage.

Endogenous sources of ROS
Endogenous sources of ROS include various cellular organelles and enzymatic reactions within the body:

Mitochondria
Mitochondrial ROS generation primarily occurs at complexes I and III of the ETC due to electron leaks that fail to reach complex IV and instead univalently react with oxygen to form O2•−.40 Superoxide, a membrane-impermeable molecule, is rapidly dismutated into membrane-diffusible H2O2 by SOD.41 H2O2 is further detoxified into water by catalase, glutathione peroxidase, or thioredoxin peroxidase or participates in redox signal transduction. Alternatively, H2O2 can also undergo further reduction via the Fenton reaction to form OH•, a highly toxic molecule that causes oxidative damage.41 Additionally, mitochondrial nitric oxide synthase (NOS) produces nitric oxide (NO•), which then reacts with O2•− to form peroxynitrite (ONOO−), a potent oxidant42 (Fig. 1b).

Peroxisome activity
Peroxisomes, the major sites of intracellular H2O2, represent another important source of ROS. In addition, xanthine oxidase (XO) and xanthine dehydrogenase (XDH), which are involved in purine metabolism, generate O2•⁻ while catalyzing the oxidation of hypoxanthine to xanthine and xanthine to uric acid (UA).43 XO also reduces nitrate and nitrite to NO•, which reacts with O2•⁻ to form ONOO⁻, a reactive nitrogen species (RNS). Additionally, an electron transport chain in the peroxisomal membrane contributes to O2•⁻ production. H2O2 is released during these catalytic activities and can form OH• via Fenton reactions43 (Fig. 1b).

Endoplasmic reticulum (ER)
The ER, an organelle critical for protein synthesis, folding, maturation, and assembly, is another significant source of ROS.44 Oxidative protein folding within the ER is facilitated by oxidoreductases such as protein disulfide isomerases (PDIs), ERp72, and ERp57, with PDI catalyzing thiol-disulfide exchange reactions to form native disulfide bonds in proteins.45 During this process, PDI is oxidized by endoplasmic reticulum oxidoreductin-1 (Ero1), which transfers electrons from reduced PDI to oxygen, producing H2O2. Ero1 also promotes the conversion of reduced glutathione (GSH) to glutathione disulfide (GSSG), contributing to ROS accumulation and ER stress.46 Disruptions in these pathways can lead to protein misfolding and accumulation, triggering ER stress and the unfolded protein response (UPR), thereby inducing cancer. While the UPR initially supports tumor cell survival and propagation, prolonged ER stress can ultimately induce apoptosis in tumor cells44 (Fig. 1b).

NOX
The NOX family of enzymes was the first system identified to produce ROS as a primary function rather than as a metabolic byproduct.47 Evidence strongly correlates NOX-driven ROS generation with leukemogenesis, disease progression, and drug resistance. In 2023, Germon et al. linked FLT3 mutations in AML to increased protein oxidation, phosphorylation, and elevated NOX2 complex activity.48 NOX2 inhibition synergistically enhances apoptosis in FLT3-mutant AML cells and reduces FLT3 phosphorylation and cysteine oxidation, identifying NOX2 as the primary source of ROS.48 This finding aligns with previous in vitro studies demonstrating significant reductions in intracellular ROS levels following NOX inhibition or genetic knockout of NOX isoforms or subunits49 (Fig. 1b).
Besides, thymidine phosphorylase (TP) enhances ROS production by increasing NADPH levels, which activate NADPH oxidase, and through metal-catalyzed oxidation of excess 2-deoxy-D-ribose-1-phosphate (DR1P).50 Polyamines (PAs), such as spermine and spermidine, produce H2O2 through catabolic processes involving enzymes like spermine oxidase (SMOX). Additionally, diamine oxidase (DAO) and acetylpolyamine oxidase (APAO) produce H2O2 during polyamine oxidation.50 The cytochrome P450 (CYP) enzymes in the endoplasmic reticulum and mitochondria, as well as the cyclooxygenases (COXs) and lipoxygenases (LOXs) involved in eicosanoid metabolism, contribute to ROS formation.51 The PI3K/AKT/phosphatase and TENsin homolog deleted on chromosome 10 (PTEN) signaling pathway influences NOX-derived ROS production.52,53 Enzymes such as protein kinase C (PKC), MAPK, and protein kinase A (PKA) modulate NOX activation, whereas the tumor suppressor p53 regulates both antioxidant and prooxidant responses, promoting ROS production and apoptosis in cancer cells.52,53

Exogenous sources of ROS
Exogenous sources of ROS include external factors outside the body or environmental exposures:
External factors such as pollutants, tobacco smoke, radiation, and certain drugs induce ROS through various mechanisms.54 For example, tobacco smoke contains more than 4000 chemicals, such as superoxide and hydroxyl radicals. Ionizing radiation can also generate hydroxyl radicals, either directly by oxidizing water or indirectly through the formation of secondary ROS. Transition metals such as iron, copper, zinc, and aluminum catalyze the Fenton and Haber-Weiss reactions, leading to the formation of highly reactive OH• and hydroxyl anion (OH−) from H2O2.55 Carcinogenic metals like antimony, arsenic, and chromium similarly induce ROS through these reactions. Additionally, pathogenic invasion can trigger immune responses, leading to the activation of neutrophils and macrophages, which generate ROS as part of the host defense mechanism.56 Collectively, various intracellular and extracellular stimuli contribute to ROS formation in cancer cells (Fig. 1b).

Redox homeostasis
Redox homeostasis refers to the balance between the production of ROS and the antioxidant mechanisms that neutralize these ROS.4 While basal ROS levels support signaling and defense, mild oxidative stress damages biomolecules, increasing the risk of mutations and diseases like cancer.57 Conversely, excessive ROS overwhelm the cell’s antioxidant defenses, triggering oxidative damage and, ultimately, programmed cell death.4 To prevent such damage, cells maintain redox homeostasis through an elaborate antioxidant defense system (Fig. 2a, b).
Nrf2 plays a critical role in maintaining redox homeostasis. It is a redox-sensitive transcription factor that regulates the expression of antioxidant and ROS-detoxifying enzymes. Under oxidative stress, ROS modify the Keap1-Nrf2 interaction, allowing Nrf2 to escape degradation, accumulate in the nucleus, and activate AREs.57 This triggers the transcription of genes encoding antioxidants such as SOD, catalase, GSH, glutathione reductase (GR), GSH S-transferase (GSTP), GPX, TRXR, NADPH quinone oxidoreductase 1 (NQO1), and heme oxygenase-1 (HO-1), which neutralize ROS and protect cells against oxidative stress, especially in conditions such as cancer, where ROS levels are elevated57 (Fig. 1b).
When initiated, SOD catalyzes the dismutation of the superoxide anion into H2O2 and oxygen (O2), reducing the risk of oxidative damage from superoxide, a highly reactive ROS. While H2O2 is still a reactive molecule, it is less harmful and is further detoxified by catalase and GPX, thereby protecting cells from oxidative stress.6 The GSH system, where GSH is the primary intracellular antioxidant, neutralizes ROS and restores oxidized protein thiols, converting GSH into GSSG.6 The GSH/GSSG ratio reflects the redox state of the cell, with a high ratio indicating a healthy, reducing environment. Glutathione reductase regenerates GSH from GSSG using NADPH. Similarly, the TRX system maintains redox balance by reducing oxidized proteins and scavenging ROS. TRXR, also NADPH-dependent, keeps TRX in its reduced, active state.58 Both systems, involving enzymes like glutathione peroxidase and peroxiredoxin, are vital for neutralizing peroxides and preventing oxidative stress damage (Fig. 2a).
While Nrf2 signaling-mediated redox balance is beneficial for healthy cells, it also regulates redox homeostasis in cancer cells, ensuring that ROS do not reach toxic levels that could hinder cell proliferation.59

ROS and biology of cancer
ROS, particularly H2O2, can induce reversible oxidation of cysteine residues in several signaling proteins. H2O2 can also directly modulate the activities of several antioxidants, protein kinases, and transcriptional regulators for redox regulation. Besides, ROS trigger different phosphatases that dephosphorylate and thus inactivate kinases, thereby promoting signaling pathways. Therefore, ROS are key modulators of a wide array of pathways and transcription factors associated with cell proliferation, differentiation, immune response, epithelial-mesenchymal transition (EMT), and cell death. These pathways vary in subcellular localization: cytoplasmic-localized pathways include epidermal growth factor receptor (EGFR), extracellular-regulated kinase 1/2 (ERK1/2), ASK1, and ATG4, although some (such as ERK1/2 and EGFR) also exert nuclear effects upon activation.60 In contrast, nuclear-localized factors such as p53 and PARP1, upon ROS-mediated activation, function predominantly in transcription regulation and DNA repair, although neucleocytoplasmic shuttling of p53 is also well documented.60,61 A third redox-sensitive pathway group—including Wnt/β-catenin, PI3K/AKT/mTOR, aryl hydrocarbon receptor (AhR), TGF-β, calcium signaling, Janus kinase (JAK)/signal transducers and activators of transcription (STAT), hypoxia-inducible factor 1 α (HIF1α), NF-κB, Nrf2, c-Jun N-terminal kinase (JNK), p38, and Ataxia-telangiectasia mutated (ATM)—features components that shuttle between the cytoplasm and nucleus, allowing complex regulation of gene expression and cellular fate.60
In terms of cellular progression or defining cell fate, ROS-modulated prosurvival pathways include Wnt/β-catenin, PI3K/AKT/mTOR, AhR, TGF-β, calcium signaling, EGFR, ERK1/2, JAK/STAT, HIF1α, NF-κB, and Nrf2 pathways.3 The prodeath pathways modulated by ROS include JNK, p38, ASK1, p53, ATG4, PARP1, ATM, and AMP-activated protein kinase (AMPK) pathways.3 Prosurvival pathways, when hyperactivated, driven by moderate increase in ROS, are considered to induce pro-oncogenic phenomena, while prodeath pathways, when initiated under excessive ROS, are considered to exert antitumor activity. Interestingly, categorically assigning whether ROS activate or inhibit a specific pathway is overly simplistic. For example, while moderate ROS often drive tumorigenic processes by enhancing prosurvival signaling, high ROS levels can inhibit these same pathways, such as PI3K/AKT, leading to growth arrest or cell death.62,63 Again, it is not uncommon for prosurvival signaling pathways to also induce the expression of cell death-related molecules; for example, JAK/STAT signaling promotes proliferation but can also mediate necroptosis via ZBP1-RIPK3-MLKL activation in the absence of RIPK1.64 Thus, ROS-mediated signaling is highly context dependent, influenced by ROS concentration, cellular metabolic state, and intrinsic genetic factors, underscoring the complex and dynamic interplay between redox balance and cancer biology.

ROS, cell cycle regulation, and carcinogenesis
ROS are key regulators of the cell cycle, with their effects determined by concentration, duration, and cellular context. In the G1 phase, ROS-mediated oxidation/phosphorylation of cyclin-dependent kinase 2 (CDK2) promotes its activation, facilitating the G1/S transition.65 Interestingly, Auranofin-induced H2O2 inhibits cyclin D-CDK4/6 activity by inducing disulfide bond formation between CDK4 and cyclin D, reducing cell proliferation.66 During the S phase, ROS modulate CDK2-cyclin A activity and intra-S phase checkpoints, with mitochondrial ROS accelerating replication onset.67 A moderate increase in ROS levels may increase CDK2 activity, which is vital for bypassing senescence and enabling cancer cell immortalization.67 Conversely, in the G2/M transition (particularly redox-sensitive), sustained ROS have also been reported to oxidize and inactivate Cdc25 phosphatases, preventing CDK1 activation and blocking mitotic entry, potentially leading to cell death.68 Moreover, excessive ROS hyperphosphorylate Aurora A, disrupting spindle assembly,69 and oxidize APC/C components like APC11, inhibiting cyclin B degradation and prolonging mitotic arrest.70
Additionally, physiological levels of ROS, particularly H2O2, can modulate numerous prosurvival pathways and transcription factors involved in cell cycle progression. For example, Wnt/β-catenin signaling promotes β-catenin nuclear translocation, where it interacts with T-cell factor/lymphoid enhancer-binding factor (TCF/LEF) transcription factors to activate the expression of target genes regulating cell proliferation, such as c-Myc and cyclin D1.71 AhR translocates to the nucleus, forms a complex with ARNT, and binds to xenobiotic response elements (XREs) or dioxin response elements (DREs) located in the promoter regions of target genes involved in cell cycle progression and proliferation.72 However, the ability of the AhR pathway to induce a G1/S cell cycle block, as evidenced following human cytomegalovirus (HCMV) infection, highlights the need for further research considering its role in various cancers and cellular contexts.72 TGF-β enhances cyclin D1 transcription via β-catenin signaling, enhancing cyclin D1 promoter activity.73 Ca²⁺ signaling activates transcription factors such as the DREAM complex and heat shock transcription factors (HSFs), driving immediate early gene (IEG) expression.74 These IEGs are responsible for transitioning cells from a resting state (G0) back into the cell cycle, initiating proliferation. HIF1α promotes cell growth by enhancing glycolysis under hypoxia and regulating growth factors like IGF-2 to support proliferation and survival.75 Once activated, EGFR triggers multiple downstream pathways, including the Ras-Raf-MAPK pathway and the PI3K-AKT pathway, ultimately leading to cell division and growth.76 JAKs phosphorylate receptors and STAT proteins, leading to STAT dimerization and nuclear translocation to drive transcription of genes for growth, proliferation, and survival.77 NF-κB promotes the G1-S transition by upregulating cyclin D1 and inhibiting p53, driving cell growth and apoptosis resistance.67
The PI3K/AKT pathway is a key signaling cascade linked to cell proliferation and is often dysregulated in cancer. Regarding PI3K/AKT pathway, activated PI3K phosphorylates PIP2 into PIP3, which recruits AKT to the plasma membrane, where it is phosphorylated at Thr308 and Ser473, triggering full activation.78 AKT activation regulates cell survival, growth, and metabolism by inhibiting p27, activating cAMP-response element binding protein (CREB), inactivating tuberous sclerosis complex 2 (TSC2), sequestering FOXO in the cytoplasm, and engaging a feedback loop with mTORC2 (AKT and mTORC2 form a critical feedback loop in which mTORC2 activates AKT by phosphorylating it at serine (Ser)473. This increased AKT activity might subsequently promote further mTORC2 stimulation).79 These coordinated activities collectively promote protein synthesis and cell cycle progression. The ERK pathway activates kinases such as p90RSK and MSK1, which phosphorylate CREB at Ser133.80 When CREB is phosphorylated, it acts as a transcriptional activator of various genes involved in cell growth and proliferation, including c-Myc and CDK1. As described earlier, hyperactivation of these pathways due to increases in ROS levels may contribute to carcinogenesis.
Conversely, excessive ROS concentrations can exert an antitumor effect by either modulating prodeath signaling pathways or inhibiting prosurvival pathways. For example, supraphysiological levels of ROS induces G1 arrest by stabilizing c-Fos binding to chromatin, suppressing Fra-1-mediated cyclin D1 expression, and activating p21 through the ATM/p53 signaling pathway.81 During S-phase, oxidative DNA damage, including 8-oxoG lesions and strand breaks, activates the MRN-ATM-Chk2 cascade, halting replication to allow repair via p53/p21 activation.82 Besides, drug-induced ROS can inhibit the PI3K/AKT pathway to suppress cancer cell proliferation by disrupting key pathways critical for tumor growth and survival, as evidenced by studies with (+)-anthrabenzoxocinone in NSCLC and 6-methoxydihydrosanguinarine in breast cancer.62,63 While the exact basis for this paradoxical role remains unclear, excessive ROS levels induced by these drugs may underlie their inhibitory effects (Fig. 2c).

ROS and angiogenesis
Elevated ROS levels (particularly NOX4-induced H2O2) promote angiogenesis through redox-mediated pathways, with HIF1α playing a central role by directly upregulating VEGF and angiogenic factors like angiopoietin-2 and endothelin-1 (EDN1), particularly in colon cancer.83 AKT-mediated FOXO inhibition prevents VEGF suppression while enhancing Sp1-driven VEGF transcription.78 Wnt/β-catenin signaling similarly activates VEGF and EDN1 expression via TCF/LEF complexes, as evidenced in HEK293T cells.84 In addition to activating VEGF, NF-κB initiates interleukin-8 (IL-8) and matrix metalloproteinase 9 (MMP9), and interacts with the PI3K pathway to enhance the expression of angiogenic factor with G patch and FHA domains 1 (AGGF1).85 TGF-β1 induces angiogenesis primarily by enhancing VEGF secretion under hypoxia and upregulating signaling proteins such as Smad3/4 and HIF1α, which further increase VEGF transcriptional activity through histone acetyltransferase (p300), emphasizing the role of chromatin remodeling.86 TGF-β1 also upregulates Flt-1, a key VEGF receptor, further amplifying the angiogenic response.
EGFR signaling promotes the Ras-Raf-MAPK pathway or recruits tumor-infiltrating neutrophils via IL-8, which enhances the permeability of newly formed vessels, facilitating tumor cell migration.76 Neutrophils also deliver a specialized form of MMP9 that enzymatically activates VEGF. Under hypoxia, STAT3 forms transcriptional complexes with HIF1α, CBP/p300, and Ref-1/APE to enhance VEGF expression, as observed in multiple cancers.87 STAT3 also directly binds to the VEGF promoter, as observed in NSCLC, and regulates MMP2 and MMP9 secretion, facilitating ECM degradation and angiogenesis.88 Additionally, cofactors like HIF1α and Sp1 collaborate with STAT3 in endothelial cells to further promote VEGF expression, supporting angiogenesis87 (Fig. 3a).
Conversely, excessive ROS can inhibit VEGF signaling by damaging receptors and depleting NO• through ONOO− formation, disrupting vasodilation and endothelial function.89 ROS may also activate antiangiogenic pathways, such as p53-mediated production of thrombospondin-1.90 Furthermore, elevated ROS levels can trigger endothelial apoptosis via mitochondrial dysfunction and caspase activation.91 The ability of excessive ROS to inhibit angiogenesis is now being leveraged in the development of ROS-mediated cancer therapeutics.

