The physiological activities of normal cells are carried out in a state of redox homeostasis ( Figure 3 ). Various factors cause redox imbalance, leading to oxidative stress. Excessive ROS in normal cells is related to tumorigenesis. Tumorigenesis is the formation of a new tumor from malignant transformed somatic cells, one of the first stages of which is DNA damage. Excessive ROS can directly induce oxidative DNA damage and cause genomic instabilities, leading to mutations. This effect is seen especially in proto-oncogenes and tumor suppressor genes, in which unrepaired mutations can trigger tumor initiation (36). Types of DNA damage include DNA double-strand breaks and the formation of oxidative DNA adducts such as mutagenic 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodG) and 8-hydroxy-2′-deoxyguanosine (8-OHdG). These compounds are widely considered markers of endogenous DNA oxidative damage as well as risk factors for cancer initiation (37). 8-OHdG causes a shift in histone methylation from an activated to a more inhibited status, which in turn results in aberrant methylation of tumor suppressor genes, leading to their inactivation and dysfunction in human hepatocarcinogenesis (38). In response to oxidative stress, Nrf2 prevents carcinogenesis through the rapid enzymatic modification and efflux of chemical carcinogens and the elimination of ROS or alleviation of oxidative damage through expression of its target genes (39). However, the intense damaging effects of oxidative stress in normal cells can cause redox imbalances. Therefore, disruption of redox homeostasis contributes to the initiation and development of cancer (40).

Cancer cells strive for a balanced redox environment to support their growth, proliferation, and progression. Disruption of redox homeostasis is detrimental to cancer cells. From the perspective of oxidation, the effect of ROS on tumors depends on the context—namely, the concentration, species, and site of action—of ROS. For instance, low to moderate concentrations of ROS have positive effects on cancer cells because they can act as signaling molecules involved in cancer cell growth, proliferation, and metabolism. As one of the lines of evidence, downregulation of SDHC expression resulted in decreased mitochondrial SDH activity and increased ROS levels in hepatocellular carcinoma (HCC) cells, thereby promoting HCC cell growth and metastasis in vitro and in vivo (41). ROS can induce tumor cell proliferation and enhance survival via numerous pathways. ROS oxidize and inactivate mitogen-activated protein kinase (MAPK) phosphatases, activating the MAPK pathway and promoting cancer cell growth, proliferation, and development (42). In addition, the ROS-activated MAPK pathway regulates the synthesis and activation of activator protein 1 (AP-1). AP-1 is a transcription factor that influences cell proliferation, invasion, cell cycle, and apoptosis and is activated by c-Jun N-terminal kinase (JNK), extracellular regulated protein kinases 1/2 (ERK1/2), and p38 (43). Activation of AP-1 is beneficial to cell proliferation owing to its enhancement of the expression of growth-stimulatory genes, such as cyclin D1, and simultaneous inhibition of the cell cycle inhibitor p21 (44). In addition to activating transcription factors, ROS regulate the expression of genes related to cell proliferation and progression via epigenetic modifications. Activation of protein kinase C (PKC) by phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA) can augment NOX2 activity following ROS production in MCF7 breast cancer cells. These ROS can induce histone H3 acetylation in the slug promoter region and thus upregulate slug expression. Slug is a main transcription factor involved in epithelial–mesenchymal transition (EMT), and its expression favors cell survival, proliferation, migration, and invasion (45). However, ROS perform these functions when present at the appropriate level and in the early stage of tumor formation. In other words, ROS in cancer cells can exert their cancer-promoting effects only under redox homeostasis. When the level of ROS is too high, ROS can lead to death and cell cycle arrest in cancer cells. Cunning tumor cells also have strategies to circumvent these effects and balance the excess ROS.

