Many kinds of gases, such as argon, helium, nitrogen, oxygen, air, and mixed gases, are currently used in CAP (18). Different methods and equipment for producing CAP have been developed on the basis of different biomedical conditions (19). Two main types of methods are used directly and indirectly produce plasma, namely, dielectric barrier discharge (DBD) and plasma jets (20). Many laboratories have developed their own specific CAP production equipment because it is easy and inexpensive to make. However, the use of different equipment makes comparing the results of plasma therapy under the same conditions difficult. Therefore, the standardization of CAP devices is particularly important. Currently, the following types of medical products have been Conformite Europeenne (CE) certified: the DBD device PlasmaDerm^® VU-2010 (CINOGY Technologies GmbH), the atmospheric pressure plasma jet (APPJ) kINPen^® MED (INP Greifswald/neoplas tools GmbH), and the SteriPlas^® (Adtec Ltd., London, United Kingdom). (19)These devices have laid the foundation for the application of plasma therapy to solid tumors.

To standardize plasma sources, Mann et al. released DINSPEC91315 in 2014, which specified physical and technical standards (temperature, thermal power, radiation emission, ultraviolet radiation intensity, gas emission, and current flow) of plasma sources for biomedical applications, as well as basic criteria for characterizing biological plasma effects (namely, the effects on microorganisms in vitro) (21). DINSPEC91315 laid the foundation for biomedical research on plasma equipment. However, the standard is developed through existing standards and guidelines (DINENISO12100, DINEN60601-1, DINEN60601-1-6 and DINEN60601-2-57), and therefore, it does not define specific limits, as the evaluation of different parameters depends to a large extent on the expected effectiveness and application of the plasma equipment. Other characteristics of medical plasma equipment need to be added to the standards, including physical evaluation criteria, biological evaluation standards and chemical substance detection. The physical standards include gas temperature, gas composition, heat output, electromagnetic radiation and leakage current, while biological standards include the effects on microorganisms, eukaryotic cells and cancer cells. Furthermore, the composition of ROS and RNS and the effect of pH need to be evaluated through chemical detection (22). Collectively, specification of these parameters may improve the DINSPEC91315 and lead to a unified international standard for medical plasma equipment.

Recently, CAP has been applied to a variety of the clinical situations, especially in the field of dermatology, where it has been used for disinfection, the treatment of atopic eczema, ichthyosis/epidermal barrier defects, wound healing promotion, and scar and herpes zoster treatments (19, 23). CAP therapy of solid tumors is not yet being applied in the clinic [only a few studies have reported CAP cancer applications, and they describe CAP applied to the palliative treatment of end-stage head and neck cancer and grossly contaminated tumor ulcerations (24, 25)]; it is only in the laboratory research stage (26). In recent research, CAP treatment is mainly categorized into two types: direct treatment, that is, the use of a CAP-generating device to directly affect tumors, and indirect treatment, that is, the use of a CAP-generating device to treat a solvent that is subsequently applied to tumors.

These two treatment modalities have unique advantages and disadvantages. Because many ROS are short-lived and therefore only have a very short free diffusion path length, it is hard for them to reach their biological target (27). However, the applications for this treatment are relatively few, and recent research on direct CAP use has mainly been focused on subcutaneous tumor treatment (Figure 1A). We are committed to making plasma devices more portable for the treatment of other tumors. Recently, a team has developed a small and flexible plasma gun, and animal experiments have been carried out with it to prove CAP treatment efficacy against melanoma (28). We assume that a flexible microplasma gun combined with endoscopy (29) will be a promising device for treating tumors such as gastric, colorectal, lung or abdominal cavity metastases. As a consequence of indirect CAP treatment, the long-lived species hydrogen peroxide and nitrite, present in CAP-activated liquids, can interact in a complex set of reactions and generate the short-lived species singlet oxygen. Singlet oxygen then can inactivate catalase that is specifically present on the surface of tumor cells, but not on nonmalignant cells. As a consequence, the tumor cells start to generate secondary singlet oxygen through the interaction between hydrogen peroxide and peroxynitrite that are generated by the tumor cells themselves. This autoamplificatory mechanism leads to the inactivation of the protective membrane-associated catalase of tumor cells and thus allows specific ROS/RNS-mediated apoptosis-inducing signaling specifically in tumor cells (30–32). Indirect plasma therapy can be used in more types of tumors than direct plasma therapy because it can be injected into solid tumors or added to an abdominal lavage to treat disseminated peritoneal carcinomatosis (11) (Figure 1B).

With respect to the safety of plasma therapy, direct therapy has been safely used in dermatology applications, as mentioned above. In addition, although no clinical safety tests have been performed on indirect plasma therapy, many teams have verified the safety of plasma-activated water (PAW) in animal experiments. For example, a toxicity assessment with CD1 mice confirmed that long-term exposure to PAW had no negative effect on homeostasis, did not induce functional or histological changes in important organs, did not cause changes in hematological or biochemical blood parameters, and did not promote inflammation. (33) Xu et al. also proved indirect therapy safety by treating immunodeficient nude BALB/mice with an oral lavage of PAW daily for 2 weeks (34). In addition, Xu et al. proved that PAW can be safely injected into the bone marrow of New Zealand rabbits without inducing obvious toxicity (35).