ROS and metastasis
Moderate ROS levels promote metastasis and cancer cell invasion by hyperactivating prosurvival pathways like Wnt/β-atenin pathway, which drives EMT through β-catenin’s nuclear interaction with transcription factors such as ZEB1/2 and SNAIL to repress epithelial markers.92 PI3K/AKT activation triggers mTORC2, a key regulator of EMT,79 whereas TGF-β signaling disrupts cell polarity via Par6 phosphorylation by TGF-β type II receptor (TβRII) and RhoA degradation, which is mediated by Smurf1, and induces miR-155 expression to destabilize tight junctions.93 Interestingly, TGF-β signaling activates RhoA, promoting stress fiber formation to enhance migration and invasion, while its activation of mTORC2 further supports RhoA activity, facilitating EMT and cell motility.94 Additionally, TGF-β signaling activates SNAIL, SLUG, and ZEB, thus promoting a more migratory mesenchymal phenotype.94 Calcium signaling through calcineurin activates the nuclear factor of activated T cell (NFAT), which translocates into the nucleus to induce ZEB1/2 expression, modulating E-cadherin downregulation and N-cadherin and vimentin upregulation.95 EGFR activation triggers the MAPK and PI3K-AKT pathways, which directly regulate the expression of mesenchymal and epithelial markers.76 HIF1α drives EMT via TWIST and ZEB1,96 whereas activated STATs promote MMP expression and the expression of EMT-associated transcription factors such as SNAIL, TWIST, and ZEB1.88 A recent study reported that chronic oxidative stress disrupts the circadian rhythm of neutrophils and promotes neutrophil extracellular trap (NET) formation via glucocorticoid release, creating a metastasis-supporting microenvironment.97
Conversely, excessive ROS have been well documented to inhibit metastasis in several studies. GPX2-knockdown-induced ROS suppress gastric cancer progression and metastasis by disrupting the KYNU-Kyn-AhR signaling pathway,98 which is known to promote metastasis, as evidenced in chronic lymphocytic leukemia (CLL) in mice.99 Similarly, a cholesterol oxidase-loaded Co–PN3 single-atom nanozyme effectively suppresses tumor metastasis by enhancing ROS generation100 (Fig. 3b).

ROS and drug resistance
Moderate levels of ROS are linked to the upregulation of drug efflux pumps like P-gp and MRP1. ROS regulate several signaling pathways involved in regulating drug efflux pumps. For example, β-catenin, after nuclear translocation, upregulates the expression of drug resistance genes such as ABCB1 (encoding P-gp) and ABCC1 (encoding MRP1).101 The PI3K/AKT pathway promotes multidrug resistance by upregulating ABCB1, ABCC1, and ABCG2 (encoding BCRP), and activating antiapoptotic proteins like X-linked inhibitor of apoptosis (XIAP).102 HIF1α activates H19 transcription in NSCLC cells, with long noncoding RNA H19 overexpression linked to increased P-gp and MRP1 expression.103 Furthermore, the EGFR and ERK pathways induce p-gp expression by activating PI3K signaling, while TGF-β1 enhances P-gp through the HOTAIR/miR-145 axis, providing a novel mechanism for drug resistance in cancer.104,105 A study demonstrated that oxidative stress-induced Nrf2 promotes drug efflux by upregulating ABCG2, as observed in biliary tract cancer cells.106 Interestingly, targeting the Nrf2 pathway to overcome drug resistance in liver cancer involves the induction of excessive ROS.107 Another recent study suggested that the 2D-CuPd nanozyme overcomes Tamoxifen resistance in breast cancer by leveraging the ROS-mediated inhibition of the PI3K/AKT/mTOR pathway.108 Many other studies are also exploring the vulnerability of cancer cells to excessive ROS to overcome drug resistance. (Fig. 4a).

ROS in immunosuppression
In the tumor microenvironment (TME), excessive ROS exert profound immunosuppressive effects, undermining antitumor immunity and promoting cancer progression. Cancer cells, cancer-associated fibroblasts (CAFs), and myeloid-derived suppressor cells (MDSCs) are major contributors to elevated ROS levels, which suppress T-cell responses and promote the apoptosis of T cells due to their weak antioxidant systems.109 ROS impair the function of cytotoxic CD8+ tumor-infiltrating lymphocytes (TILs) by inducing mitochondrial dysfunction, which reduces their ability to attack cancer cells.110 Similarly, natural killer (NK) cells are rendered less effective, as ROS increase their apoptosis and functional impairment.111 Furthermore, ROS contribute to the upregulation of immune checkpoint molecules like PD-L1 on cancer cells via FGFR1 signaling, suppressing T-cell activity and enabling immune evasion.112 ROS also favor the survival of immunosuppressive cells such as MDSCs and M2 macrophages, which have enhanced antioxidant capacities, allowing them to thrive under oxidative conditions.113 Through nitration and nitrosylation, ROS impair antigen presentation and TCR functionality, exacerbating immune dysfunction.114 Hydrogen sulfide, an RSS, promotes regulatory T (Treg) cell differentiation by activating FOXP3.115 ROS enhance the TGF-β/Smad2/3 pathway and induce NF-κB activation, collectively suppressing antitumor immunity by inhibiting cytotoxic T cells and NK cells, promoting the expansion of Tregs and MDSCs, and upregulating immune checkpoints like PD-1 and PD-L1.116 Additionally, ROS stabilize HIF1α, promoting immunosuppression in NSCLC through inducing EMT, reducing CD8+ TILs, and activating the HIF1α/LOXL2 pathway, which fosters a hypoxia-driven immunosuppressive microenvironment117 (Fig. 4b).
Conversely, the immunostimulatory role of moderate levels of ROS is also well evidenced. ROS, including O2•⁻, H2O2, and OH•, serve as critical signaling molecules that bolster immune responses through their role in both innate and adaptive immunity. Upon T-cell receptor (TCR) activation, mitochondrial oxidative phosphorylation (OXPHOS) and NOXs are upregulated, leading to increased ROS production, which fuels T-cell proliferation and interleukin (IL) secretion.118 Moderate ROS levels also activate NFAT to stimulate IL-2 secretion, promoting T-cell proliferation and survival.119 Besides, ROS can also induce the type 2 cytokine Fc–IL-4, which revitalizes exhausted CD8+ T cells by increasing LDHA-driven glycolysis and NAD+ generation, effectively restoring their antitumor activity.120 ROS play crucial modulatory roles in amplifying distal TCR pathways, including Ca2+-calcineurin-NFAT, IKK-NF-κB, Ras-ERK1/2, and mTOR cascades, which in turn enhance T-cell activation, proliferation, and cytokine production (e.g., IL-2).121 Additionally, ROS-activated AhR in effector T cells supports their transition into memory T cells, ensuring long-term immune surveillance.122 Peroxynitrite stimulates TLR4 and NF-κB activation and cytokines (release of TNF-α, IL-1β, and IL-8)123, and NOX-induced ROS play a role in inflammasome activation, such as the NLRP3 inflammasome, which drives caspase activity and cytokine production to enhance immune cell recruitment and function.124 Additionally, the loss of transmembrane p24 trafficking protein 4 (TMED4)-induced ROS reduces FOXP3 stability and the suppressive function of Tregs in an IRE1α/XBP1 axis–dependent manner125 (Fig. 4b).

ROS, cellular metabolism and cancer

Carbohydrate metabolism
ROS modulate cancer metabolism by influencing multiple pathways critical for tumor survival and proliferation. In glycolysis, ROS oxidize GAPDH and PKM2, altering their activity, thereby inhibiting glucose uptake, while simultaneously enhancing glucose uptake and metabolism through HIF1α stabilization.126 Interestingly, inhibition of GADPH and PKM2 redirects glucose-6-phosphate (G6P) flux away from glycolysis and towards the oxidative arm of the PPP, which produces NADPH, thereby boosting NADPH production to counter oxidative stress.127 In mitochondria, ROS impact the TCA cycle by oxidizing enzymes such as aconitase and α-ketoglutarate dehydrogenase (αKGDH), modulating metabolic flux, whereas superoxide generated at the ETC amplifies redox signaling.128 Glutaminolysis supports ROS detoxification by enhancing glutathione regeneration. ROS directly oxidize the glutaminase protein, altering its structure and function, and modulating metabolic pathways, including elevated mitochondrial ROS production, which further affects glutaminase activity. Interestingly, ROS can increase glutaminase activity by activating p53, which increases GLS2 expression and activity.129 Additionally, serine‒glycine one-carbon metabolism (SGOC) provides NADPH and nucleotides, with Nrf2 upregulating this pathway to mitigate oxidative damage.130 Cancer cells balance ROS levels to promote redox signaling and survival while avoiding oxidative stress, a vulnerability that therapies targeting these pathways aim to exploit, selectively inducing oxidative damage and cell death in tumors (Fig. 5a).

Lipid metabolism
Elevated ROS activate AMPK, which phosphorylates and inhibits acetyl-CoA carboxylase (ACC), reducing malonyl-CoA levels and thereby relieving the inhibition of carnitine palmitoyltransferase 1 (CPT1), the rate-limiting enzyme in fatty acid oxidation (FAO).131 This metabolic shift promotes FAO, leading to increased ATP production and regeneration of NADPH via TCA cycle-linked pathways. In many tumors, enhanced FAO provides metabolic flexibility, sustaining ATP and NADPH levels during oxidative and metabolic stress, which promotes survival, therapy resistance, and metastatic potential.132 Concurrently, ROS, particularly OH•, initiate lipid peroxidation by attacking polyunsaturated fatty acids (PUFAs) within cellular membranes. This peroxidation process leads to the formation of 4-hydroxy-2-nonenal (4-HNE).133 4-HNE readily forms covalent adducts with nucleophilic amino acid residues—particularly cysteine, histidine, and lysine—through Michael addition or Schiff base formation, frequently resulting in protein misfolding, unfolding, or aggregation. Notably, in glioma cells, fatty acid-binding protein 7 (FABP7) binds and internalizes PUFAs to promote lipid droplet formation, whereas deletion of FABP7 decreases lipid droplets accumulation and leads to elevated ROS levels6 (Fig. 5a).

Nucleic acid metabolism
ROS significantly influence nucleic acid metabolism through direct oxidative modifications and by regulating transcriptional, epigenetic, and posttranscriptional mechanisms.

Redox regulation of transcription
ROS modulate the activity of redox-sensitive transcription factors (TFs), such as Nrf2, NF-κB, HIF1α, and FOXO, either through posttranslational modifications (PTMs) or by impacting upstream signaling pathways (such as Keap1 in Nrf2 signaling, IKK in NF-κB signaling, and PHD2 in HIF1α signaling).134 Under ROS stress, Nrf2 binds to ARE, which then induces noncoding RNA (ncRNA). While ncRNAs play a major role in therapeutic resistance, they can function as regulated cell death (RCD) accelerators that modulate therapeutic sensitivity. Besides, translation of ncRNAs can produce unwanted peptides in ROS-associated diseases, including cancer. Aberrant polyadenylation events within introns of tumor suppressor genes such as DICER1 and FOXN3 result in the production of truncated proteins lacking essential tumor-suppressive functions, thereby promoting tumorigenesis.135 Additionally, certain long ncRNAs (lncRNAs) can be translated into peptides bearing hydrophobic C-terminal tails, enabling their localization to cellular membranes, where they may act as tumor-associated antigens.136 Interestingly, normal cells and cancer cells seem to have different lncRNA transcripts in response to H2O2 stress, which might contribute to the different sensitivities to ROS observed in normal and cancer cells. ROS-induced PTMs also activate NF-κB signaling, leading to the upregulation of the antiapoptotic proteins BCL-XL and IAP137 (Fig. 5b).

Redox regulation of epigenetics
Redox signaling exerts long-term cellular effects through epigenetic mechanisms, including DNA and histone methylation, acetylation, and other PTMs. Mito-ROS regulate the expression of DNA methyltransferases (DNMTs), leading to changes in the expression of oxidative stress-related genes.138 Furthermore, the methyl donor of epigenetic methylation is from methionine metabolism, which can also be regulated by H2O2. Acetylation of histones at lysine residues serves as a transcriptional activator. This reaction is catalyzed by histone acetylases (HATs) using acetyl-CoA as a donor substrate. Histone deacetylases (HDACs), on the other hand, are regulated by redox modifications—for instance, S-nitrosylation of HDAC2 at cysteine residues induces its release from chromatin, which promotes chromatin remodeling and transcriptional activation of nearby genes.9 Furthermore, S-glutathionylation of histone 3 affects nucleosome stability and alters chromatin structure, which enhances the binding of the replication machinery to DNA.139 Sirtuins are an additional class of NAD+-dependent HDACs whose activity is indirectly redox-regulated through their dependence on the NAD+/NADH ratio maintained by the TCA cycle and affected by mitochondrial ROS production.140 Besides, 8-oxo-2′-deoxyguanosine, a product of DNA oxidation, functions as an epigenetic modifier that regulates gene expression. Additionally, RNA modification is a posttranscriptional event that can be regulated by ROS.6 The generation of N6-methyladenosine is a very common type of RNA modification affected by H2O2 through the activation of methyltransferases or demethylases.141 Notably, N6-methyladenosine modification can alter the stability and translational efficiency of mRNAs, leading to drug resistance (Fig. 5b).

Protein metabolism
Redox signaling intricately regulates protein metabolism at multiple stages, including protein synthesis, folding, and degradation. ROS, particularly H2O2, negatively influence protein synthesis by inducing phosphorylation of eukaryotic initiation factor (eIF2) via oxidative stress-sensitive kinases such as PERK and GCN2, thereby repressing translation.142 Additionally, H2O2 suppresses mTORC1 activity, reducing phosphorylation of 4EBP1, which then binds eIF4E, preventing interaction with eIF4G, thereby blocking translation initiation. In the ER, protein folding is redox-regulated through the formation of structural disulfide bonds, which are catalyzed by Ero1 and PDI, both of which utilize redox-active cysteine residues.143 H2O2 also modulates protein degradation by promoting the formation of the immunoproteasome through IFN-γ signaling, which involves dissociation of the 19S regulatory particle from the 20S core, assisted by Hsp70.144 Moreover, H2O2 enhances proteasome activity via S-glutathionylation of cysteine residues in the 20S subunit.145 ROS also induce protein carbonylation by oxidizing amino acid residues such as proline, arginine, lysine, and threonine, forming reactive carbonyl groups—markers of irreversible oxidative damage.146 Accumulation of carbonylated proteins disrupts proteostasis and contributes to tumor progression; however, excessive ROS-induced ER stress can also exert antitumor effects by overwhelming cancer cell survival mechanisms (Fig. 5c).