ROS levels in cancer cells are significantly higher than in normal counterparts partly due to mitochondrial dysfunction and hypermetabolism. Excessive ROS are destructive to cancer cells. To survive under conditions of excessive oxidative stress, cancer cells are constantly trying to enhance their antioxidant defenses to eliminate excess ROS. Malignant or metastatic cells utilize reductive stress to promote their viability. Hyperactivation of Nrf2 is a common mechanism. Sustained accumulation or activation of Nrf2 and subsequent upregulation of antioxidants results in an environment conducive to the survival and growth of malignant cells (46). KEAP1-mutant non-small-cell lung cancer (NSCLC) induces Nrf2 to activate the expression of thioredoxins and peroxiredoxins, which are critical for Nrf2-dependent growth and proliferation in KEAP1-mutant cell lines (47). By maintaining the redox homeostasis, Nrf2 restrains ROS-mediated activation of apoptosis-inducing JNK/p38 signaling cascades, thereby contributing to cancer cell proliferation and growth (48). In addition, cancer cells activate other signaling pathways and transcription factors to promote their proliferation and growth under conditions of redox homeostasis. FOXO proteins belong to the forkhead transcription factors subfamily, which participates in diverse cellular events and plays an important role in regulating cellular processes such as cell proliferation and apoptosis. To date, four FOXO members have been discovered in mammals: FOXO1, FOXO3a, FOXO4, and FOXO6. ROS regulate the activity and function of FOXO proteins, which are involved in the regulation of intracellular redox state. In response to excessive ROS, acetylated FOXO3a translocates from cytoplasm to nucleus, where it is deacetylated through interaction with nuclear sirtuin 1 (SIRT1) (49). An inadequate supply of reducing equivalent, such as NADH, can lead to the activation of SIRT1, a regulator of cellular redox activity (50). When deacetylated FOXO proteins become abundant in the nucleus, the expression of genes implicated in cell cycle and DNA repair is upregulated, as well as the expression of antioxidant enzymes, including SOD and CAT (51). FOXOs are essential for SIRT1-dependent cancer cell vitality. In addition, the overexpression of peroxisome proliferator-activated receptor δ (PPARδ) has been found to facilitate colorectal cancer cell growth and promotion by increasing the expression of vascular endothelial growth factor (52). Nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) is an important nuclear transcription factor that participates in cellular inflammatory and immune response and regulates DNA transcription, cytokine production, and cell differentiation and apoptosis. Activation of NF-ĸB can be accelerated in a redox-dependent manner via pyruvate dehydrogenase kinase 1 (PDK1). Through that mechanism, epithelial growth factor receptor (EGFR) exerts proproliferative effects, which was found to drive the appearance of pancreatic precancerous lesions in Kras mouse models (53). Therefore, maintaining the redox balance is very important for tumor cells. On the one hand, the reduction system maintains ROS at a certain concentration and inhibits the apoptotic effect of ROS on tumors; on the other hand, the level of ROS after redox regulation can allow ROS to exert a protumor effect.

Metastasis is actually an inefficient and difficult process for tumor cells. Metastatic cancer cells migrate through a variety of environments that are significantly different from that of their tumor origin, such as blood and lymph nodes, resulting in disruption of redox homeostasis through various mechanisms, a partial reason that the vast majority of tumor cells die during metastasis (54). To adapt to oxidative stress, the few surviving metastatic cells undergo reversible changes, thereby acquiring resistance to oxidative stress.

Cancer cells deactivate some metabolic pathways during metastasis, partially to preserve reducing equivalents to control oxidative stress. Some anabolic pathways consume reducing equivalents, weaken the antioxidant system, and aggravate oxidative stress. In turn, oxidative stress can also inhibit cell anabolism by consuming reducing equivalents. For example, reducing equivalents from NADPH are required for lipogenesis; decreasing acetyl-CoA carboxylase limits NADPH consumption by fatty acid synthesis and conserves more NADPH, promoting resistance to oxidative stress and other biological processes (55). Therefore, metastatic neoplastic cells downregulate or block these anabolic pathways to preserve more antioxidants for oxidative stress resistance to facilitate metastasis. Cells with metastatic potential and metastatic cells undergo epigenetic and transcriptional changes, upregulate antioxidant production and enhance the antioxidant capacity. The expression of Prx5 is upregulated in oral squamous cell carcinoma (OSCC) cells, thereby scavenging ROS and promoting OSCC cell invasion and growth (56). Melanoma cells in hypoxic areas of primary tumors demonstrate increased expression of the lactate transporter monocarboxylate transporter 1 (MCT1), which mitigates oxidative stress and improves the vitality of cancer cells in the blood (57). MCT1 upregulates lactate uptake, thereby decreasing the NAD+/NADH ratio, which increases pentose phosphate pathway (PPP) activity, a primary source of NADPH for alleviating oxidative stress (58). Oxidative stress is further increased when metastasizing cancer cells enter the bloodstream (59) because of the high levels of oxidants and fluid shear stress in the blood. Fluid shear stress can induce apoptosis in metastatic breast cancer cells by increasing mitochondrial generation of O 2 − . Circulating metastatic cancer cells demonstrate upregulated mitochondrial MnSOD expression, converting O 2 − to relatively stable H₂O₂, endowing cells with apoptosis resistance (60). In addition, some metastatic cancer cells enter the blood in clusters, which may promote their survival in the blood partially by decreasing their exposure to oxidants, thus attenuating the damaging effects of ROS (61). To improve the efficiency of metastasis, cancer cells may have an increased tendency to undergo lymphatic metastasis because the lymphatic system facilitates cancer cell migration and invasion. Vascular endothelial growth factor C (VEGFC) and multiple chemokines induce and accelerate metastatic cancer cells entry into lymphatic vessels, promoting metastatic spread. Overexpression of VEGFC in mouse lungs elevates the lymphatic vessel density and contributes to the spread of cancer cells from the lung to other regions (62). Moreover, evidence indicates that melanoma cells in the lymphatic circulation undergo less oxidative stress and form more metastatic lesions than metastatic cells in the blood circulation (63). Lymph contains fewer oxidants than blood and protects metastatic melanoma cells from damage induced by oxidative stress.

Although excessive ROS generally inhibits the survival of cancer cells during metastasis and invasion, ROS also promote tumor metastasis in some contexts. For instance, inhibition of TP53 induced glycolysis regulatory phosphatase (TIGAR), an enzyme that facilitates the influx of glucose into the PPP, and increases the levels of ROS in pancreatic ductal adenocarcinoma (PDAC) cells, resulting in enhanced invasive and metastatic ability of cancer cells (64). Moreover, H₂O₂ induces activation of AP-1 in a JNK-dependent manner, leading to upregulation of matrix metallopeptidase 7 (MMP-7) and increased metastasis in human colon cancer cells (65). These effects may depend on the concentration, type, and source of ROS. Moderate concentrations of ROS induce the metastasis of cancer cells. Different kinds of ROS have different roles on cancer cells. For instance, H₂O₂ generated via mitochondria may contribute to cancer cell metastasis (66), while lipid peroxides generated by membrane lipid peroxidation may inhibit cancer cell migration and invasion (63).

Cell death is an intricate process and is considered the endpoint of cell fate for both normal cells and cancer cells. Through extensive and thorough research, cell death modalities have been divided into two categories: accidental cell death and regulated cell death (RCD). The latter type of cell death, predominantly including apoptosis, necroptosis, pyroptosis, and ferroptosis, is characterized by controlled signaling pathways and definite effector mechanisms (67). Various stimuli or signals, including but not limited to oxidative stress, induce regulated death in cancer cells, which is also a mechanism and strategy for tumor treatment. However, cancer cells can avoid death through various methods, collectively called apoptosis resistance, which is a general hallmark of cancer (68). This review mainly describes the control of apoptosis, necroptosis, pyroptosis, and ferroptosis by cancer cells through redox regulation.

Autophagy is a catabolic phenomenon by which intracellular molecules and organelles are engulfed and encapsulated into double-membrane vesicles called autophagosomes, which interact with lysosomes for degradation of their components, thereby recycling intracellular components and maintaining cellular homeostasis (101). Autophagy not only happens at a basic level in cells but also can be triggered by diverse signaling pathways and cellular regulators and result in selective or nonselective degradation. Just as oxidative stress has a dual effect on tumor initiation and development, autophagy also has different effects on cancer based on the context, such as the stage of cancer development or the level of autophagy. Autophagy is believed to prevent cancer initiation by removing oncogenic molecules, toxic unfolded proteins, and dysfunctional organelles, but once the tumor forms, increased autophagic levels typically allow cancer cells to survive, grow, and migrate (102).