ROS and cell death
The specific type of RCD that cancer and immune cells undertake in response to ROS stress is determined by the intrinsic genetic wiring, metabolic state, and level and type of ROS. Intrinsic apoptosis occurs in stressed epithelial or cancer cells via mitochondrial BAX/BAK activation and caspase-9-mediated pathways when antioxidant defenses fail but the apoptotic machinery remains intact.147 Superoxide and hydroxyl radicals drive intrinsic apoptosis by disrupting mitochondrial membrane potential, whereas H2O2 promotes extrinsic apoptosis by inducing c-FLIP ubiquitination and proteasomal degradation, enabling caspase-8 activation. Extrinsic apoptosis dominates as ROS synergize with death receptor signaling (e.g., TNFR/Fas), activating caspase-8 through DISC formation.148
Necroptosis and extrinsic apoptosis share initial signaling pathways involving death receptors such as TNFR1 or Fas, but necroptosis occurs in caspase-8-deficient contexts (e.g., inflamed tissues or resistant cancers) via RIPK1/RIPK3/MLKL activation.149 Under specific conditions, ROS can bypass these pathways to induce pyroptosis, driven by inflammasome activation (e.g., NLRP3) in response to PAMPs/DAMPs, the predominance of inflammatory caspases (caspase-1/4/5/11) over caspase-8 or RIPK3, and cell-specific mechanisms prioritizing cytokine release (IL-1β/IL-18) in immune cells like macrophages.150 In such cases, ROS-activated inflammasomes cleave gasdermin-D to form membrane pores, enabling pyroptosis while suppressing apoptosis and necroptosis through mechanisms like c-FLIP degradation or RIPK1 ubiquitination.
Ferroptosis, characterized by GPX4 inhibition and iron-dependent lipid peroxidation, occurs in iron-rich cells such as glioblastomas and neurons, and is often triggered by decreased SLC7A11 expression or cystine depletion.151 In contrast, disulfidptosis arises in cancer cells under glucose deprived conditions with high SLC7A11 expression, leading to cystine accumulation, disulfide stress, and cytoskeletal collapse, ultimately resulting in cell death.152
Oxeiptosis occurs when extreme oxidative damage causes lysosomal membrane permeabilization, particularly when other death pathways are impaired.153 In contrast, cuproptosis, which is regulated by cuproptosis-related genes (CRGs), is promoted by genes such as FDX1, LIAS, and PDHA1, which increase cellular sensitivity, whereas genes like MTF1 and CDKN2A reduce sensitivity.154 Cells with mutations affecting Fe-S cluster formation or stability are especially vulnerable to cuproptosis because of weakened mitochondrial function and increased susceptibility to copper-induced toxicity.155 And ROS-induced NETosis occurs when tumor-derived factors like cytokines and DAMPs activate neutrophils via NOX2, resulting in excessive ROS production.156

Apoptosis
Excessive intracellular ROS activate proapoptotic BH3-only proteins (e.g., BAD, BIM, PUMA, NOXA), which displace antiapoptotic BCL-2 family members (e.g., BCL-2, BCL-XL), enabling BAX/BAK activation and mitochondrial outer membrane permeabilization (MOMP).157 This releases cytochrome c and apoptogenic factors such as AIF, Smac/Diablo, and endonuclease G, initiating apoptosome formation and caspase (-9, -3, -6, and -7) activation.158 Effector caspases then cleave cellular proteins, causing DNA fragmentation, cytoskeletal dismantling, and controlled cell death.
In addition, ROS modulate several signaling pathways to promote apoptosis. These pathways include the ASK1 pathway, which initiates proapoptotic signaling through JNK and p38 pathways, leading to the phosphorylation of proapoptotic proteins like BAD, BIM, and BAX.159 PARP1 also activates BAX and BAK, contributing to apoptosis via mitochondrial dysfunction and chromatin remodeling.160 Additionally, PARP1 drives apoptosis via interactions with signaling molecules such as apoptosis-inducing factor (AIF), p53, caspase-3, caspase-7, and ATM.161 ATG4 plays a significant role in apoptosis, as ATG4 may shift cells from survival to apoptosis when autophagy is inhibited.162 Additionally, p53, through its activation by AMPK, ATM and other stress-related pathways, promotes BAX, PUMA, and NOXA,163 whereas TGF-β enhances the expression of BIK while downregulating BCL-XL, causing an increased apoptotic response.164 HIF1α binds to the hypoxia-responsive element (HRE) of the BCL-2 interacting protein 3 (BNIP3) gene, upregulating the expression of this apoptosis-inducing factor in various cancers.165 ROS also interact with calcium signaling at the mitochondrial–ER interface, where Ca2+ surges induce mitochondrial swelling and rupture via disrupted ion exchange and adenine nucleotide translocase (ANT) channel dysfunction.166 Besides, Ca2+ overload activates calpain, which cleaves BCL-2 proteins and activates BID and AIF, thereby promoting apoptosis. Ca2+-activated calcineurin dephosphorylates BAD, enhancing its proapoptotic activity.166
ROS promote extrinsic apoptosis by accelerating the ubiquitin-mediated degradation of c-FLIP, thereby enhancing DISC formation and caspase-8 activation through adapter proteins such as Fas-associated death domain (FADD) and TNF receptor-associated death domain (TRADD). ROS also activate pathways such as the AhR, p53, JNK, and ASK1 pathways, enabling extrinsic apoptosis. Recent findings revealed that ROS-driven AhR/CYP1B1 signaling induces Fas via p53 activation, contributing to 6PPDQ-induced cardiac dysfunction in zebrafish embryos.167 Additionally, p53 can bypass classical death receptors by upregulating Reprimo (RPRM), which activates the Hippo–YAP/TAZ–p73 axis, leading to the expression of proapoptotic genes.168 JNK and ASK1 modulate death receptor signaling, amplifying extrinsic apoptosis.169
Moderate levels of ROS, on the other hand, inhibit apoptosis by activating prosurvival pathways such as the PI3K/AKT, ERK, and NF-κB pathways. The PI3K/AKT pathway inactivates BAD and BAX while also suppressing caspase-9 and caspase-3 activation.170 Similarly, ERK signaling phosphorylates and neutralizes BAD, BIM, and BMF (Ser77), preventing their interaction with BCL-2 family proteins and enhancing the expression of BCL-2 and BCL-XL.171 Additionally, NF-κB upregulates antiapoptotic genes, including FLIP, BCL-XL, c-IAP, XIAP, TRAF1, and TRAF2, further blocking cell death172 (Fig. 6a).

Autophagy
Autophagy is a tightly regulated cellular degradation process in which lysosomes eliminate protein aggregates, damaged organelles, and invading pathogens.173 ATG4 uniquely regulates autophagy by processing and deconjugating LC3/ATG8-PE. ROS-induced ATG4 oxidation enhances LC3 lipidation and early autophagosome formation,174 although phagophore growth can occur independently of LC3/ATG8.162 ROS also regulate autophagy indirectly through the activation of MAPKs, including JNK and p38.175 JNK phosphorylates c-Jun, enhancing transcription of Beclin 1, crucial for autophagosome formation. and regulates LC3, facilitating autophagy.176 p38β MAPK induces autophagy by phosphorylating ULK1 at Ser555, particularly in response to cancer-related stimuli.177 The p53 pathway induces autophagy by upregulating Beclin 1 and damage-regulated autophagy modulator (DRAM).178 ROS-Ca2+ signaling also facilitates autophagy via calcium-permeable ion channels such as those of the transient receptor potential (TRP) family.179 TRPV1 mediates prosurvival autophagy in thymocytes in response to capsaicin,180 whereas TRPC1-driven Ca2+ entry triggers autophagy under hypoxia and nutrient deprivation.181 Additionally, TRPM2 induces Beclin 1 phosphorylation at Ser90 via CaMKII, increasing hepatocyte vulnerability to cell death during oxidative stress.181 Besides, In response to oxidative DNA damage, PARP1 activates AMPK, which in turn suppresses mTOR signaling—a key inhibitor of autophagy.5 This pathway induces a prosurvival form of autophagy in NPC cells (Fig. 6b).
Interestingly, autophagy might modulate ROS to promote cancer cell survival and drug resistance, depending on the context. In pancreatic cancer, autophagy mitigates Tetrandrine-induced ROS, preserving mitochondria and reducing apoptosis, thereby contributing to drug resistance.182 HIF1α-BNIP3-mediated mitophagy reduces mitochondrial ROS and NLRP3 inflammasome activation, which is protective in conditions like renal fibrosis but potentially promotes tumor progression.183 Therefore, context-based autophagy targeting could be a potential antineoplastic therapeutic strategy.

Necroptosis
ROS amplify RIPK1 activation via a feedback loop. TNF-α-induced RIPK3 activation generates ROS, which promote RIPK1 autophosphorylation at Ser161 and facilitate the formation of disulfide bond-linked aggregates through cysteines 257, 268, and 586.184 This enhances RIPK1 recruitment of RIPK3, further increasing ROS production and driving necrosome formation. Additionally, ROS oxidize RIPK1 and reduce the levels and activation of executioner caspases-3, -6, and -7, priming the hyperglycemic shift from apoptosis to necroptosis before TNF-α engages death receptors.185 Besides, ROS-driven JAK/STAT signaling upregulates Z-DNA binding protein 1 (ZBP1), which forms a complex with RIPK3 in the absence of RIPK1, triggering MLKL-mediated necroptosis.64 Notably, RIPK3 links necroptotic signaling to cellular metabolism by interacting with glutamate dehydrogenase 1 (GLUD1), which catalyzes the conversion of glutamate to alpha-ketoglutarate.186 Interestingly, insulin-induced ROS might induce necrosis by initiating PI3K/AKT, a prosurvival pathway,187 suggesting context-specific ROS modulation of necroptosis (Fig. 6b).

Ferroptosis
This process begins with the inhibition of the cysteine/glutamate antiporter system xCT, which is composed of solute carrier family 7 member 11 (SLC7A11) and solute carrier family 3 member 2 (SLC3A2) and limits the extracellular cysteine uptake required for GSH synthesis.188 Depletion of GSH impairs GPX4, halting the detoxification of lipid hydroperoxides and triggering unchecked lipid peroxidation189. Lipid peroxides react with ferrous iron (Fe2+) through Fenton reactions, generating ROS that damage cellular membranes.190 Dysregulated iron metabolism, characterized by increased transferrin receptor 1 (TFR1) expression and decreased ferritin levels, leads to iron overload and further amplifies ROS production.190 Antioxidant systems, including ferroptosis suppressor protein 1 (FSP1)-ubiquinol (CoQH2)-vitamin K hydroquinone (VKH2), dihydroorotate dehydrogenase (DHODH)-CoQH2, and GTP cyclohydrolase 1 (GCH1)-tetrahydrobiopterin (BH4), provide protection by mitigating oxidative damage and inhibiting ferroptosis.188
Under excessive ROS, p53 plays a crucial role in promoting ferroptosis by negatively regulating SLC7A11, leading to decreased GSH and increased sensitivity to ferroptosis by limiting the antioxidant capacity of cells.191 Additionally, p53 influences iron homeostasis through enzymes like ferredoxin reductase (FDXR), enhancing iron accumulation, which is pivotal for ferroptosis.192 Paradoxically, ROS can modulate ferroptosis by regulating SLC7A11 expression through the Nrf2-mediated pathway.193 Interestingly, a recent study identified a GPX4- and FSP1-independent regulation of ferroptosis involving the phospholipid-modifying enzymes MBOAT1 and MBOAT2, which are regulated by sex hormones.194 Ferroptosis may induce immunosuppression, as ferroptosis in MDSCs results in the release of oxygenated lipids that impair T-cell function (Fig. 6b).

Disulfidptosis
Disulfidptosis is a form of PCD induced by glucose starvation in SLC7A11-overexpressing cells, where reduced NADPH production impairs cysteine-to-cystine conversion, causing cysteine accumulation and disulfide stress.195 Excessive OH• depletes NADPH, leading to abnormal disulfide bond formation in actin cytoskeletal proteins, resulting in F-actin aggregation, cytoskeletal contraction, detachment from the plasma membrane, and structural disruption, ultimately culminating in cell death.196 Besides, ROS might induce disulfidptosis by regulating the JNK and NF-κB pathways.197,198 Recent studies have shown that cells with high disulfidptosis scores are more vulnerable to glucose deprivation199. This metabolic vulnerability can enhance sensitivity to certain anticancer drugs, although in some contexts, disulfidptosis induction might paradoxically contribute to drug resistance in cultured cells through adaptive stress responses199 (Fig. 7a).

Pyroptosis
ROS play dual roles in pyroptosis, which is mediated by inflammasomes and gasdermin proteins (GSDMs).200,201 Under ROS stress, thioredoxin-interacting protein (TXNIP) dissociates from TRX and activates NLR family pyrin domain containing 3 (NLRP3) inflammasomes, leading to caspase-1 activation and GSDMD cleavage, forming membrane pores and releasing proinflammatory cytokines.202,203 Mito-ROS oxidize GSDMD cysteine residues, enhancing its cleavage by caspase-1/caspase-11.204 Interestingly, NLRP3 and GSDM pore formation in the mitochondrial membrane can induce ROS production by causing mitochondrial dysfunction.205 Iron-induced ROS recruit BAX to mitochondria via translocase of outer mitochondrial membrane 20 (TOMM20), promoting cytochrome C release, caspase-3 activation, GSDME cleavage and pyroptosis.206 ROS-induced TNF signaling can also drive RIPK/FADD-mediated caspase-8 activation to cleave GSDMD or GSDMC and induce pyroptosis.207 ROS elevate intracellular Ca2+ levels, triggering ER stress, Caspase-3 activation, and GSDME-mediated pyroptosis in TNBC cells.208 Additionally, ROS can trigger the ASK1-JNK-caspase-3-GSDME pathway to promote pyroptosis.209 Interestingly, ROS can inhibit pyroptosis through the PI3K/AKT-mTORC2 pathway, suppressing ASK1 activation. ROS-p53 signaling transcriptionally upregulates GSDME and caspases, switching apoptosis to pyroptosis.210 ROS stress can activate NF-κB, which induces the expression of inflammasome-associated genes. Additionally, ROS-induced AhR signaling can either activate the NLRP3 inflammasome,211 or inhibit macrophage pyroptosis by upregulating ornithine decarboxylase 1 (Odc-1), enhancing polyamine biosynthesis (e.g., spermine), which suppresses K+ efflux and inflammasome assembly, as observed in ulcerative colitis patients212 (Fig. 7a).
However, acute NLRP3 activation induces tumor-suppressive pyroptosis and antitumor immunity, whereas chronic activation promotes metastasis through IL-1β/IL-18-driven inflammation.213 Notably, GSDME expression enhances tumor cell phagocytosis by tumor-associated macrophages and boosts the activity of tumor-infiltrating natural killer cells and CD8+ T lymphocytes, amplifying the antitumor immune response.214

Oxieptosis
ROS oxidize specific cysteine residues (e.g., Cys151, Cys273, and Cys288) on KEAP1, which acts as a cellular ROS sensor.215 At low to moderate ROS levels, KEAP1 releases Nrf2, while at excessive ROS levels, KEAP1 releases PGAM5, a mitochondrial serine‒threonine phosphatase that dephosphorylates AIFM1 at Ser116 to activate the oxeiptosis pathway.216 This caspase-independent cell death process selectively removes ROS-damaged cells, preventing inflammation and preserving tissue homeostasis (Fig. 7a).

NETosis
ROS induce NETosis through both NOX-dependent and NOX-independent mechanisms. In NOX-dependent NETosis, NOX generates ROS, activating the MAPKs-ERK, p38 MAPK, and JNK pathways.217 This promotes the migration of myeloperoxidase (MPO) and neutrophil elastase (NE) to the nuclear envelope, where NE partially degrades histones, leading to chromatin decondensation and the release of neutrophil extracellular traps (NETs).6 In NOX-independent NETosis, ROS are produced by mitochondria and are triggered by stimuli such as calcium ionophores or ultraviolet light.218 Elevated calcium activates peptidyl arginine deiminase 4 (PAD4), causing histone citrullination, chromatin decondensation, and NET formation.6 Both pathways highlight ROS as key initiators of NETosis.
NETosis promotes chronic inflammation and cancer progression by inducing DNA damage, proliferation, and angiogenesis via MMP9 and VEGF.219 It facilitates metastasis via IL-17/granulocyte colony-stimulating factor (G-CSF) signaling, high mobility group box 1 (HMGB1) activation, and carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1)-mediated adhesion.220 NETosis shifts neutrophils to a protumor N2 phenotype, promoting an immunosuppressive TME, EMT, and chemoresistance.221 NETs shield tumor cells from immune detection, driving immunotherapy resistance via PD-L1 expression, T-cell exhaustion, and IL-8-mediated immunosuppression, while NETosis in the hypoxic TME enhances chromatin decondensation, tumor cell trapping, and immune evasion, making NETs a critical therapeutic target221 (Fig. 7a).

Cuproptosis
ROS induce cuproptosis through Fenton-like reactions, where copper cycles between Cu+ and Cu2+, generating toxic OH• that damages biomolecule and disrupts mitochondrial function.222 This oxidative stress promotes copper accumulation, leading to the aggregation of lipoylated TCA cycle proteins and the loss of iron‒sulfur cluster proteins—key hallmarks of cuproptosis. In cancer, this mechanism exploits the heightened copper demand and metabolic vulnerabilities of tumor cells, particularly those reliant on OXPHOS.222 By selectively targeting cancer cells with copper ionophores like Elesclomol or ROS-inducing agents, including radiotherapy or combination therapy, cuproptosis can be triggered, offering a promising strategy to overcome cancer.223,224 However, balancing copper toxicity and therapeutic efficacy remains critical to minimize off-target effects while maximizing anticancer outcomes (Fig. 7a).

ROS selectivity in signaling pathway modulation
ROS activate various signaling pathways through selective interactions dictated by the chemistry of specific ROS and the biochemical properties of target molecules. O2•⁻ reacts selectively, reducing transition metals and forming ONOO⁻ with nitric oxide, while also damaging Fe–S cluster enzymes critical to metabolism.11,225 H2O2 is less reactive but selectively oxidizes cysteine residues with low pKa values and reacts with transition metals to generate OH•.12 OH• is highly indiscriminate, reacts at near diffusion-controlled rates and drives oxidative damage.226 ONOO⁻ reacts with thiols, metals, and CO₂, generating oxidants such as NO2• and CO3•⁻.225 ¹O2 is a nonradical but highly reactive species, central to photobiological and oxidative processes.11
The redox potential of specific residues, particularly cysteine, plays a critical role in ROS selectivity. Cysteine residues with low pKa values (e.g., 4.7–5.4) are more nucleophilic and prone to oxidation. For instance, the catalytic cysteine in PTPs has a low pKa because of the Cys-X5-Arg motif, making them highly susceptible to oxidation by H2O2, leading to the formation of sulfenic acid (SOH), disulfide bonds (S‒S), or sulfenyl-amide (S‒N) bonds.227 Proteins containing redox-sensitive motifs, such as the CXXC motif in nucleoredoxin (NRX) or TRX, are preferential targets for ROS.228 These motifs are structurally designed to sense and respond to redox changes. For example, the oxidation of Cys205 and Cys208 in NRX disrupts its interaction with disheveled (Dvl), stabilizing β-catenin and activating the Wnt/β-catenin pathway. The sensitivity of cysteine residues to redox changes is influenced by their steric accessibility and the pKa value of their thiol groups. Typically, the pKa value of free cysteine is approximately 8.2. However, when cysteine residues are located near positively charged residues, their pKa decreases to less than 6.5, increasing their susceptibility to oxidation.229 This shift in pKa is associated with the transformation of redox-sensitive thiols into potent nucleophiles in the presence of basic residues. Cellular redox state might modulate the sensitivity of specific molecules to ROS and subsequent reactions. For example, reducing conditions strengthen some interactions (e.g., NRX-Dvl), and oxidizing conditions weaken them in modulating the Wnt/β-catenin pathway.230 Proteins such as TRX and glutaredoxin (GRX) might modulate ROS interactions by reducing oxidized cysteine residues, thereby restoring protein function. For example, TRX reduces oxidized ASK1, preventing its activation, whereas GRX regulates the redox state of proteins like Ras, influencing MAPK signaling.58

Mechanism of the ROS-mediated modulation of signaling pathways

PI3K/AKT/mTOR pathway
ROS enhance PI3K/AKT signaling by suppressing PTEN through H2O2-induced oxidation (Cys124/Cys71) or Ser380 phosphorylation, inhibiting membrane binding,231 forming disulfide bridges and promoting migratory shifts.232 Interestingly, PTEN degradation also occurs independently of H2O2, as observed in pancreatic cancer cells with elevated prostaglandin production via 5-LOX and COX-2 overexpression.232,233 H2O2 also oxidizes cysteines in PTPs, enabling receptor tyrosine kinase (RTK) phosphorylation,234 Insulin-induced ROS inhibit PTP1B, preventing insulin pathway dephosphorylation and promoting PI3K/AKT signaling.235 H2O2 also increases threonine phosphorylation in Caco-2 cells by downregulating PP2A via oxidation of its redox-sensitive Cys266 and Cys269, forming inhibitory disulfide bonds,236 particularly through oxidation of the B55α isoform.237 Notably, some studies suggest that ROS-mediated inhibition of the PI3K pathway regulates cell death and enhances chemotherapy sensitivity in cancers, although the exact mechanisms are unclear238,239 (Table 1, Fig. 7b).

Wnt/β-catenin pathway
NOX1 generates H2O2, which oxidizes the Cys205 and Cys208 residues of NRX.71 This oxidation inactivates NRX, which normally binds to disheveled (Dvl) and represses its ability to inhibit the β-catenin destruction complex, thereby enabling it to stabilize β-catenin, facilitating ROS-driven Wnt/β-catenin signaling.71 Thus, a redox-dependent interaction between NRX and Dvl is essential for Wnt/β-catenin signaling modulation by ROS (Table 1, Fig. 7b). Interestingly, NRX can indirectly promote β-catenin signaling by modulating Dvl stability. KLHL12, an E3 ubiquitin ligase adapter, targets Dvl for proteasomal degradation, negatively affecting the canonical Wnt pathway. However, reduced levels of NRX disrupt the interaction between KLHL12 and Dvl, preventing Dvl degradation and preserving the pool of Dvl necessary for sustaining Wnt/β-catenin signaling.240

Nrf2 pathway
Under stress, ROS oxidize Keap1’s cysteine-rich domain (Cys151),59 disrupting its interaction with Nrf2 and activating Nrf2. Consequently, Nrf2 levels increase in the cytoplasm, where Caspase-3-cleaved PKCδ further activates Nrf2 via Ser40 phosphorylation, enabling its nuclear translocation to activate ARE expression.241 Mitochondrial ROS activate macrophage-stimulating 1/2 (Mst1/2) kinases via autophosphorylation at Thr183/Thr180.242 Activated Mst1/2 relocates with Keap1 to mitochondria, phosphorylating its N-terminus, disabling Keap1 dimerization and subsequent Nrf2 activation243 (Table 1, Fig. 7b).

JAK/STAT pathway
ROS, particularly H2O2, activate JAK2 and TYK2 signaling by phosphorylating JAK2 (Tyr1007/1008) and TYK2 (Tyr1054/1055),77 leading to STAT activation via Tyr705 phosphorylation.244 In vascular smooth muscle cells, ROS activate c-Abl, which phosphorylates PKCδ (Tyr512/523),245 leading to PYK2 phosphorylation (Tyr402/579) and subsequent JAK2 transphosphorylation.246 Additionally, ROS activate SRC kinase through Cys-oxidative modification, initiating JAK/STAT signaling.247 Besides, ROS oxidize PTP1B, inhibiting its phosphatase activity and thereby preventing dephosphorylation of signaling components such as the IL-4 receptor, which prolongs STAT signaling248 (Table 1, Fig. 7b).

TGF-β pathway
Ionizing radiation-induced ROS activate latent TGF-β1 by oxidizing Met253 in its LAP, triggering conformational changes, suggesting the isoform specificity of ROS in TGF-β activation.249 This isoform-specific activation is due to variations in LAP homology (34–38%), despite the high similarity (75%) among TGF-β cytokines. ROS can activate LTGF-β indirectly by potentiating MMP2 and MMP9, which modify thiol residues (e.g., Cys73) to sulfenic acid,250 activating MMPs, leading to LAP cleavage and the release of active TGF-β in the ECM (Table 1, Fig. 8a).

HIF1α pathway
ROS regulate HIF1α by oxidizing Fe2+ to Fe3+, which inhibits prolyl hydroxylases (PHDs) that degrade HIF1α via hydroxylation.251 Notably, PHDs use iron, ascorbate, oxygen, and 2-oxoglutarate to hydroxylate and inhibit HIF1α, while ROS disrupt these interactions, causing HIF1α accumulation. Hypoxia-induced mito-ROS, particularly H2O2, can also regulate HIF1α by promoting succinate accumulation (which inhibits PHD expression) via the reverse succinate dehydrogenase (SDH) activity.252 Additionally, nitric oxide (NO) induces HIF1α expression by S-nitrosylating Cys800 or Cys520, preventing its binding to the Von Hippel‒Lindau (VHL) protein and proteasomal degradation.253 ROS can also indirectly activate HIF1α through pathways such as the AMPK, PI3K/AKT, and MAPK pathways, which phosphorylate HIF1α at specific residues (e.g., Ser419, Ser641, and Ser643)254,255 (Table 1, Fig. 8a).

AhR pathway
ROS, especially H2O2, activate the AhR pathway by accelerating the production of endogenous ligands like FICZ and kynurenine (Kyn).256 Kou et al. reported that H2O2 increases oxindole levels, a tryptophan catabolite, with its major form, 2-oxindole, effectively activating the AhR pathway.257 Besides, H2O2 and 2-oxindole were observed to promote PD-L1 and indoleamine-2,3-dioxygenase-1 (IDO1), two immune checkpoint proteins.257 In glioma cells, IDO1 and TDO were shown to infuse the transformation of Trp into Kyn, followed by AhR activation.258,259 Opitz et al. reported that Kyn is an endogenous oncometabolite that has the potential to activate AhR.258 Moreover, UV-induced ROS phototransform Trp into FICZ, a potent AhR ligand, highlighting the role of ROS in modulating the tumor milieu via AhR signaling.257 (Table 1, Fig. 8a).

NF-κB pathway
The NF-κB pathway involves five proteins (p50/p105, p52/p100, p65/RelA, RelB, and c-Rel) and operates through two pathways: the canonical pathway, triggered by receptors like TLRs and TNFR, activates p50-RelA via IKKβ and NEMO, while the noncanonical pathway, initiated by BAFF-Rs, activates p52-RelB through NIK and IKKα.260 NF-κB is typically inhibited by IκB proteins, which are degraded upon phosphorylation by IKKs, enabling NF-κB activation.
ROS, like H2O2 and superoxide, act as activators or inhibitors of NF-κB, while NF-κB has pro- or antioxidant effects depending on oxidative stress levels. Superoxide and H2O2 activate TAK1, which phosphorylates IKKβ at Ser177 and Ser181, enabling IKKβ to degrade IκBα at Ser32 and Ser36, triggering NF-κB signaling.261 TNF-α-induced ROS oxidize the dynein light chain (LC8), enabling its binding to IκB and promoting IκBα degradation by IKK, leading to NF-κB activation.262 Besides, externally added H2O2 directly phosphorylates IκBα at Tyr42 via SYK/SRC kinases, which induces the phosphorylation and nuclear translocation of the p65 protein.263
Excessive ROS, on the other hand, oxidize the Cys62 residue of NF-κB p50, reducing its DNA binding ability; Notably, this residue is prone to oxidation in the cytoplasm but sensitive to reduction in the nucleus.264 Moreover, prolonged and excessive ROS production may inactivate the proteasome, which plays a key role in degrading IκB, an inhibitor of NF-κB.263,265 H2O2 reduces TNF-α-induced IKK activity by oxidizing cysteine residues, impairing IκB degradation and diminishing NF-κB activation.266 The same outcome was reported in the case of arsenite267 and nitric oxide,268 as the Cys179 residue of IKKβ was oxidized,269 causing the inhibition of IKK.
In case of noncanonical pathway, IL-1β-induced H2O2 activates NIK, leading to IKKα phosphorylation (Ser176/Ser180), which converts p100 to p52, initiating NF-κB.270 Moreover, H2O2 was also shown to mediate NF-kB activation in an IKK-dependent way by regulating the PI3K/PTEN/AKT signaling cascade.271 However, this interaction remains debated owing to the complexity of the crosstalk between ROS and NF-κB signaling, as well as the variable outcomes in different experimental models (Table 1, Fig. 8a).

ATG4 pathway
H2O2 can induce the oxidation of the Cys81 (in the vicinity of the catalytic Cys77) residue of ATG4A and ATG4B into sulfenic acid or through reversible oxidation of Cys77 and Cys81 into disulfide bridges.272 However, Zheng et al. identified Cys292 and Cys361 as key sites of ATG4B oxidation function and subsequent autophagy mediation.273 Interestingly, Zheng et al. also reported that H2O2 treatment-induced oxidative modification of the Cys292 and Cys361 residues of ATG4 to restrict LC3 lipidation, leading to autophagy inhibition.273 The discrepancy between these studies is due to varied autophagy-inducing systems and levels of ROS274 (Table 1, Fig. 8a).

Calcium pathway
Several studies have reported ROS-induced Cys oxidation and subsequent overexpression of different TRP channels, which in turn activate Ca2+. influx. For instance, ROS oxidize the Cys616, Cys621/Cys258 and Cys742 residues of TRPV1;275 the Cys553 and Cys558 residues of TRPC5;276 and the Cys621, Cys421, Cys641 and Cys665 residues of TRPA1.277 H₂O₂ activates TRPM2 by targeting Glu829 and Arg845, producing NAD metabolites such as ADP-ribose, which opens calcium channels and induces cell death.278 ROS-mediated Ca2+ activation in antitumorigenic signaling involves the potentiation of the Ca2+ emission channel mucolipin-1 (MCOLN1/TRPML1) through palmitoylation of cysteine residues Cys565, Cys566, and Cys567 within the L564CCC motif on its C-terminal tail.279 However, it remains unclear whether all three cysteine residues undergo palmitoylation or if modifications to Cys565 and Cys567 are sufficient to optimize Cys566 palmitoylation (Table 1, Fig. 8a).

EGFR pathway
ROS promote ligand-independent initiation of EGFR in several ways, involving EGFR phosphorylation at the Tyr845 residue,280 and/or inhibition of EGFR inhibitory PTPs (at Cys124 and Cys71) and/or Cys797 oxidation of EGFR.281 Interestingly, EGFRT790M activation was observed to promote NOX2-regulated ROS formation, which in turn promoted the oxidation of the Cys797 and Met790 residues of EGFR.282,283 While Met790 oxidation is reversible, Cys797 oxidation on EGFRT790M further upregulates EGFR-NOX2, which increases ROS production, resulting in aberrant EGFR expression. This excessive phosphorylation of EGFR with a disrupted dimer structure leads to the development of TKI resistance in ROS-exposed EGFR-sensitive NSCLC cell lines.282,283 On the other hand, high ROS levels have been reported to induce apoptosis and surmount TKI resistance, as when exposed to excessive ROS, ASK1 potentiation results in the initiation of JNK and p38 MAPK and the induction of apoptosis.284 Leung et al. suggested that Sanguinarine aggregated excessive ROS by promoting the activation of NOX3 and consequential inactivation of MsrA and Cys peroxidation of EGFRT790M in EGFR-modulated TKI-resistant cells, which ultimately led to EGFR degradation and cell death284 (Table 1, Fig. 8b).

Ras/MAPK (ERK) pathway
The most prominent mechanism of ROS-mediated ERK activation starts with the potentiation of EGFR. Once activated, EGFR interacts with SHC, leading to Tyr317 phosphorylation, which facilitates Grb2 binding and the formation of the SHC-Grb2-SOS complex, which is modulated by hydrogen peroxide.281 After relocating to the plasma membrane, SOS activates Ras by promoting GTP binding.281 Activated Ras initiates Raf, which phosphorylates MEK1 (Ser218, Ser222) or MEK2 (Ser222, Ser226), and MEK1/2 then activates ERK by phosphorylating Tyr185 and Thr183.285 In addition, SRC kinases activated by ROS mediate ERK activation via Raf or EGFR signaling.286 Furthermore, ROS can mediate the Ras-dependent activation of ERK by oxidizing the Cys118 residue of Ras. In addition, commensal bacterium-induced ROS oxidize DUSP3 Cys124, relieving ERK inhibition and activating ERK287 (Table 1, Fig. 8b).

JNK (MAPK) pathway
ROS-induced JNK pathway activation involves MAPKKK (MEKK1/2/3/4, MLK, and ASK1)-induced phosphorylation of a MAPKK (MEK3/4/6/7) at Ser or Thr residues, which then phosphorylates a highly critical Thr-Pro-Tyr (TPY) motif of JNK, leading to the induction and nuclear translocation of JNK.288 ROS reversibly oxidize TRX at the Cys32 and Cys35 residues, dissociating it from ASK1 (activation), which, in turn, activates JNK.289 Additionally, ROS dimerize GSTP, disrupting its ability to suppress JNK, thereby enhancing JNK activity, with potential dimerization sites at Arg70, Arg74, Asp90, Asp94, and Thr67.290 Besides, ROS directly activate ASK1 through homo-oligomerization and autophosphorylation at Thr838 or hetero-oligomerization with ASK2, where ASK2 phosphorylates Thr838 of ASK1 after its own autophosphorylation at Thr806291 (Table 1, Fig. 8b).

p38 (MAPK) pathway
ROS-induced p38 activation mechanism involves several initial proteins involved in the JNK pathway, such as ASK1. ROS directly or indirectly affect ASK1, MEKK1/2/3/4, and MLK3, which subsequently activate MEK3/6 (highly conserved in the p38 pathway).288 MEK activation further induces double phosphorylation of a critical TPY motif of p38 on tyrosine and threonine residues.288 Besides, ROS have been found to mediate heterodimerization between Cys104 of MKK3 and Cys119 and Cys162, which forms a disulfide bond between p38α and MKK3, leading to p38α overexpression292 (Table 1, Fig. 8b).

PARP1 pathway
Four PARP-1 domains—F1, F3, WGR, and CAT (catalytic)—are crucial for DNA damage-dependent activity, whereas the CAT domain, comprising the Helical Domain (HD) and ADP-ribosyltransferase (ART), is vital for activation.293
Under oxidative stress, ROS-induced DNA damage triggers HD unfolding via a “leucine switch” (Leu698 and Leu701 repositioning), relieving autoinhibition and enabling PARP1 DNA binding and autoPARylation.293 Regulatory factors like histone PARylation factor 1 (HPF1) modulate PARP1 activity during stress-induced DNA damage. HPF1 interacts with several critical residues (Trp1014, Ser1012, Leu985, His826, and Leu1013) in the catalytic domain of PARP1 through its Lys307, Asp283, Cys285, Phe268, and Phe280 residues in the C-terminal domain.294–296 This binding induces ADP-ribosylation of serine residues in the HD domain of PARP1 (Ser499, Ser507, and Ser519).295,296 HPF1 also recruits Arg239 to position Glu284 for serine ADP-ribosylation, relieving PARP1 HD autoinhibition. The PARP1-HPF1 complex enhances ADP-ribosylation on Asp/Glu/Ser residues in histones, PARP1, and chromatin factors294,296 (Table 1, Fig. 8b).

ATM pathway
At moderate ROS levels, a few enzymes like PRDX2 and TRX1, can induce chemical alteration of the Cys2991 residue on ATM to generate a reversible disulfide bond.297 Upon dimerization, ATM undergoes autophosphorylation at the Ser1981 residue, leading to the initiation of ATM.298 Under excessive ROS levels, ATM triggers downstream sites such as p53 and checkpoint kinase 2 (CHK2) for activation even when the DNA damage response (DDR) is absent, indicating that the activation pathway has no impact on the role of ATM at a certain level, whereas ATM has many other functions independent of the DDR297 (Table 1, Fig. 8b).

p53 pathway
Upon oxidative DNA damage, ROS-activated ATM autophosphorylates and activates p53 (via Ser15 phosphorylation) and CHK2, which further phosphorylates p53 at Ser20.299 Stress-activated JNK and p38MAPK have also been shown to affect p53 activation upon exposure to H2O2.299 p38 MAPK has been found to induce the phosphorylation of p53 at Ser15, 33, 37, and 46,299 whereas JNK has been implicated in the phosphorylation of p53 at the Thr81 and Ser20 residues.300 As mentioned earlier, H2O2 expression results in dimerization and subsequent initiation of ASK1, which initiates p38 MAPK and JNK, stabilizing p53.288 Notably, DNA damage-induced p53 activation is typically modulated by ATM, whereas redox signaling-dependent p53 activation is regulated by p38 MAPK signaling.299 Although diamide-induced ROS initiate both JNK and p38 MAPK, p53 activation by diamide is regulated entirely by p38 MAPK.299 Interestingly, while H2O2 activates all the ATM, JNK, and p38 MAPK signaling, H2O2-induced p53 activation requires only ATM and JNK299 (Table 1, Fig. 8b).