Autophagy is regulated by a rigorous and precise set of signaling pathways mainly involving autophagy-related genes (ATGs). The process of autophagy consists of five stages: initiation, phagophore nucleation, phagophore elongation, phagophore fusion, and content degradation. Autophagy is initiated upon the release of the ULK1 (ATG1) complex from mTOR inhibition. The ULK1 complex, consisting of ULK1, ULK2, FIP200, ATG101, and ATG13, leads to phagophore nucleation, which is subsequently induced by a class III PI3K complex that comprises VPS15, VPS34, ATG14, Beclin 1, UVRAG, and AMBRA1. ULK1 phosphorylates Beclin 1, which performs as a protein scaffold for the PI3K complex and accelerates the recruitment of other proteins in the phagophore. Two ubiquitin-like conjugation systems participate in phagophore elongation to form autophagosomes. The first conjugation system is the association of phosphatidylethanolamine (PE) with intracellular LC3-I to form LC3-II, the lipidated form of LC3; this conversion is promoted by ATG4B, ATG3, and ATG7 and results in the incorporation of LC3-II into the growing double membrane. The second system is mediated by ATG7 and ATG10 and results in the formation of the ATG5–ATG12 complex. Finally, the protein syntaxin 17 (STX17) induces the fusion of the autophagosome with the lysosome, leading to the formation of the autophagolysosome and degradation of the autophagosome contents (103).

Autophagy has a complicated relationship with redox regulation in cancer cells. Oxidative stress can induce autophagy in cancer cells. H₂O₂ can promote the accumulation of LC3-PE by inhibiting the delipidating activity of ATG4 in the cytoplasm and induce autophagy in cells during amino acid starvation (104). In MCF-7 breast cancer cells, oxidative stress activates the transcription factor forkhead box O3 (FOXO3), stimulating the transcription of LC3 and adenovirus E1B 19-kDa-interacting protein 3 (BNIP3), which are involved in autophagy (105). Moreover, oxidative stress can induce autophagy through AMP-activated protein kinase (AMPK). AMPK is sensitive to ROS and can be phosphorylated by the upstream kinase AMPK kinase (AMPKK) following H₂O₂ accumulation (106). Activated AMPK not only directly phosphorylates and activates ULK1 to induce autophagy, but also inhibits mTORC1 through phosphorylation of TSC2 and raptor during glucose deprivation, thereby reducing the inhibitory effect of mTORC1 on ULK1 (107).

From the perspective of the tumor-promoting effect of autophagy, autophagy can be considered a mechanism by which cancer cells resist oxidative stress. Cancer cells in hypoxic regions, characterized by high levels of ROS, have high autophagic flux to promote their survival, which may prevent an increase in ROS production (108). Hypoxia-inducible factor-1α (HIF-1α) is a major factor used by cells to adapt to hypoxia and oxidative stress. HIF-1α induces the transcription of the Bcl-2 and BNIP3 genes, whose expressed proteins compete with beclin-1 to combine with BCL2, thereby increasing cellular beclin-1 and promoting autophagy (109). Moreover, HIF-1α triggers selective mitophagy through the expression of BNIP3 to prolong cell survival during hypoxia (110). Accordingly, HIF-1α-induced autophagy and mitophagy reduce ROS levels in cancer cells during hypoxia to promote cell survival. Moreover, studies have shown that PDAC cell lines exhibit high autophagic flux and that downregulation of autophagy following ATG5 or ATG7 knockout decreases PDAC cell survival (111). This effect occurs partially because autophagy can maintain cellular redox homeostasis by preventing the accumulation of cytotoxic amounts of ROS, thereby allowing continuous tumor growth. Impairment of autophagy in these cells is associated with excessive oxidative stress, increased mitochondria impairment and decreased cell survival (112). Mitophagy, a subtype of selective autophagy, is a way to eliminate ROS and mitigate oxidative stress for cancer cells. In addition to the abovementioned BNIP3-induced mitophagy, membrane depolarization of dysfunctional mitochondria activates PTEN-induced putative kinase 1 (PINK1), which induces the activation of the E3 ligase parkin (PARK2) to ubiquitinate mitochondrial outer membrane protein substrates, providing a recognition signal for the autophagic machinery (113). This in turn leads to selective clearance of damaged mitochondria and decreases ROS production. Therefore, in cancer cells, ROS are eliminated through high autophagic flux to prevent the toxic effects of ROS, thereby promoting tumor progression.