AMPK pathway
H2O2 activates AMPK in HEK-293 cells via an AMP/ADP-dependent pathway involving mitochondrial respiratory chain inhibition.301 Another mechanism by which hypoxia-induced ROS activate AMPK is via an AMP-independent pathway involving Ca2+ release from the ER. Stromal interaction molecule 1 (STIM1) detects ER calcium depletion, translocates to ER-plasma membrane junctions, and activates Ca2+ release-activated channels (CRAC) by tethering Orai proteins, triggering CaMKKβ activation, which phosphorylates AMPK at Thr172.302 In addition, H2O2 directly activates AMPK in HEK-293 cells by oxidatively modifying Cys299 and Cys304 of the AMPKα subunit303 (Table 1, Fig. 8b).
However, Hinchy et al. reported decreased AMPK activity after 30 min of H2O2 addition, as ATP/ADP ratios were restored and PRDX-SO2/3 formation was increased.304 The underlying mechanism might be that oxidative stress oligomerizes PRDXs, which may compete with other cellular proteins for thioredoxin-catalyzed cysteine thiol reduction.

Exploring ROS and ROS-driven mechanisms for therapeutic targeting
In pancreatic cancer, oxidative stress initially activates TIGAR and Nrf2 for survival, but prolonged stress reduces TIGAR, increases ROS, and promotes metastasis.305 Blocking antioxidants delays tumor initiation but accelerates metastasis, whereas antioxidant intervention (e.g., N-acetyl-L-cysteine) suppresses metastasis in TIGAR-deficient models. In melanoma and lung cancer, antioxidants have been shown to promote metastasis, underscoring the potential efficacy of a pro-oxidant approach in this context.306,307
Additionally, the dual roles of ROS enable the development of context-specific therapeutic strategies, ranging from ROS induction to scavenging, for combination cancer therapies. Cancer cells exhibit exceptional mitochondrial plasticity and metabolic reprogramming, allowing them to adapt to the nutrient-deficient tumor microenvironment and evade single-agent therapies by activating alternative survival pathways.308 Although excessive ROS can effectively induce cancer cell death and serve as a promising anticancer strategy, they may also cause immunosuppression. Thus, combining ROS-elevating therapies with immunotherapy could optimize outcomes by counteracting immune suppression while targeting cancer cell resilience.
Targeting downstream molecules critical for ROS-mediated pathways, such as the NRX (Wnt/β-catenin), IKK (NF-κB), and ASK1 (MAPK) pathways, might be a potential antitumor therapeutic strategy. By targeting these signaling nodes, this strategy precisely modulates ROS-driven pathways, counteracting ROS-induced signaling without necessarily altering ROS levels. Finally, the heterogeneity of redox responses across cancer types, patients, and stages emphasizes the need for personalized approaches. A comprehensive set of parameters—including redox status, antioxidant enzyme expression, cell signaling profiles, and cancer-specific signaling—constituting a “redox signaling signature” for individual patients, is currently awaiting development.309 Notably, regardless of the therapeutic strategy, ROS-targeted therapies fall into two main categories: modulating redox adaptation mechanisms (antioxidant defenses) and targeting ROS-generating systems (e.g., NOX, iNOS, and mitochondria).

Therapeutic leverage of ROS depletion in chemoprevention and cancer treatment
Antioxidants are commonly used as over-the-counter supplements to prevent cancer or in combination with chemo-/radiotherapy to reduce side effects and improve patients’ quality of life during treatment. This section and Table 2 summarize antioxidant agents with antineoplastic potential that modulate ROS in cancer and their ROS-mediated mechanisms of action.

Modulation of GSH metabolism
N-acetyl-L-cysteine (NAC) supports GSH metabolism by providing cysteine for GSH replenishment. A preclinical study suggested its selective anticancer potential by depleting ROS below the levels required for telomerase activation, disrupting redox homeostasis, and suppressing cancer cell proliferation.310 Interestingly, NAC paradoxically lowers the GSH/GSSG ratio, as cancer cells require some ROS to maintain antioxidant capacity. Besides, reduced glutathione and NOV-002 are being clinically investigated for their potential to improve the quality of life of patients receiving chemo-/radiotherapy.

Inhibition of NOX-mediated ROS generation
Agents capable of targeting NOXs to scavenge ROS could offer considerable potential for cancer therapy. Among several small-molecule NOX inhibitors, Ebselen is under clinical study for mitigating chemotherapy-induced toxicity.311 Diphenylene iodonium (DPI) suppresses early tumorigenesis in colitis-associated cancer by inhibiting ROS-driven NF-κB, STAT3, and ERK pathways, reducing inflammation.312 GKT137831 disrupts the NOX4-ROS-YY1-IL-8-PD-L1 axis, lowering IL-8 and PD-L1 levels to impair angiogenesis, immune evasion, and Gefitinib resistance.313 The optimized successor of GKT137831, Setanaxib, overcomes αPD-1/PD-L1 resistance by targeting NOX4 in cancer-associated fibroblasts, reversing their immunosuppressive phenotype, and enhancing CD8+ T-cell infiltration.314 Cyclodiaryl-iodonium (CDAI) kills pancreatic cancer cells by blocking NOX, reducing nonmitochondrial ROS while inducing lethal mitochondrial ROS, disrupting energy metabolism and triggering p53/NF-κB/GnRH pathway chaos with minimal toxicity.315 Fangchinoline inhibits NSCLC metastasis by reversing EMT and suppressing the NOX4-ROS-AKT-mTOR pathway.316

Direct ROS scavenging
ROS scavengers directly neutralize ROS by donating electrons or hydrogen atoms. ROS scavengers, such as Vitamin E, Vitamin C, Vitamin D, Melatonin, Carotenoids, and Lycopene, interact with ROS by donating electrons or hydrogen atoms to free radicals. To date, numerous studies have explored the antineoplastic potential of ROS scavengers. Among them, a phase I clinical trial (NCT00985777) provided evidence that Tocotrienol (Vitamin E) may induce apoptosis in PDAC cells.317 Regarding anti-inflammatory activity, Vitamin D3 has been reported to decrease TNF-α-induced inflammation in lung epithelial cells through a reduction in mitochondrial fission and mitophagy.318 A study reported that Melatonin inhibits angiogenesis by downregulating the HIF1α/ROS/VEGF pathway in HUVECs,319 while its pro-oxidant activity may also contribute to anticancer effects, highlighting a context-dependent mechanism that warrants further investigation.320 A phase II trial (NCT00450749) suggested that Lycopene may slow prostate cancer growth, although the results were inconclusive. In PANC-1 cells, lycopene inhibits ROS-mediated NF-κB signaling and induces apoptosis.321

Keap1 inhibition
Keap1 inhibition activates Nrf2 to upregulate antioxidant enzymes that neutralize ROS. For example, in cutaneous T-cell lymphoma (CTCL), DMF therapy has shown clinical efficacy by lowering mSWAT scores, relieving pruritus, and inhibiting NF-κB, as demonstrated in the NCT02546440 trial.322 Sulforaphane targets Cys151 to inhibit the Keap1–Nrf2 interaction, boosting NAD(P)H:NQO1 expression and reducing lung inflammation in vivo.323 A clinical trial (NCT00843167) revealed that Sulforaphane reduces the expression of Ki-67 and other proliferation markers in DCIS and high-risk lung cancer patients.324 Another trial (NCT01265953) revealed that Sulforaphane downregulates the prostate cancer-associated genes AMACR and ARLNC1.325 Besides DMF and Sulforaphane, other Keap1 inhibitors include Resveratrol, Quercetin, RTA-402, and RTA-408. Resveratrol induces caspase-3-mediated apoptosis (39% increase; NCT00920803) and reduces Ki-67-linked proliferation (5% decrease; NCT00433576) in hepatic metastases and colorectal cancer.326,327 Quercetin activates Nrf2/Keap1 to reduce ROS,328 while inducing ROS-dependent cell death in malignant cells,329,330 underscoring its potential as both a chemopreventive and chemotherapeutic agent. Bardoxolone methyl (RTA 402) upregulates Nrf2 targets, suppresses NF-κB and cyclin D1, and induces tumor regression in advanced cancers. In trials (NCT00529438, NCT00508807), it achieved responses in several hematologic and solid tumors, with stabilization lasting 6–12 months.331,332 Omaveloxolone (RTA 408) has also been reported to induce time- and dose-dependent activation of Nrf2 antioxidant genes (NCT02029729). In addition, Genistein and Procyanidin B2 reduce carcinogen-induced ROS and DNA damage via Nrf2/ARE signaling in bronchial epithelial cells.333

Utilization of SOD mimetics
Although SOD mimics have a lower rate constant than natural SOD enzymes, they effectively function in extracellular fluids lacking antioxidant enzymes. SOD mimics include Metalloporphyrins, Mn (II) polyamines, Mn(III) salens, Mn(III) corroles, and Mn(IV) biliverdins.334 The Mn(II)-based compound GC4419 showed safety but no remission in mCRPC (NCT01080352). Additionally, certain SOD mimics like Motexafin gadolinium, act as both prooxidative and antioxidative agents.

Xanthine oxidoreductase (XOR) inhibition
Inhibiting XOR (Xanthine oxidase and Xanthine dehydrogenase) offers a promising cancer therapeutic approach for reducing ROS generation during purine catabolism, thereby mitigating DNA damage, inflammation, and tumor progression.335 Preclinical studies have shown that XOR inhibition enhances the effectiveness of the TKIs currently used in clinics against BCR-ABL in CML.336
Hericium erinaceus, an edible mushroom, may treat breast cancer by inhibiting Xanthine oxidase, scavenging ROS, and suppressing cancer cell growth with minimal impact on normal cells.337

iNOS inhibition
iNOS (inducible NOS) is highly overexpressed in many cancers and is correlated with poor prognosis in TNBC, making it a promising biomarker and therapeutic target.338 Therefore, inhibiting iNOS can reduce ROS production, suppress tumor growth, and modulate immune responses. Andrographolide was shown to exert antitumor activity across multiple cancers by inhibiting iNOS.339 In the NCT02834403 trial, combining L-NMMA with Taxane showed efficacy in treating chemorefractory breast cancer, particularly in advanced cases, with manageable toxicity.340

Therapeutic leverage of ROS elevation in cancer treatment
Since the 1950s, various strategies/drug research have been implemented based on this concept, such as administering ROS or ROS-generating enzymes to tumor cell lines or murine models with tumors.335 As documented in this section and Table 3, antineoplastic agents/methods with ROS-inducing potential and their ROS-mediated antitumor mechanisms—compiled from sources including the FDA database (www.fda.gov), ClinicalTrials.gov, and DrugBank (http://go.drugbank.com)—demonstrate diverse mechanisms of action ranging from metabolic interference to direct oxidative damage.

GSH depletion
GSH inhibition causes severe ROS accumulation and increased oxidative stress, leading to cancer cell death.341 BSO, a gamma-glutamylcysteine synthetase inhibitor, disrupts redox balance and sensitizes cancer cells to chemotherapy,342 while Imexon generates ROS, causing DNA damage and apoptosis in multiple myeloma.343 The hybrid anticancer drug OSamp depletes glutathione and generates ROS, selectively inducing oxidative stress and apoptosis in cancer.344 Similarly, PEITC, a cruciferous phytochemical, selectively kills cancer cells by targeting GSH (along with ETC-III and the UPR) to induce ROS-mediated cell death,345 with its synthetic analog LBL21 targeting stem-like cancer cells in NSCLC.346 A trial revealed that Nutri-PEITC jelly improved PFS and quality of life in advanced oral cancer patients, which was correlated with p53 levels.347 Benzoyloxy dibenzyl carbonate (B2C) generates Quinone Methide (QM) intermediates (GSH scavengers) upon hydrolysis, depleting GSH, enhancing ROS accumulation, and showing anticancer therapeutic potential.348 Additionally, a recent study identified ACZ2 as a dual GSH/TRXR inhibitor, demonstrating its potential as a ROS-targeted therapeutic against gastric cancer.349 Besides, DMF, another KEAP1-inhibiting antioxidant, paradoxically depletes GSH in MS patients, increasing ROS and mitochondrial stress to induce caspase-mediated memory T-cell apoptosis, thereby reducing neuroinflammation (NCT02461069).350 Many other studies have also reported ROS inducing potential of DMF,351,352 suggesting selective anti-inflammatory mechanism of DMF that warrants further investigation.

Thioredoxin inhibition
The dysregulated TRX system often drives tumor progression and therapy resistance, making its targeted inhibition a potent ROS-mediated anticancer strategy. PX-12, a TRX1 inhibitor that binds to Cys73 of TRX, has shown promise across various cancers in preclinical and early clinical trials.353 Another TRX1 inhibitor, Diallyl Trisulfide (Allitridin), a garlic-derived compound, inhibits TRX1 via Michael addition (Cys32 and Cys35), radiosensitizing glioblastoma.354 Arsenic trioxide (ATO), an FDA-approved APL treatment,355 and the repurposed antirheumatic drugs, Auranofin (AF) and Motexafin gadolinium (MGd) inhibit TRXR, inducing ROS to exert antitumor activity.356,357 Curcumin has demonstrated anticancer activity through TRXR inhibition in preclinical and early clinical studies.358 Piperlongumine (a dual TRXR/GSH inhibitor) demonstrates selective cytotoxicity in colon cancer.359 Shikonin’s dual inhibition of TRXR and PKM2 induces necroptosis, whereas Nrf2 hyperactivation mediates resistance, which can be reversed through metabolic targeting.360 Novel agents such as LW-216 (Sec498-targeted TRXR1 degradation),361 6-shogaol (selenocysteine inhibition),362 and Nitrovin induce caspase-3-independent paraptosis via MAPK/Alix modulation.363

SOD inhibition
Inhibiting SOD, especially SOD1, prevents superoxide conversion, increasing ROS, thereby selectively targeting cancer cells. ATN-224 (Tetrathiomolybdate), a copper-chelating agent, depletes cellular copper, a cofactor essential for SOD1 activity, causing oxidative stress and cancer cell death, with promising preclinical and clinical results.364,365 A recent study suggested that the E3 ligase TRIM22 functions as a tumor suppressor in breast cancer by degrading CCS (copper chaperone of SOD1), thereby reducing the copper supply to SOD1, impairing SOD1 activity, increasing ROS levels, and suppressing STAT3-dependent oncogenic pathways.366

NQO1 inhibition
NQO1, overexpressed in cancers but minimally expressed in normal tissues, offers a tumor-specific therapeutic target through redox cycling, generating cytotoxic ROS. β- Lapachone (ARQ761) induces DNA damage, PARP1 hyperactivation, and apoptosis in an NQO1-dependent manner (NCT02514031).367 Another NQO1 substrate, ARQ501, was shown to cause stable disease in a phase 1 trial,368 and DNQ is 10-fold more potent than β-lapachone.369 However, these agents cause methemoglobinemia, a limitation overcome by DNQ derivative Isopentyl-DNQ (IP-DNQ) in pancreatic cancer.370 Cryptotanshinone (CTS) binds NQO1 noncatalytically, triggering ROS/JNK1/iron/PARP/calcium-mediated necrosis,371 while Daphnetin inhibits NQO1 to induce ferroptosis in ovarian cancer.372

Inhibition of glutamine (Gln) metabolism
Inhibiting glutamine metabolism disrupts GSH and NADPH synthesis, elevating ROS, which impairs tumor growth and induces cancer cell death.129 Glutaminase inhibitors, such as CB-839 (Telaglenastat), BPTES, and Compound 968, disrupt Gln metabolism, depleting GSH and increasing ROS to induce cancer cell death.129 Additionally, the ASCT2 inhibitor, V-9302 blocks Gln uptake, impairing tumor growth and enhancing immunity via ROS-mediated B7H3 degradation.373 A novel glutaminase inhibitor, BIX01294, suppresses pancreatic cancer by inhibiting KDM6B-mediated glutaminase expression, thereby disrupting redox balance.374

Glycolysis inhibition
Glycolysis inhibitors (2-DG, Metformin, Lonidamine, and 3-BP) target hexokinase to disrupt the Warburg effect in cancer cells, impairing energy production and inducing ROS-mediated selective cell death.375 In a recent study, Chloroquine (CQ) was found to inhibit hypoxia-induced growth of colorectal cancer cells by suppressing glycolysis and NAD⁺ production through PDK1 inhibition.376 This leads to mitochondrial dysfunction, excessive ROS generation, reduced membrane potential, and apoptosis via PARP cleavage and caspase activation. Despite their antitumor efficacy, glycolysis inhibitors lack selectivity because of the role of glycolysis in normal cells; targeting cancer-specific isoforms (e.g., PKM2) or using nanocarriers could enhance specificity and reduce toxicity.