Autophagy and Nrf2 are inextricably linked through p62. p62 is an adaptor protein that not only can induce autophagy, including that of itself and of other target proteins, but also can release Nrf2 from KEAP1 and exert its antioxidant effect by directly binding KEAP1 (114, 115). Meanwhile, the expression of p62 can also be regulated by Nrf2, indicating that there is a positive feedback mechanism in the p62–KEAP1–Nrf2 axis (116). As p62 is an autophagy substrate, inhibition of autophagy leads to accumulation of p62 and subsequent activation of Nrf2. Liver-specific deletion of autophagy gene ATG7 was shown to cause p62 accumulation, translocation to the nucleus of Nrf2, overexpression of Nrf2-target genes, and carcinogenesis, which was prevented by p62 deficiency (117). Moreover, deficiency of p62 and Nrf2 was found to greatly suppress the progression of oncogenic RAS-driven NSCLC in mouse models (27, 118). These lines of evidence show the oncogenic roles of p62 and Nrf2. Based on this context combined with the stage of cancer initiation, autophagy may play a role in suppressing tumorigenicity. Autophagy inhibits the occurrence of tumors by degrading p62 and then inhibiting Nrf2 activation. Therefore, as mentioned above, autophagy not only plays the role of scavenging ROS to protect cancer cells under the induction of oxidative stress, but also can inhibit tumorigenesis by degrading carcinogens.

Cancer cells are highly metabolically active and undergo metabolic reprogramming by which they can autonomously modify their metabolic efficiency and activity to satisfy the increased biosynthetic demand as well as mitigate oxidative stress, which is required for their proliferation and metastasis.

The discovery of the Warburg effect produced a torrent of cancer metabolism research and showed that cancer cells utilize glucose differently than normal cells. Cancer cells exhibit high uptake of glucose and preferential metabolism of glucose via glycolysis instead of mitochondrial OXPHOS, even in the presence of adequate oxygen; this preference is a common metabolic characteristic of cancer cells (119). How do cancer cells achieve this state of high glucose uptake and glycolysis? Dysregulated expression of genes encoding key glycolytic enzymes has been observed in cancer cells. The oncogene MYC has been shown to upregulate lactate dehydrogenase A (LDHA), an important glycolytic enzyme that is necessary to increase the glycolytic rate and tumorigenic potential of cancer cells (120). HIF1 induces an increase in glucose transporter 1 (GLUT1) activity, resulting in increased glucose intake (121). These mechanisms increase glycolysis through transcriptional upregulation of glucose transporters and glycolytic enzymes.

Why do cancer cells exhibit the Warburg effect, which uses more glucose but inefficiently produces ATP? This question has been studied in recent decades and urgently needs to be answered. From the perspective of cellular redox, the Warburg effect has been considered to favor cancer cells because a high level of glycolysis in cancer cells can decrease ROS generation and the induction of oxidative stress. Compared with untransformed cells, cancer cells have defects in mitochondrial metabolism (122). Glucose deprivation more readily induces oxidative stress in cancer cells, thereby increasing the susceptibility of cancer cells to glucose deprivation-induced cytotoxicity, which is detrimental to cancer cells (123). Although oxidative stress following glucose deprivation induces activation of Lyn kinase (Lyn) and JNK1 and increases the expression of c-Myc (124), cancer cells require increases in glucose metabolism to compensate for this defect (125). Cancer cells express high levels of pyruvate dehydrogenase kinases (PDKs) to inhibit PDH, which is responsible for catalyzing the conversion of pyruvate to acetyl-CoA, which flows into the tricarboxylic acid (TCA) cycle (126). By inhibiting PDH, detached cancer cells steer glucose away from the TCA cycle, thereby reducing the production of mitochondrial ROS, limiting the cytotoxic effects of oxidative stress, and inducing resistance to anoikis, a form of programmed cell death in which cells detach from the extracellular matrix (ECM). The inhibition of PDKs and activation of PDH can make some cancer cells sensitive to anoikis and decrease their metastatic efficiency due to ROS production (126). Moreover, upregulation of MnSOD sustains the Warburg effect through activation of the AMPK pathway to enhance the malignancy of tumor cells (127). Therefore, cancer cells can maintain redox homeostasis through the Warburg effect to acquire a survival advantage by restricting mitochondrial respiration and decreasing the generation of mitochondrial ROS.