Inhibition of GPXs
GPX inhibition disrupts redox balance, inducing oxidative stress and offering a ROS-targeted cancer therapy approach.14 For instance, Lenvatinib blocks GPX2 in liver cancer, causing ROS buildup and apoptosis.377 Similarly, targeting xCT/GPX4, a critical regulator of lipid peroxidation, with compounds such as FIN56, Erastin, Sulfasalazine and RSL3 induces ferroptosis.378–381 Interestingly, a recent study suggested that RSL3 may induce cell death primarily through the inhibition of TRXR1 rather than the suppression of GPX4, highlighting the need for further studies to elucidate the precise mechanisms by which RSL3 triggers ferroptosis.380 FINO2 indirectly inhibits GPX4 activity (unlike Erastin and RSL3) without depleting GPX4 protein (unlike FIN56), while also oxidizing iron to drive ferroptosis.382

Inhibition of the ubiquitin–proteasome pathway
Proteasome inhibitors represent a promising anticancer strategy by disrupting protein homeostasis and inducing ROS.383 Bortezomib treats multiple myeloma by blocking the 26S proteasome, increasing ROS, and upregulating proapoptotic proteins such as NOXA.384 Disulfiram (in combination with copper) has been shown to inhibit both the proteasome and SOD1, amplifying oxidative stress.222 Experimental inhibitors such as MG132, Epoxomicin, Marizomib, and Lactacystin also induce ROS-dependent apoptosis across various cancers.385 Another novel proteasome inhibitor, CEP-18770, demonstrates superior and more sustained dose-dependent suppression of tumor proteasome activity than Bortezomib in medulloblastoma.386

GSTP inhibition
Overexpression of GSTP is frequently observed in multiple cancers, positioning it as a compelling therapeutic target for ROS-driven anticancer strategies.387 While known inhibitors (e.g., Ethacrynic Acid derivatives, NBDHEX, TLK199, LAS17, and CNBSF) exist, their ROS-linked mechanisms remain incompletely characterized.387 A recent study identified GSTP1 as the target of Tryptanthrin (TRYP), which binds and inhibits GSTP1, triggering ROS accumulation, DDR, and senescence initiation.388

Nrf2 inhibition
Nrf2, frequently overexpressed in therapy-resistant cancers, represents a promising therapeutic target for ROS-mediated anticancer strategies, as demonstrated by Tamoxifen-resistant breast cancer cells that develop irradiation cross-resistance through Nrf2-driven antioxidant upregulation and a glycolytic shift, with Nrf2 inhibition restoring radiosensitivity in both cellular models and patient tumors (NCT00738777).389 This approach is further validated by the agent β-Sitosterol, which indirectly modulates Nrf2 through ASS1 and then induces ROS-mediated cell death in ovarian cancer.390

NOX activation
NOX activation drives ROS generation in cancer therapy, exemplified by ZIO-101 (Darinaparsin), which induces NOX-mediated ROS production and mitochondrial disruption, triggering apoptosis in hematologic and solid tumors.391 While its precise mechanism—including potential NOX subunit (e.g., p67phox, p47phox) involvement—requires further elucidation, ZIO-101’s dual action of ROS induction and mitochondrial impairment effectively promotes cancer cell death.56,391 A study on recurrent glioblastoma observed that elevated inositol 1,4,5-triphosphate (IP3) kinase B (ITPKB)—due to reduced Trim25-mediated degradation—suppresses NOX enzyme activity, lowering ROS production and enabling resistance to Temozolomide (TMZ). Depleting ITPKB or inhibiting it with GNF362 reactivated NOX, increased ROS levels, and restored TMZ sensitivity in resistant GBM cells.392

Targeting ETC
Elesclomol and Fenretinide exploit this mechanism: Elesclomol binds to mitochondrial copper, disrupting ETC function and increasing ROS,393 whereas Fenretinide interferes with mitochondrial complex II to elevate superoxide.394 Loss of Fhit protein is observed in most cancers. Atovaquone, a repurposed antimalarial drug, acts as a selective OXPHOS inhibitor by targeting the CoQ10-dependent mitochondrial complex III, causing electron leakage, ROS production, and cell death, as observed in Fhit-deficient lung cancer cell lines.395 Phenformin, a metformin analog, inhibits complex I to generate ROS and shows promise in preclinical and early-phase trials for treating HCC and glioma.396

Pharmacological ascorbate
Ascorbate (vitamin C) acts as a pro-oxidant under certain conditions, generating ROS in the presence of transition metals like iron.397 In a Phase II trial (NCT02420314) for advanced NSCLC, Ascorbate combined with Carboplatin and Paclitaxel inhibited tumor growth through H₂O₂ generation, iron modulation (higher transferrin linked to improved PFS), and immune activation (4.2-fold increase in CD8+ T cells), achieving a 34.2% response rate and a 12.8-month median OS.398 Another trial (NCT02344355) in glioblastoma patients reported improved survival (19.6 vs. 14.6 months), with the greatest benefit in IDH-mutant tumors (53.1 months) and those with high MRI-detected iron, where vitamin C-induced Fe3+ reduction to Fe2+ enhanced chemo-radiation sensitivity.397

Conventional anticancer agents
Chemotherapeutic drugs (e.g., Doxorubicin, Paclitaxel, Cyclophosphamide, and 5-Fluorouracil) often induce ROS via a secondary mechanism (including mitochondrial dysfunction, DNA damage, depletion of NAD+ and ATP, and destabilization of microtubules) to enhance their cytotoxic effects.399 That their anticancer efficacy can be regulated by exposure to antioxidants suggests that their ROS-inducing capacity can be exploited for developing cancer therapies. Ionizing radiation (IR) produces ROS via water radiolysis, causing oxidative damage.400 PDT employs photosensitizers that generate ROS, primarily ¹O2, upon light activation, causing oxidative damage and inducing cancer cell death.401 PDT has shown efficacy in Phase III trials for basal cell carcinoma, improving tumor reduction and quality of life, but is limited by light penetration depth and photosensitivity. A previous study suggested that MLu (lutetium texaphyrin), an analog of MGd, acts as a photodynamic sensitizer under light irradiation to generate excessive ROS, which activate caspase-3. Caspase-3 then cleaves GSDME, releasing its pore-forming N-terminal fragment, thereby leading to pyroptosis.402
In addition to conventional approaches, several strategies targeting metabolic vulnerabilities and stress response pathways have emerged as promising ROS-mediated antitumor therapies. ER stress induction—via agents like Bortezomib or natural compounds such as Triptolide—triggers UPR activation and ROS-mediated cell death in malignancies with a high proteostatic burden.403 IDH1/2 inhibition disrupts NADPH homeostasis, weakens redox defenses and induces oxidative damage. Vorasidenib, a brain-penetrant IDH1/2 inhibitor, showed efficacy in a Phase III trial for IDH-mutant glioma (NCT04164901) by reducing GSH, increasing ROS, and enhancing therapeutic effects.404 DDR pathway inhibitors (e.g., ATM, CHK1, and PARP inhibitors) exacerbate oxidative stress by impairing repair processes and antioxidant functions.405

ROS-responsive pathway/cellular process-targeted therapy: potential targets
This approach is particularly relevant in diseases like cancer, where ROS dysregulation promotes pathological modulation of cellular processes/pathways. This section and Table 4 include different potential targets specific to different ROS-mediated cellular processes or pathways, as well as potential therapeutic strategies for mitigating cancer.

Potential targets in the ROS-responsive pathway
As PTP inhibition might modulate several oncogenic pathways, developing small molecules to activate or protect PTPs from oxidative modification offers therapeutic potential. A study observed TRXR1 to exhibit potential in protecting PTP1B from oxidative inhibition by H2O2.406 NRX, a redox-sensitive regulator of Dvl proteins, modulates Wnt/β-catenin signaling; its ROS-mediated inactivation promotes excessive self-renewal of hematopoietic stem/progenitor cells (HSPCs) in myelodysplastic syndrome (MDS).407 Although no direct activators exist, antioxidants (e.g., NAC, Ascorbic acid) and redox modulators (e.g., Ebselen, Curcumin) may influence NRX activity. Notably, as a TRX subfamily member, NRX itself supports antioxidant defense by protecting enzymes like catalase from ROS-induced oxidation, enhancing H2O2 detoxification.408 Furthermore, its downregulation might reduce β-catenin levels and cause a carcinogenic shift,240 warranting further exploration of its therapeutic potential in redox control and Wnt signaling.
Targeting JAK2 in JAK/STAT pathway might offer a promising strategy to disrupt the ROS-STAT pathway. JAK2 inhibitors such as AG490 and Ruxolitinib reverse ROS-induced STAT activation and mitigate pSTAT3-driven ICAM-1 and PD-L1 induction.409 Similarly, inhibiting Ras and MEK1/2 offers another anticancer therapeutic approach by blocking the ROS-ERK pathway. For example, Alendronate sodium blocks Ras prenylation to induce G1 arrest and apoptosis by inhibiting ROS-mediated ERK1/2 signaling.410 The MEK inhibitor PD98059 suppresses ROS-induced ERK1/2 activation, reducing glial fibrillary acidic protein (GFAP) expression (a marker of gliosis).411
Targeting ASK1 activation in the JNK and p38 pathways represents another potential therapeutic strategy, as various compounds can induce apoptosis through the ROS-ASK1-JNK/p38 axis. Flexicaulin triggers ROS/ASK1/JNK activation in esophageal cancer,412 while Quercetin induces AMPK-α, ASK1, and p38/caspase-dependent apoptosis.413 Daidzein and Gefitinib synergistically promote c-Jun translocation via this axis.414 Additionally, Catechol-containing polyphenols induce ROS via tyrosinase-mediated oxidation, triggering ASK1/JNK/p38 signaling and apoptosis in A375 melanoma cells.415

Potential targets in ROS-responsive cellular processes
The BCL-2 family and caspase family are potential therapeutic targets in ROS-mediated apoptosis.157,158 Targeting these molecules involves many pathways and molecules, including ER stress pathways (UPR sensors, CHOP, and calcium signaling), p53 tumor suppressors (which regulate the BAX/BCL-2 balance), mitochondrial apoptosis (cytochrome c release and caspase activation), and antioxidant systems (TRXR inhibition and GSH depletion).44,157,158 Therefore, therapeutic strategies involve ROS-inducing agents (prooxidant drugs, TRXR/GSH inhibitors), ER stress inducers, p53 reactivators, and combination therapies (e.g., with PARP inhibitors or immunotherapy).
The ferroptosis defense network includes multiple parallel antioxidant systems—xCT/GPX4, FSP1/CoQ10, DHODH, and GCH1/BH4—which prevent lipid peroxidation, presenting key therapeutic targets. Besides xCT/GPX4 inhibition (by Erastin/Sulfasalazine/RSL3),379–381 FSP1 inhibitors (iFSP1, Liproxstatin-1) and DHODH blockers (TPP-brequinar/B2, Leflunomide) disrupt organelle membranes and mitochondrial antioxidants, respectively.416–419 GCH1 suppression (by Gambogenic acid) amplifies ROS by depleting BH4, as evidenced in NSCLC.420 Notably, multitarget agents like Brequinar (DHODH/FSP1) exploit the system’s redundancy, selectively targeting cancer cells, especially in OXPHOS-dependent or immunotherapy-resistant malignancies.421 The NLRP3 inflammasome-GSDMD-caspase-1/NF-κB axis represents a pivotal but context-dependent target in ROS-mediated pyroptosis, but NLRP3 has dual roles in disease pathogenesis.422 Therefore, precision modulation is critical: inhibitors (MCC950) for inflammatory diseases versus activators (Saikosaponin-D, Baicalin) for immunogenic tumor killing.422–424 Additionally, emerging agents like Gambogic acid targets ROS/caspase-3/GSDME, while Shikonin, GEFT, and Paris saponin-7 modulate ROS-GSDM signaling to induce pyroptosis in several cancers.425–428
Targeting the Keap1/PGAM5/AIFM1 axis in ROS-mediated oxieptosis could selectively amplify oxidative stress to induce programmed cell death in cancer cells, offering a novel therapeutic strategy. In cancer cells, including CRC cells, ROS inducers like Sanguinarine (SNG) trigger Keap1-PGAM5-dependent cytotoxicity by inducing oxieptosis.216
RIPK1/3 and MLKL, as central mediators of necroptosis, are promising therapeutic targets in ROS-driven necroptosis.184 However, RIPK1 modulation is governed by phosphorylation, with activation driven by autophosphorylation at Ser161 and Ser166 and ROS-induced modifications.429 Conversely, inhibitory phosphorylation by TAK1, MAP kinase-activated protein kinase 2 (MAPKAPK2), IKKα/β (at Ser320/Ser321, Ser336, and Ser25), and ULK1 (at Ser357) restrains RIPK1-mediated cell death,430 suggesting residue-specific modulation of these targets. Additionally, agents like Acetylshikonin, RETRA, and Erigeron breviscapus induce RIPK1/RIPK3/MLKL-dependent necroptosis, indicating their antitumor potential.431–433
MPO, NE, and PAD4 are key targets in ROS-mediated NETosis. While in early-stage cancers, antitumor N1 neutrophils dominate, suggesting that NETosis induction could enhance tumor cell killing,434 its inhibition in advanced stages can block protumor N2 neutrophil effects. ROS scavengers like Kaempferol suppress NETosis and reduce lung metastasis in a mouse breast tumor model by targeting the ROS-PAD4 pathway.435

Targeting cancer vulnerabilities via a combination approach
ROS-modulating combination therapies have become integral components of many adjuvant and neoadjuvant cancer treatment regimens, addressing the complexity and adaptability of ROS-mediated cancer cell biology. The anticancer combination therapeutic strategies leveraging ROS for anticancer effects are discussed in this section and in Table 5. The data concerning the antitumorigenic strategies have been collected from www.fda.gov, http://clinicaltrials.gov, http://go.drugbank.com, and different works of literature.

Combination therapeutic strategies targeting ROS elevation

TRX system in combination therapy
TRX inhibition often drives cancer cells to alternative metabolic pathways.436 For example, in MYC-high HGSOC, TRX inhibition shifts cells to glutamine metabolism, and combining Auranofin with CB-839 exacerbates ROS accumulation and induces metabolic stress and apoptosis.436 In NSCLC, the complementary roles of the TRX and GSH antioxidant systems are also evident, as simultaneous inhibition—via Auranofin treatment combined with GSR (GSH pathway) loss—synergistically elevates ROS levels, overwhelms cellular redox defenses, and triggers apoptosis, thereby sensitizing cancer cells to TRXR inhibition.437 Additionally, in Nrf2-activated NSCLC with Keap1 mutations, resistance to β-lapachone arises from enhanced antioxidant defenses. Targeting the TRX system and SOD1 has been found to disrupt redox balance synergistically, amplifying β-lapachone-induced ROS, DNA damage, and cell death.438
In glioblastoma, TRX inhibition by Auranofin upregulates GSTP1 as compensation, while Piperlongumine (PL) directly inhibits GSTP1, disabling the GSH system. Dual inhibition disrupts redox homeostasis, amplifies ROS, and lowers the IC50 to nanomolar levels, highlighting their potential for repurposing in glioblastoma therapy.439 A recent study suggested that simultaneous inhibition of the TRX system (Auranofin) and the proteasome (Bortezomib) synergistically induces paraptosis in breast cancer cells by promoting GSH depletion through the ATF4/CHAC1 axis, leading to proteotoxic stress and characteristic dilation of the ER and mitochondria.440 This combination selectively targets cancer cells at lower doses, offering an effective, less toxic strategy against apoptosis-resistant cancers. Inhibition of the TRX system has also been shown to effectively enhance chemosensitivity through ROS-dependent mechanisms. In HCC, Withaferin A synergizes Sorafenib to promote ROS-mediated ER stress, DNA damage, and apoptosis.441 Targeting both the TRX system and the DDR has demonstrated promising outcomes across several cancers. A study demonstrated that Cyst(e)inase, an enzyme depleting L-cysteine/cystine, when combined with Auranofin, amplified ROS levels and DSBs, inhibiting prostate cancer growth.442

Nanoparticle-driven combination therapies targeting the TRX system
TRXR inhibitors, including GNPs, Auranofin, CONPs, Curcumin derivatives, and others, are utilized as sensitizers in RT, PDT, CDT, and ultrasound (US) irradiation via nanoparticle-based delivery. This approach enhances treatment efficacy, overcomes resistance, and reduces required doses, minimizing side effects.
Gold nanoparticles (GNPs) act as nanoradiosensitizers, enhancing localized energy deposition and ROS generation, leading to DNA damage. A novel chemopiezocatalytic therapy using Piperlongumine (PL)-loaded, Auranofin (Au)-decorated, PEG-coated zinc oxide nanorods (PL-Au@P-ZnO NRs) leverages the piezoelectric properties of ZnO under US to generate ROS.443 Auranofin enhances ROS via Fenton-like activity, and PL increases specificity, inducing selective tumor cell death. In vivo, PL-Au@P-ZnO NRs with US suppress tumor growth without causing systemic toxicity.443
Motexafin Gadolinium (Xcytrin®), a Gadolinium (Gd3+)-coordinated Texaphyrin (Gd-Tex) radiosensitizer, has been investigated in clinical trials for glioblastoma multiforme and pancreatic cancer, but the outcomes remain unpublished. Limited efficacy has led to the development of Gd-Tex-lipid-based nanovesicles (Gd-NTs) for improved tumor accumulation and radiosensitization.444 Incorporating O2-dependent myoglobin (Mb) into Gd-NTs (MbO₂@Gd-NTs) alleviates tumor hypoxia, amplifies Gd-Tex-induced ROS production, enhances radiosensitization, and induces long-term antitumor immune memory.444
Curcumin is being evaluated as a radiosensitizer in prostate (NCT01917890) and cervical cancer (NCT05947513) but faces challenges due to its low bioavailability, short half-life, and HIFs, necessitating nanoparticle-based delivery systems. Preclinically, a biomimetic nanoplatform using mesoporous silica nanoparticles (MSNs) loaded with Chlorin e6 (Ce6) and Curcumin (Cur) and coated with cancer cell membranes (CMs) demonstrated enhanced tumor targeting and uptake in oral carcinoma.445 Cur inhibited TRXR1, disrupting ROS defenses and amplifying Ce6-mediated ROS under laser irradiation, significantly inhibiting tumor growth.

Targeting mitochondria in combination therapy
Combining mitochondria-targeting drugs with ROS-modulating therapies effectively overcomes cancer cell metabolic plasticity and enhances efficacy through synergistic pathways compared to monotherapy. For instance, Artemisinin generates ROS and induces apoptosis while weakly inhibiting GPX4 to trigger mild ferroptosis. Combining Artemisinin with GPX4 inhibitor, RSL3 synergistically enhances ferroptosis, highlighting the potential of this combination for effective cancer therapy.446 Targeting mitochondria and NQO1 may overcome drug resistance in NQO1-overexpressing BC. FRV-1 targets ETC-I, inducing ROS and mitochondrial dysfunction in NQO1-negative cells but is inactivated in NQO1-positive MCF7 cells. Combining FRV-1 with the NQO1 inhibitor Dicoumarol has been reported to restore sensitivity.447
Mitochondria-DDR targeting combinations show synergistic lethality in BRCA1-mutated cancers. Defective base-excision repair (BER) sensitizes cells to mito-ROS damage, making Elesclomol effective; moreover, combining Elesclomolwith PARP inhibitors, Rucaparib or Talazoparib amplifies ROS-induced DNA damage and enhances synthetic lethality.448 Targeting the ER and mitochondria have demonstrated anticancer efficacy in ovarian cancer. Dual-targeting nanoparticles (EMT-NPs), which synergistically induce ER stress and mitochondrial dysfunction by promoting Ca2+ efflux, ROS production, and apoptosis, have recently been developed.449 They also enhance imaging in xenograft models, offering a drug-free, precision cancer-related prognostic strategy. Finally, to overcome the limitations of IR-820, IR-820@NBs were developed, which target mitochondria under ultrasound to generate lethal ROS,450 while in combination with proteasome inhibitors, MG-132 further enhances SDT efficacy by inducing apoptosis and autophagy in HCC cells.
Targeting mitochondria has shown promise in enhancing chemosensitivity. The Elesclomol–Paclitaxel combination exploits Elesclomol’s ROS generation and mitochondrial apoptosis induction capacity to sensitize cancer cells to Paclitaxel. This combination has been evaluated in several clinical trials. A Phase II trial in patients with metastatic melanoma revealed that compared with Paclitaxel alone, Elesclomol-Paclitaxel doubled the median PFS, reduced DPR by 41.7%, and improved OS.451 However, a subsequent Phase III trial (NCT00522834) in chemotherapy-naive advanced melanoma patients failed to show efficacy overall, although patients with normal lactate dehydrogenase (LDH) levels responded better.452 These findings suggest the reliance of Elesclomol on OXPHOS, with the use of glycolysis inhibitors potentially enhancing its efficacy in high-LDH tumors. In support of this, a study in breast adenocarcinoma cells (MCF7 and MDA-MB-231) demonstrated greater cytotoxicity with Elesclomol and glycolysis inhibitors (2-DG or 3-BP) compared to single-agent therapies.453 Similarly, EP13, an OXPHOS inhibitor targeting complex I, increased ROS levels and synergized with glycolysis inhibitors (Oxamate and 2-DG) in aggressive breast cancers.454

Nanoparticle-driven combination therapies targeting mitochondria
Recent nanotechnology advances integrate mitochondria-targeted therapeutics with multimodal strategies to address hypoxia and enhance efficacy. A polymer micelle (CAT/CPT-TPP/PEG-Ce6, CTC) achieves triple-synergistic amplification of ROS via PDT and chemotherapy, inducing immunogenic cell death.455 Another nanoparticle system, AuCu-Ce6-TPP (ACCT), combines RT, RDT, and CDT to combat hypoxia by integrating Ce6 for sensitization, Curcumin (Cu) for OH• generation and hypoxia alleviation, Auranofin (Au) for glutathione depletion, and triphenylphosphine (TPP) for mitochondrial targeting, demonstrating potent anticancer effects in hypoxic 4T1 cells.456
The CFMB nanoplatform, loaded with Metformin (MET) and BAY-876 (BAY) on CuFe₂O₄ (CF), targets glycolysis and OXPHOS, depleting tumor energy via MET-mediated HK2 inhibition and mitochondrial disruption, and BAY-mediated GLUT1 inhibition. In high-GSH environments, CFMB releases Cu+/Fe2+, catalyzing H2O2 into OH• to enhance CDT, with NIR light boosting ROS production and enabling PTT for tumor suppression without toxicity.457 Similarly, the CPGMB nanoplatform, composed of polydopamine-coated CeO2-x nanorods co-loaded with MET and GOx, inhibits glycolysis via HK2 inhibition and glucose depletion. GOx-generated H2O2 is converted to O2 by Ce4+, alleviating hypoxia, while Ce3+ transforms H2O2 into OH• for CDT.458 Another study developed a mitochondriona-targeted bioreactor using TPP, GOx, CAT, and protoporphyrin IX (PpIX) that integrates glucose (starvation therapy), oxygen generation, and PDT to amplify ROS production, inducing mitochondrial destruction and apoptosis with strong in vivo antitumor efficacy.459
To address copper (Cu) ionophore limitations, a ROS-sensitive polymer (PHPM) was developed to coencapsulate Elesclomol (ES) and Cu, forming NP@ESCu nanoparticles.460 Upon ROS-triggered release in cancer cells, ES and Cu synergistically amplify ROS, inducing cuproptosis and stimulating immune responses. In a bladder cancer model, NP@ESCu reprogrammed the tumor microenvironment and, when combined with anti-PD-L1, enhanced therapeutic efficacy.460 Another bioreducible exosome with redox-cleavable diselenide linkers delivers mitochondria-targeting sonosensitizers (T-Ce6) and glycolysis inhibitors (FX11) to tumors.461 Under US irradiation, T-Ce6 induces ROS-mediated mitochondrial damage, whereas FX11 inhibits glycolysis, amplifying antitumor effects of SDT.

Glutaminase inhibition in combination therapy
Glutaminase inhibition in combination with other ROS modulation pathways has been shown to be effective in overcoming therapeutic resistance in various cancers. In metastatic RCC, CB-839 plus Cabozantinib achieved a 42% ORR and 100% DCR (NCT02071862),462 with Phase II trials ongoing (NCT0342821). For PIK3CA-mutant CRC, CB-839 combined with Capecitabine improved PFS, even in fluoropyrimidine-resistant patients (NCT02861300).463 In high-risk MDS, CB-839 and Azacitidine achieved a 70% marrow complete response in a Phase Ib/II study (NCT03047993).464 Overcoming Osimertinib resistance in NSCLC involves targeting glutamine metabolism, as resistant cells are glutamine dependent. Dual inhibition of ASCT2 and GLS1 has been shown to outperform GLS1 inhibition alone, effectively combating resistance in preclinical models.465 Additionally, high NQO1 expression in resistant cells supports redox homeostasis, contributing to resistance. Combining NQO1 targeting with glutamine metabolism inhibition offers a promising strategy to overcome resistance.466 Moreover, a nanobooster, C9SN, comprising the glutaminase inhibitor C968 and the photosensitizer Chlorin e6, enhanced PDT efficacy by depleting GSH and inducing oxidative stress.467 It triggered immunogenic cell death, CTL activation, and macrophage polarization, suppressing tumors and remodeling the tumor microenvironment. In anaplastic thyroid cancer (ATC), glutaminolysis inhibition alone induces growth arrest but not cell death due to compensatory ATF4-mediated one-carbon metabolism, which is amplified during progression from papillary to anaplastic thyroid cancer. Dual targeting of PHGDH, a key enzyme in one-carbon metabolism, and glutamine metabolism synergistically increases ROS, inducing growth arrest and sensitizing ATC cells to anticancer drugs.468

Glycolysis inhibition in combination therapy
Combining glycolysis inhibition with oxidative stress modulation enhances therapy by sensitizing cancer cells to ROS-mediated damage and addressing metabolic reprogramming, resistance, and nonselectivity. For instance, 2-DG synergizes with Etoposide to increase cytotoxicity, promote tumor-specific T-cell activation, and induce immunogenic cell death.469 Clinically, a Phase I study (NCT00247403) explored 2-DG as a radiosensitizer for intracranial neoplasms but was withdrawn due to discontinuation of drug manufacturing. This combination disrupts mitochondrial integrity, induces ROS-mediated apoptosis, and suppresses the expression of EMT markers, offering a promising metabolic reprogramming strategy for BC treatment. Glycolysis inhibition in CRC is limited by systemic toxicity and resistance via TRX1 upregulation, which enhances SLC1A5 expression, glutamine transport, and GSH synthesis to counteract cytotoxicity.470 TRX1 inhibitor PX-12 or SLC1A5 knockdown synergistically enhances the cytotoxic effects of 2-DG by disrupting redox balance in vitro and in vivo.470 Moreover, while single treatments showed minimal efficacy, coinhibition of glycolysis (2-DG), TRX (Auranofin), and GSH (BSO) synergistically suppressed tumor growth, metastasis, and tumor-initiating capacity in TNBC patient-derived xenografts by inducing ROS-mediated apoptosis in both epithelial and mesenchymal breast cancer stem cells.471

GSH depletion in combination therapy
GSH depletion enhances cancer therapy sensitivity. For example, the limited endogenous H2O2 and elevated GSH levels in tumor cells reduce the effectiveness of the iron-based nanocarrier 10-hydroxycamptothecin (HCPT) in inducing ferroptosis.472 To overcome this, a ferric phosphate nanotherapeutic system, FeP@HCPT-HA, was developed. FeP@HCPT-HA targets CD44-overexpressing tumor cells and degrades in the acidic tumor microenvironment to release HCPT, Fe2+, and Fe3+. The released Fe3+ depletes GSH, downregulates GPX4, and enhances lipid peroxidation, whereas HCPT induces apoptosis and supplies H2O2 for the Fe2+-mediated Fenton reaction. These synergistic mechanisms effectively induce ferroptosis and apoptosis, significantly inhibiting tumor growth. Additionally, to enhance PDT, a study designed a star-shaped polymer, PEG(-b-PCL-Ce6)-b-PBEMA, that boosts ROS generation, depletes GSH, and delivers chemotherapy. It combines PDT with Ce6 for ROS production, an H2O2-labile group for GSH depletion, and Doxorubicin delivery. This triple strategy significantly increased antitumor efficacy both in vitro and in vivo, suggesting a promising approach to amplify oxidative stress in cancer treatment.473 A Phase I trial (NCT00327288) evaluated Imexon-Docetaxel across various cancers to determine maximum tolerated dose, but the clinical outcomes are pending. Targeting GSH and glucose transporters also exploits the metabolic dependencies of cancer cells, suggesting a promising therapeutic approach. Cancers resist ferroptosis by upregulating SLC7A11 and increasing cysteine uptake and GSH synthesis. Glucose restriction causes cysteine accumulation and NADPH deficiency, triggering disulfidptosis. To exploit this vulnerability, FCSP@876 MOFs were developed, combining a ferroptosis-inducing MOF with the GLUT1 inhibitor BAY876. This nanoplatform induces two-pronged ROS attack: first, ferroptosis via lipid peroxidation, and second, disulfidptosis via glucose starvation-induced NADPH deficiency.474

GPX inhibition in combination therapy
GPX4 inhibition overcomes therapeutic resistance, as evidenced in BRAFV600E colorectal cancer. GPX4 upregulation in response to BRAFi ± EGFRi therapy blocks therapy-induced ferroptosis and drives resistance. Targeting GPX4 restores ferroptosis, enhances the efficacy of BRAFi ± EGFRi, and overcomes resistance in in vitro, in vivo, organoid, and patient-derived xenograft models.475 GPX-TRXR inhibition also overcomes cisplatin resistance in bladder cancer.476 Combining the TRXR1 inhibitors jolkinolide B or Auranofin, with GPX4 inhibitors offers a promising strategy for overcoming cisplatin resistance.476 Additionally, GPX inhibition may enhance PDT and PTT. To effectively eliminate residual tumor cells, a composite ZCND-Erastin/PAA:F127 hydrogel was developed, incorporating zinc-centered carbon nano-dodecahedrons (ZCND) and Erastin.477 ZCND PDT/PTT generates ROS, which reduce HSP70, destabilizing GPX4 and promoting ferroptosis, while Erastin also destabilizes GPX4 to enhance ferroptosis. In vitro and in vivo studies confirmed that this hydrogel effectively suppressed tumor recurrence after surgical resection. Coinhibition of GPX and Nrf2 also has a synergistic effect on ovarian cancer, which is particularly susceptible to ferroptosis due to its “iron addiction.” A study showed that combining Nrf2 inhibitors (e.g., ML385) with GPX4 inhibitors synergistically suppressed ovarian cancer by inducing ROS accumulation, lipid peroxidation, and caspase-3 activation, promoting ferroptosis and apoptosis.478 Coinhibition of GSTP and GSH enhances chemotherapy, as observed in a study on breast cancer, as Chlorambucil’s (CHL) efficacy is limited by GSH- and GSTP-pi-mediated resistance. A study developed a GSH-responsive prodrug (EA-SS-CHL) encapsulated in nanoparticles (ECPPs) that releases ethacrynic acid (EA) and CHL to deplete GSH, inhibit GSTP-pi, and amplify ROS, enhancing CHL’s cytotoxicity.479

Proteasome inhibition in combination therapy
Bortezomib has been shown to increase the efficacy of Actinomycin D in Wilms tumor cells by disrupting protein homeostasis and restoring apoptosis.480 Additionally, a Phase II trial (NCT01769209) explored the ability of Bortezomib to induce ROS in ALL blast cells and its potential to improve chemotherapy outcomes; the findings remain unpublished. Proteosome inhibition sensitizes CDT by elevating ROS levels and, with a macrophage membrane coating, ensures targeted tumor accumulation, as evidenced in a study on mCRC. This study used the MIL-88-MG132@M nanoplatform to overcome TME-induced CDT resistance.481 This Fe-MOF-based system, loaded with MG132, inhibited proteasome activity and NF-κB p65 phosphorylation, leading to the accumulation of proapoptotic proteins and the disruption of tumor homeostasis. Dual proteasome inhibition with Carfilzomib and Bortezomib has also demonstrated anticancer activity in melanoma while minimizing toxicity. A study showed that the low-dose Carfilzomib-Bortezomib combination amplifies antitumor effects through ROS-dependent apoptosis and ER stress, reducing Bortezomib-related toxicity.482
Co-inhibition of Glutaminase and proteasome might overcome Proteasome inhibitors (PIs) resistance, as evidenced in plasma cell myeloma (PCM). Combining the ASCT2 inhibitor V9302 with Carfilzomib synergistically elevated ROS levels, induced apoptosis, and activated a catastrophic UPR characterized by the upregulation of spliced Xbp1, ATF3, and CHOP, enhancing cytotoxicity in both PI-sensitive and resistant PCM cells.483 Coinhibition of the proteasome and Nrf2 can overcome proteasome inhibitor resistance by blocking the Nrf2-mediated antioxidant defense activated upon proteasome inhibition. For example, CuONPs inhibit proteasomes in vascular endothelial cells, inducing autophagy while stabilizing Nrf2 to increase antioxidant defenses. Conversely, Nrf2 inhibition blocks antioxidant gene activation, resulting in ROS accumulation, macromolecular damage, and cell death.484 This approach shows potential for use in ROS-mediated cancer therapy, warranting further exploration. Immune therapy-PI combination might enhance PI sensitivity, as evidenced in multiple myeloma (MM), where IL-33 enhances Bortezomib sensitivity by inducing excessive ROS.485 The IL-33-Bortezomib combination suppressed MM cell proliferation by boosting ROS levels, inhibiting NF-κB-p65 nuclear translocation, and downregulating the expression of stemness genes—effects reversed by the addition of ROS scavengers.

NQO1 inhibition in combination therapy
Coinhibition of GSTP1 and NQO1 shows antitumor potential in GBM, where EGFR mutations upregulate these enzymes to suppress ROS and drive tumor growth. The small-molecule inhibitor MNPC potently targets both NQO1 and GSTP1, disrupting redox homeostasis, elevating ROS, and inducing apoptosis in EGFRvIII-mutant GBM cells.486 Additionally, therapy-resistant CSCs, including those in TNBC, rely on DPP9-upregulated Nrf2-mediated antioxidant defenses (NQO1, SOD1) to suppress ROS and survive. Targeting these vulnerabilities, the NQO1-bioactivatable agent IB-DNQ generates ROS to kill CSCs, with its efficacy enhanced by concurrent SOD1 inhibition, which is elevated in TNBC.487
NQO1 inhibition may help overcome therapeutic resistance, as demonstrated in several preclinical and clinical trials. A study revealed that DPP9 promotes chemoresistance in liver cancer by binding to KEAP1, stabilizing Nrf2, and upregulating NQO1 expression to lower ROS and reduce chemotherapy efficacy.107 Inhibiting NQO1 with Dicoumarol restores ROS accumulation and enhances the effectiveness of chemotherapy in liver cancer. ARQ761 is under clinical evaluation with Gemcitabine and Nab-paclitaxel (NCT02514031) for treating NQO1-overexpressing tumors. A study enhanced CDT by using nanovesicles (NV-IONP-Lapa) loaded with iron oxide nanoparticles (IONPs) and β-Lapachone (Lapa). NQO1 catalyzed Lapa to generate H₂O₂, while IONPs released iron ions in the acidic tumor environment, converting H₂O₂ into cytotoxic OH• via the Fenton reaction. This system showed excellent tumor-targeting potential with minimal side effects, offering a promising approach for treating NQO1-overexpressing tumors.488
Targeting DDR and NQO1 may synergistically enhance antitumor efficacy while minimizing β-Lapachone-associated toxicity. Combining β-Lapachone with proliferating cell nuclear antigen (PCNA) inhibitor T2AA blocks DNA repair complex assembly, exacerbates ROS-induced DSBs, and enhances antitumor efficacy, as shown in Lewis lung carcinoma models.489 Similarly, combining the PARP inhibitor Rucaparib with β-Lapachone amplifies ROS generation and shifts the cell death mechanism from necrosis to caspase-dependent apoptosis, improving tumor-selective killing in NQO1-positive xenograft models.490 A clinical trial (NCT03575078) evaluating ARQ761 with Olaparib was initiated but withdrawn due to lack of patient enrollment.