On the other hand, cancer cells metabolize glucose through the PPP to support redox homeostasis. The PPP is the main source of NADPH, which can neutralize ROS and maintain intracellular levels of GSH (128). Glycolytic enzymes such as pyruvate kinase (PK) are involved in redirecting glycolytic flux through the PPP for detoxification of ROS. In cancer cells, ROS inhibit pyruvate kinase M2 (PKM2) to divert glucose flux into the PPP and thereby generate NADPH (129). Therefore, glucose metabolic reprogramming, including glycolysis and PPP activity, is involved in decreasing ROS levels in cancer cells.

Another metabolic pathway intimately related to redox homeostasis in cancer cells is glutaminolysis. Glutaminolysis is a metabolic process by which glutaminase (GLS) catalyzes the formation of ammonia and glutamate from glutamine. Then, glutamate is metabolized by glutamate dehydrogenase (GDH1) to alpha ketoglutarate (α-KG), a TCA cycle intermediate (130). Cancer cells exhibit high levels of glutamine uptake and glutaminolysis. The transcription factor c-Myc is exploited to increase glutamine input by increasing the expression of glutamine transporters including system N transporter 2 (SN2) and alanine-serine-cysteine transporter 2 (ASCT2) in cancer cells (131). In addition, c-Myc promotes glutaminolysis by upregulating the expression of GLS and GDH1 in cancer cells (132). Moreover, the glutaminolysis pathway is triggered by the oncoprotein K-Ras in PDAC (133).

Increased glutaminolysis is also an essential component of tumor metabolic reprogramming. Why are cancer cells so addicted to glutaminolysis? In addition to providing the glutaminolysis product α-KG to flow into the TCA cycle to meet the anabolic needs of cancer cells, glutaminolysis can also directly or indirectly participate in ROS detoxification and resistance to oxidative stress. On the one hand, glutaminolysis generates antioxidants that protect cancer cells against oxidative stress. The production of reduced GSH, composed of glutamate, cysteine, and glycine, therefore depends on glutaminolysis to provide glutamate and on the X c − antiporter to import the precursor of cysteine, cystine. Increased glutaminolysis is beneficial to the survival of sorafenib-resistant HCC cells via the NADPH-dependent antioxidative defense (134). On the other hand, glutaminolysis produces the intermediate α-KG for the TCA cycle to support a process called glutamine anaplerosis. This process can also generate other antioxidant molecules. For instance, the process of converting malate to oxaloacetate (OAA) in the TCA cycle can simultaneously reduce NADP+ to NADPH (135). In addition to producing antioxidants, metabolites of glutaminolysis in the TCA cycle can counteract oxidative stress through enzymes or transcription factors. For instance, α-KG can bind to calcium/calmodulin-dependent protein kinase kinase 2 (CAMKK2) and stimulate the activity of AMPK, a substrate of CAMKK2, thereby inhibiting the mTOR pathway, balancing redox status and ultimately reactivating anoikis resistance signaling in LKB1-deficient lung cancer (136). Fumarate generated in the TCA cycle can interact with and increase the activity of GPX1 to use GSH to eliminate excessive ROS (137). Furthermore, fumarate can modify the cysteine residues on KEAP1 to activate the antioxidant response of Nrf2 (138). In general, glutaminolysis is necessary for TCA anaplerosis, maintenance of the bioenergetic capacity, and regulation of redox status in diverse kinds of cancer.

The TME, composed of tumor cells, immune and inflammatory cells, fibroblasts, the ECM, tumor lymphatics, capillaries, and various cytokines and chemokines, is a complex integrated system. It takes part in many aspects of tumors and generates the tumor vasculature, which is closely implicated in metastatic progression. Synergistic interplay between cancer cells and cells within the TME, such as cancer-associated fibroblasts (CAFs) and immune cells, forms a tumor-promoting microenvironment. Redox regulation in cancer cells participates in this process, and the TME in turn maintains redox homeostasis cancer cells.

In the past, through studies based on the carcinogenic effects of ROS, dietary supplementation with antioxidants was believed to be able to prevent or treat cancer by reducing ROS levels. The effectiveness of antioxidants in cancer treatment has been proven in numerous animal models and in vitro studies. For example, in vitro, overexpression of MnSOD or CuZnSOD significantly decreased breast cancer cell growth (159). The combination of N-acetylcysteine (NAC) and vitamin C was found to prevent the onset of cancer in a model of MYC-dependent human B lymphoma (160). However, the results of many clinical trials launched to confirm whether dietary supplementation with antioxidants can reduce cancer incidence or inhibit tumor progression were disappointing. Large-scale clinical trials indicating that long-term vitamin E supplementation does not prevent cancer have been published (161). The SELECT study even demonstrated that dietary supplementation with vitamin E significantly increased the risk of prostate cancer among healthy men (162). Moreover, many studies on antioxidant dietary supplementation have demonstrated insufficient evidence that antioxidants can prevent cancer or suppress cancer progression (163). Thus, the hope of using antioxidant dietary supplementation to prevent cancer has waned. However, the use of some antioxidants in combination with traditional radio-chemotherapies appears promising in the treatment of cancer. A phase I clinical trial of ascorbic acid with gemcitabine in the control of metastatic and node-positive pancreatic cancer showed well-tolerated and some preliminary efficacy (164). Results from a phase I clinical trial of pharmacological ascorbate combined with radiation and temozolomide for newly diagnosed glioblastoma showed that the combination is safe and warrants further investigation (165). Antioxidants combined with radiotherapy may also reduce the incidence of side effects. Results of a phase IIb randomized double-blind trial of GC4419 (a superoxide dismutase mimetic) versus placebo showed that GC4419 significantly reduced severe oral mucositis induced by concurrent radiotherapy and cisplatin in head and neck cancer (166). Certainly, these clinical studies require further investigation to confirm the therapeutic effect of antioxidants combined with chemoradiotherapy.

Considering that cancer cells upregulate antioxidant production to eliminate excess ROS and maintain redox homeostasis, prooxidant therapy has been confirmed to treat tumors at the redox level, which means that this therapy works by increasing ROS levels and exacerbating oxidative stress in cancer cells to promote their death and inhibit tumor progression. Widely used chemotherapeutic agents, including procarbazine, paclitaxel, daunorubicin, doxorubicin, alkylating agents, cisplatin, carboplatin, topotecan, and irinotecan, can increase ROS in cancer cells and kill these cells by exacerbating oxidative stress (167). Some selective drugs that inhibit mitochondrial SDH, such as α-Tocopheryl succinate (α-TOS), have also been widely studied, which can induce cancer cell apoptosis by targeting SDH to generate ROS (168). On the other hand, in addition to production of excessive ROS, targeted suppression of the antioxidant system in cancer cells is another element of prooxidant therapy. Some small-molecule drugs with prooxidant effects have also been widely studied. Erastin induces ferroptosis by selectively inhibiting system X c − and decreasing GSH synthesis. In one study, folate-targeting exosomes were used to deliver erastin to triple-negative breast cancer cells. This folate-vectorized exosome-encapsulated erastin selectively targeted MDA-MB-231 cells and promoted ferroptosis through intracellular GSH depletion and ROS overproduction, inhibiting the proliferation and migration of these cells (169). Brusatol is an inhibitor of the Nrf2 pathway. A study showed that brusatol caused quick and transient exhaustion of the Nrf2-related protein in Hepa-1c1c7 mouse hepatic carcinoma cells and freshly isolated primary human hepatocytes, thereby sensitizing these cells to the cytotoxic effects of other chemotherapeutic drugs (170). Imexon is a prooxidant molecule that exhausts GSH, blocks GPX1 activity, and increases ROS levels; its activity against non-Hodgkin lymphoma was evaluated in a phase II trial (171).