Nrf2 inhibition in combination therapy
Nrf2 inhibition enhances antitumor therapy in various cancers by overcoming MDR linked to elevated ROS and Nrf2 levels. In a study on lung cancer, targeting the ROS/Nrf2 axis with Nrf2 siRNA or Tangeretin was shown to suppress Nrf2, overcome MDR (downregulating P-gp), and synergize with Paclitaxel and AZD9291 in A549/T xenografts.491 Similarly, to overcome hypoxia- and hyperthermia-induced phototherapy resistance, a MnO₂- and Ce6/PDA-PEG-FA-decorated silica nanonetwork loaded with Brusatol was engineered. MnO₂ generated oxygen to amplify Ce6-mediated PDT and PDA-driven PTT while targeting HSPs, and Brusatol inhibited the Nrf2‒GPX4 axis, inducing ferroptosis and phototherapeutic efficacy.492

Targeting DDR in combination therapy
Combining PARP inhibitors (PARPis) with ferroptosis inducers (FINs) may overcome PARPi resistance in BRCA-wild-type ovarian cancer. A study revealed that PARP inhibition induces ferroptosis via p53-mediated SLC7A11 downregulation, depleting glutathione and increasing ROS-driven lipid peroxidation in BRCA-mutant ovarian cancer. Ferroptosis suppression contributes to resistance in BRCA-wild-type cancers, while PARPi-FIN combinations synergistically enhance cytotoxicity in BRCA-proficient ovarian cancer models both in vitro and in vivo.493 Additionally, to address soft tissue sarcoma resistance to PDT, ZTN@COF@poloxamer nanocapsules were developed, integrating a Zr-based MOF, the photosensitizer Tetrakis (4-carbethoxyphenyl) porphyrin, and the PARP inhibitor Niraparib, where light-activated ROS generation induces apoptosis, enhancing PDT efficacy and blocking DNA repair.494
In addition, several other combination therapies targeting ROS promotion have been studied to overcome resistance to conventional anticancer agents. For example, recurrence in nasopharyngeal carcinoma often arises from radioresistance. A study showed that SOD2 knockdown amplifies radiation-induced superoxide production and lipid peroxidation, enhancing radiosensitivity via ferroptosis.495 However, DHODH inhibition in SOD2-depleted cells reduced oxidative stress, lipid peroxidation, and ferroptosis, improving survival and colony formation, suggesting that DHODH activity sustains radiation-induced ferroptosis. Therefore, caution is needed when combining DHODH inhibitors with radiotherapy, as they may counteract radiosensitization. NOX activator, Berbamine Hydrochloride has also been found to synergize with Cisplatin in lung carcinoma cells by inducing excessive ROS, impairing lysosomal acidification via NOX2 recruitment, triggering MAPK-mediated apoptosis, and increasing cell death.496 Photochemotherapy might be a promising strategy for precision nanotheranostics, as demonstrated by IMTD, a nanotheranostic coassembling ICG photosensitizer with Mannose (MAN)- Thioketal linker (TK)- Doxorubicin (DOX), for tumor-targeted, light/ROS-responsive photochemotherapy. IMTD prevents premature DOX leakage, enhances tumor accumulation via mannose receptor uptake, and enables spatiotemporal drug release through laser-triggered ROS generation and thioketal cleavage in vitro/in vivo.497
Tubeimoside-I, a heat shock protein inhibitor, has been reported to synergize with Oxaliplatin in CRC by increasing ROS levels, downregulating HSPD1, and upregulating ER stress and MAPK pathways, thereby enhancing the anticancer effects of Oxaliplatin.498 To overcome hypoxia and ROS limitations in tumor PDT, the cell membrane-targeting photosensitizer TBzT-CNQi was developed to generate ¹O2, OH•, and O2•⁻, inducing lipid peroxidation and enhancing ferroptosis.499 Moreover, combining TBzT-CNQi with ferroptosis suppressor protein 1 inhibitor (iFSP1) downregulated FSP1, reduced CoQ10, and amplified lipid peroxidation, triggering immunogenic ferroptosis. This combination activated immune responses, increased CD8+ T cells, and reduced tumor-associated macrophages, offering a promising strategy for photodynamic immunotherapy in hypoxic tumors. Immunotherapy faces challenges from poor tumor immunogenicity and immunosuppressive microenvironments. ICD inducers like Paclitaxel (PTX) stimulate immunity via DAMP release but are limited by poor targeting and immunosuppressive effects. Building on synergistic ICD inducers and IDO inhibitors, a study developed ROS-responsive PTX@PoxMTP NPs to codeliver PTX and 1-MT (IDO inhibitor).500 Tumor-specific ROS trigger PTX release, inducing ICD and amplifying ROS, whereas esterase-activated 1-MT disrupts the IDO1/kynurenine pathway to reduce immunosuppression. This synergy enhances dendritic cell maturation and T-cell infiltration, and suppresses Tregs and TAMs, resulting in robust tumor regression and metastasis control with minimal off-target effects.

Combination therapeutic strategies targeting ROS suppression
Several studies have investigated combination strategies to overcome resistance and improve tumor-specific responses by suppressing ROS levels. A novel regimen combining Sodium arsenite (NaAsO₂), Chloroquine (CQ), and Dichloroacetate (DCA) targets chemoresistance in oral squamous cell carcinoma (OSCC) by regulating ROS, autophagy, and glycolysis. NaAsO₂ induces cytotoxic ROS and inhibits glycolysis; DCA enhances this effect, while CQ blocks autophagy. This combination boosts antioxidant enzymes, restores redox balance, improves survival (90% vs. 35%), and reduces tumor progression in OSCC mouse models with minimal toxicity, offering a promising strategy against chemoresistance.501
The efficacy of immune checkpoint blockade (ICB) is often limited by MDSC-mediated immunosuppression. Tumor-derived GM-CSF activates STAT3, upregulating Fatty Acid Transport Protein 2 (FATP2) to drive lipid accumulation and ROS production, critical for MDSC function. A study demonstrated that FATP2 inhibition with Lipofermata disrupted lipid metabolism, reduced ROS, and blocked MDSC suppression, enhancing antitumor responses.502 This approach has synergized with anti-PD-L1 therapy by boosting CD8+ T-cell activity and restoring antitumor immunity, offering a metabolic reprogramming strategy to improve ICB efficacy.
In addition, many preclinical and clinical studies have been conducted and are ongoing to explore the antiproliferative, anti-invasive, antiangiogenic, and chemoresistance-modulating properties of antioxidants like Curcumin, Lycopene, and Vitamin C. A clinical trial (NCT03072992) revealed that intravenous Curcumin (CUC-1®) combined with Paclitaxel significantly improved the ORR (51%) compared to placebo (33%), with an even greater effect in patients completing treatment.503 Lycopene alone had no significant effect, but coadministration with Docetaxel showed low toxicity, favorable pharmacokinetics, and improved angiogenesis and IGF-1 signaling in metastatic prostate carcinoma (NCT0149519).504

Personalized cancer treatment utilizing the ROS signature
ROS-related biomarkers are essential for personalized medicine because of the observed heterogeneity in redox responses across different cancer types, stages, and individuals with the same cancer. Protein carbonyls, arising from oxidative cleavage, deamination, and Michael addition, along with advanced glycation end products (AGEs) like carboxymethyl lysine (CML), are among these markers.505,506 Additionally, lipid peroxidation products such as acrolein, malondialdehyde (MDA), 4-hydroxy-2-nonenal (4-HNE), and F2-isoprostanes (IsoPs) are also indicative of ROS-related damage.506,507 Nucleic acid oxidation products like 8-oxo-dG (8-oxo-2’-deoxyguanosine) are commonly used biomarker for oxidative stress and DNA damage caused by ROS.508 Elevated 8-oxo-dG indicates oxidative damage, and to date, many clinical studies have utilized plasma levels of 8-oxo-dG to monitor oxidative stress and DNA damage in cancer.
The thiol‒disulfide balance, representing the equilibrium between reduced thiols (-SH groups) and oxidized disulfides (-S‒S‒ bonds) in cells, serves as a redox biomarker.509 In a study of 62 cervical cancer patients and 61 healthy women (NCT04258553), Sezgin et al. reported a significant positive correlation between disulfide levels and cervical cancer stage, proposing thiol-disulfide balance as a potential early biomarker for this disease.510 Antioxidant enzymes are key biomarkers of oxidative stress and are extensively studied as ROS-related indicators in numerous cancer clinical trials.11 A study in ovarian cancer (NCT03470857) reported that elevated SOD activity might be linked to increased ROS. Another study (NCT01985113) reported that plasma TRX levels are significantly regulated in lung cancer patients (>7.1 U/mL), indicating its diagnostic potential.
Promoter methylation sites such as 5-methylcytosine (5-mC), RUNX3, DNMT1, and p16 are the key gene targets for hypermethylation induced by ROS, and their epigenetic alterations can serve as useful biomarkers for cancer diagnosis.511 Noncoding RNAs (ncRNAs), including microRNAs (miRNAs) and long noncoding RNAs (lncRNAs), play critical roles in regulating ROS-related pathogenesis, making them promising biomarkers and therapeutic targets in cancer.512 For example, miR-373 functions as a redox regulator and promotes tumorigenesis in various cancers, while miR-371 is linked to cancer metastasis, drug resistance, and recurrent germ cell cancers (NCT04435756).513 Additionally, miR-155, linked to oxidative stress and tumorigenesis via the Nrf2 pathway, is targeted by Cobomarsen, a Phase II trial of a miR-155 inhibitor (NCT03713320) for cutaneous T-cell lymphoma (CTL) and mycosis fungoides.514 HOTAIR, a lncRNA, regulates chromatin structure and gene expression, promoting tumor progression and metastasis in breast, colorectal, and lung cancer.515 Mitochondrial lncRNAs like ASncmtRNA, a redox sensor with oncogenic roles, are targeted by Andes-1537, an antisense oligonucleotide showing promise in Phase I trials (NCT02508441, NCT03985072) for solid tumors.516 Conversely, miR-16 suppresses tumor growth by regulating oxidative stress and promoting apoptosis. In a Phase I trial (NCT02369198), an EGFR-targeted miR-16 mimic showed promise in treating mesothelioma and NSCLC.517 Another tumor suppressor involved in redox regulation, miR-34a, has been extensively tested in animal models. MRX34, a miR-34a mimic, was developed to assess its safety, pharmacokinetics, and pharmacodynamics in a Phase I trial (NCT01829971). Besides, transcription factors like Nrf2 and p53 act as crucial ROS-related biomarkers in cancer diagnosis and treatment, balancing between protecting cells from oxidative stress and contributing to tumor growth under certain conditions.
Some ROS-related biomarkers that are clinically used to assess the redox state of the body or specific tissues and cells in cancer are listed in Table 6. The data concerning the oxidative biomarkers have been collected from www.fda.gov, http://clinicaltrials.gov, http://go.drugbank.com, and different works of literature.

Conclusion and perspective
ROS play a dual role in mammalian cells and are essential yet potentially toxic to both normal and cancerous tissues, making selective targeting through ROS-modulating therapies inherently challenging. The “Threshold concept for cancer therapy” suggests that tumor cells, with higher inherent ROS levels, can be selectively targeted through pro-oxidant therapy, exploiting their impaired redox homeostasis.518 However, the lack of comparable normal cell controls highlights the need for larger paired studies and in vivo investigations to validate its translational potential.
Developing antioxidant-based therapies poses significant challenges due to the antioxidant paradox. While antioxidants protect healthy cells by mitigating oxidative stress, cancer cells exploit elevated antioxidant levels for survival.14 Nonspecificity presents another great challenge in developing this therapeutic regime. For example, NOX inhibitors like DPI and VAS3947 affect pathways beyond NOX.519 DPI disrupts mitochondrial respiration, NOS, and other complexes, causing off-target superoxide bursts, while VAS3947 induces apoptosis via cysteine thiol alkylation, independent of NOX activity. These off-target effects can lead to misleading and unexpected outcomes in studies of NOX-related oxidative metabolism in cancer. Moreover, antioxidants like Ascorbate, Dimethyl Fumarate (DMF), and Lycopene can act as pro-oxidants or antioxidants depending on the cellular context. Given these challenges, antioxidant-based therapies may be better suited for cancer prevention, as cancer cells can exploit antioxidants for survival. Research should prioritize understanding their context-dependent pro-oxidant and antioxidant effects to improve selectivity. TME-based studies could increase treatment efficacy, addressing nonspecificity and resistance. For example, NOX4 promotes tumor growth and metastasis under hypoxic conditions, making selective NOX4 inhibitors potential treatments for hypoxia-inducible cancers.520 Furthermore, in the absence of specific agents, genetic approaches like gene knockdown offer precision therapy.
Targeting ROS promotion faces challenges due to metabolic plasticity and off-target toxicity. While polypharmacology enhances versatility, achieving tumor selectivity remains difficult. The efficacy of ROS-inducing agents depends on tumor-specific genetics and metabolism. For example, RSL3, a GPX4 inhibitor, induces ferroptosis in pancreatic cancer but varies in effectiveness based on the KRAS mutations and metabolic adaptations.521 Such observations underscore the necessity of developing context-specific therapies, which require the consideration of oncogenic drivers, metabolic dependencies, and redox regulation. ROS like H₂O₂ can amplify cancer cell death via bystander effects but may also harm normal cells, necessitating tissue- and systemic-level impact assessment.
Targeting ROS-dependent pathways for drug development is challenging because of the dual role, complex regulation, and context-dependent activity of ROS. Key considerations include several critical questions: Is the molecule’s regulation by ROS conserved, and what is its mode of regulation? How does it influence specific pathways, and what is the interplay between ROS and those pathways, including potential interference from other ROS-regulated molecules? Furthermore, does the target molecule affect a single pathway or multiple pathways? The tumor microenvironment (TME) adds another layer of complexity, as shown in cervical cancer cells: MEK inhibition (MEKi) modulates ROS differently across cell lines, with ERK reducing ROS in C33-A cells but increasing it in SiHa and CaSki cells.522 Combined ROS and ERK inhibition showed synergistic effects on CaSki and HeLa cells but not on C33-A or SiHa cells, underscoring the context-dependent ROS-ERK interplay. These complexities highlight the importance of precisely understanding ROS–target interactions and pathway dynamics to effectively leverage ROS-dependent mechanisms while minimizing unintended effects.
Combination therapeutic strategies in cancer focus on ROS induction rather than suppression, primarily owing to the complex and context-dependent role of ROS in malignancy, as exemplified by the “Antioxidant Paradox”. ROS-inducing therapies, combined with chemotherapy or radiotherapy, offer promise by amplifying oxidative stress selectively. However, challenges like bioavailability, off-target effects, TME heterogeneity, and metabolic plasticity persist.523 Nanoparticle delivery systems enable precise, TME-responsive drug release by exploiting high levels of ROS, altered pH, or elevated NQO1 levels in tumors. Hypoxia further complicates ROS-based therapies, but oxygen-independent strategies—such as Fenton-like metal catalysts (Fe/Cu), hypoxia-activated prodrugs (HAPs), SDT, and NQO1 bioactivatable drugs (e.g., β-Lapachone)—show particular promise.524 Furthermore, a major hurdle in ROS-targeted cancer therapy is the translation of mechanistic insights into clinically viable drugs. To address this, combining gene-editing tools (e.g., CRISPR or RNAi) with conventional therapies offers a promising solution—enabling precise disruption of redox-regulating genes (e.g., NOX4 and HIF1α) while leveraging the broad cytotoxicity of chemotherapy.
With respect to ROS-related biomarkers for personalized treatment, measuring specific ROS biomarkers like F2-isoprostanes or 8-oxo-dG, along with oxidative damage products, enhances sensitivity but faces challenges due to the transient and fluctuating nature of ROS.11 Stable isotopes and advanced techniques such as LC‒MS or UPLC‒MS/MS might improve detection accuracy, whereas real-time monitoring tools, such as fluorescence spectroscopy and electrochemical sensors, can capture dynamic ROS activity.525 Tissue-specific analyses and redox modulators provide deeper insights into organ- or cell-specific ROS behavior. Recent advancements, including deep tissue fluorescence imaging, metabolomics, and mass spectrometry imaging, have enabled in vivo investigations of redox mechanisms in tumors,526 but further research is needed to achieve full precision and efficacy. Ultimately, unraveling how redox modulation drives tumor initiation, progression, and resistance will guide the rational design of context specific, clinically relevant therapies that exploit tumor-specific redox vulnerabilities while minimizing harm to normal tissues.