Taken together, these findings indicate that regardless of the method or drug, it is important to destroy the redox homeostasis that cancer cells have painstakingly established. Specific and carefully adjusted interventions provide the opportunity to disrupt redox homeostasis in cancer cells. With the continuous increases in the understanding of cancers and the continuous advances in research methods and technologies, novel redox-based therapeutic approaches can improve the therapeutic effect on tumors and improve patient prognosis.

Unfortunately, cancer cells acquire resistance to chemotherapy, radiotherapy, or immunotherapy through redox regulation. Cancer cells upregulate antioxidant enzymes to develop resistance to antitumor therapies. The expression of GSH has been shown to be upregulated in NSCLC, leading to resistance to cisplatin therapy (172). Upregulation of Nrf2 is also a major cause of drug resistance in cancer cells, as Nrf2 is activated to increase antioxidant levels to detoxify ROS in cancer cells. Moreover, FOXO1 is implicated in drug resistance in cancer cells. FOXO1 activation via SIRT1-mediated deacetylation was observed to trigger overexpression of multidrug resistance protein 2 (MRP2) in tamoxifen-resistant MCF-7 breast cancer cells (173). Our group has explored the association between gemcitabine resistance and ferroptosis in PDAC; thus, here, we focus on discussing gemcitabine resistance, ferroptosis, and redox regulation in PDAC.

During the past two decades, gemcitabine has been the gold standard for systemic treatment of PDAC (174). The traditional anticancer mechanism of gemcitabine is to convert into gemcitabine diphosphate (dFdCDP) and gemcitabine triphosphate (dFdCTP) to block DNA extension and synthesis (175), but from a redox perspective, gemcitabine can generate ROS like other chemotherapeutic drugs. One of the mechanisms by which gemcitabine generates ROS is to induce NOX-derived ROS generation through an increase in the expression of p22^-phox via NF-κB activation (176). Thus, gemcitabine may also exert antitumor effects through the cytotoxic effect of excessive ROS, and this effect may also be related to ferroptosis (as discussed below).

However, the emergence of resistance to gemcitabine within weeks of treatment initiation has become a major obstacle in the treatment of PDAC with gemcitabine (177). On the one hand, this acquired resistance is related to the abundant fibrotic stroma of PDAC. The large amount of connective tissue surrounding the cancer cells may account for up to 90% of the total tumor volume and has been considered to form a physical obstacle to gemcitabine delivery (178). Among the various types of cell in the PDAC microenvironment, CAFs are the key fibrosis-generating cells. CAFs of PDAC are main secretory cells for soluble and insoluble ingredients that create the specific stroma that facilitates resistance to gemcitabine via physical barriers (177). Activated gemcitabine is entrapped within CAFs in the extracellular stroma, complicating its accessibility to cancer cells (179). On the other hand, PDAC cells possess intrinsic resistance to gemcitabine through the regulation of various molecular pathways. As a feedback mechanism, increased levels of gemcitabine-induced ROS activate Nrf2, which then triggers the transcription of cytoprotective antioxidant genes, especially genes encoding enzymes that catalyze GSH generation to eliminate the increased ROS (176). Furthermore, bioinformatic analysis showed that system X c − (SLC3A2 and SLC7A11) and GPX4, the major negative regulators of ferroptosis, were upregulated in gemcitabine-resistant pancreatic cancer cells (180). Thus, gemcitabine-mediated ROS can be inferred to further induce ferroptosis in cancer cells, implying that we can enhance the efficacy of gemcitabine or reverse gemcitabine resistance by inducing ferroptosis through targeting Nrf2, SLC3A2, SLC7A11, and GPX4 ( Figure 4 ). Knockdown of Nrf2 with Nrf2 siRNA showed that different PDAC cell lines were more sensitive to gemcitabine (176). Brusatol, an inhibitor of Nrf2, has been demonstrated to abrogate gemcitabine-induced Nrf2 activation, increase ROS accumulation, and potentiate gemcitabine-induced growth inhibition and cytotoxicity in pancreatic cancer cells (181). Another study demonstrated that plasma-treated water sensitizes pancreatic cancer cells to ferroptosis through targeted inhibition of Nrf2 and GPX4 (182). Therefore, it is worth further verifying whether combined treatment with Nrf2 inhibitors and ferroptosis inducers can reverse gemcitabine resistance and be a strategy for treating gemcitabine-resistant cells.