In general, ROS refers are composed of oxygen and have active properties in an artificial or natural environment. Herein, we discuss ROS produced by plasma. Many valuable studies have proven that ROS produced by plasma include hydrogen peroxide (H₂O₂), hydroxyl radicals (OH), superoxide radicals (O2-), O₃, atomic oxygen and singlet δ oxygen (13, 39). RONS, especially short-lived RONS, have been proven to be the main active agents exerting biochemical and molecular biological effects through direct and indirect plasma therapy (10, 40, 41). ROS can cause oxidative modification in all biomolecules. Cells utilize several antioxidant mechanisms to prevent ROS/RNS-mediated damage. For non-malignant cells, ROS are generally derived from mitochondrial respiration and external damage (radiation, ionization, etc.). To prevent ROS-induced oxidative damage to cells, cells activate an antioxidant defense system, which is categorized into enzymatic (catalase, glutathione peroxidase and superoxide dismutase) and nonenzymatic (vitamin E, vitamin C, glutathione, β-carotene, urate, bilirubin flavonoid, etc.) forms (42). During oncogenesis, most malignant cells express NADPH oxidase (NOX) in their membrane and thus generate extracellular superoxide anions. These will dismutate and form hydrogen peroxide, which is used as a proliferation signal by the malignant cells. Extracellular superoxide anions as well as hydrogen peroxide of malignant cells also establish specific ROS/RNS-mediated apoptosis-inducing signaling, based on the HOCl and the NO/peroxynitrite signaling pathway. Therefore, efficient tumor progression requires the expression of membrane-associated catalase and superoxide dismutase (SOD), which interfere with ROS/RNS-mediated apoptosis-inducing signaling (31, 32, 43). Trachootham et al. suggested that apoptosis is induced when the intracellular oxidative pressure is exceeding the antioxidant potential of tumor cells. As the experimental findings by Trachootham et al. on which this concept is based had been established through the study of transformed cells, i.e., early stages of carcinogenesis that are not protected by membrane-associated catalase, their study neglected the protection of tumor cells toward exogenous ROS and RNS through membrane-associated catalase (44). Their concept is contrasted by the findings established with bona fide tumor cells, as discussed by Bauer et al. (30, 31, 43). Nitrite, a long-lived RNS, and hydrogen peroxide, a long-lived ROS interact in a synergistic interaction, based on the generation of peroxynitrite and resulting in the formation of singlet oxygen (30, 31, 43). Singlet oxygen produced in this way can inhibit tumor cell membrane-associated catalase and SOD, resulting in the flow of CAP-derived ROS/RNS into tumor cells. (31) RNS-derived nitric oxide (NO) regulates posttranslational modifications, S-nitration and genome-wide epigenetic modifications, which can either promote or inhibit tumorigenesis (32). The fate of cancer cells is RNS concentration-dependent. In tumor cells with high levels of NO, the extra RNS produced by CAP may overwhelm the system and shift the role of NO from promoting tumor formation to inhibiting tumor formation (38).

To determine the selectivity of CAP effects on tumor cells, a team compared tumor cells with homologous normal cells. The results presented in their review showed that, among 33 evaluated cell lines, 26 cell lines showed strong CAP selectivity, five showed weak CAP selectivity, and only two showed negative CAP selectivity (45). This study proved the selectivity of CAP therapy for most tumors. Therefore, herein, we review previous studies, compare the differences in plasma-derived RONS uptake and clearance between tumor cells and nonmalignant cells and explain the mechanism of the selective CAP effect on tumor cells.

With regard to the intake of RONS, many studies have clarified the selective mechanism of the CAP effect on tumor cells, including the cholesterol content and the aquaporin (AQP) expression in the tumor cell membrane.

On the one hand, studies have shown that ROS produced by CAP can oxidize lipids in the cell membrane and tend to induce formation of pores with a size of approximately 15 Å. These pores allow different free active substances to permeate the cell. For normal cells, the high cholesterol content in the plasma membrane prevents the entry of hydrogen peroxide and other active substances (46). However, the cholesterol level in the cancer cell membrane is lower than that of healthy cells; therefore, cancer cells are more vulnerable than normal cells to the influence of cell membrane lipid oxidation, which ultimately produces RONS that penetrate the cell, promote oxidative stress and induce apoptosis (47, 48) (Figure 2, 2).

On the other hand, cancer cells tend to express more AQPs in cytoplasmic membranes, enabling cancer cells to absorb H₂O₂ produced by plasma exposure faster than normal cells (45) (Figure 2, 3). AQP is a tetramer that forms a key channel in the cell membrane (49) that promotes the transmembrane transport of H₂O₂ and other small molecules, including carbon dioxide, NO, ammonia, urea and glycerol. (50, 51) Currently, three AQPs (AQP1, AQP3 and AQP8) are known to be involved in the transport of H₂O₂ in cells (52) Yusupov et al. performed computer simulation to determine the selective mechanism of AQP action in tumor cells, and the mechanism was found to be triggered by the high expression of AQPs in the cell membrane. (53).

The selection mechanism of the CAP-induced scavenging effect on RONS in the antioxidant defense system of tumor cells has not been thoroughly studied. However, studies showed that the interactions of singlet oxygen produced by CAP triggered tumor cell production of higher concentrations of secondary singlet oxygen, resulting in the profound inactivation of protective catalase and SOD in tumor cells. (31, 43) In addition (15), the mRNA expression levels of GSTA4, MSRB3, SOD1, SOD2, CAT2 and HMOX1 in cholangiocarcinoma cells and human primary hepatocytes were compared in this study. Compared with those in the two cholangiocarcinoma cell lines, the mRNA expression of several RONS scavenging enzymes in the hepatocytes was significantly increased. In addition, other studies showed that the activity of ROS scavenging enzymes, including glutathione peroxidase, catalase and superoxide dismutase, decreased significantly after CAP treatment (54). However, CAP treatment increased the amount of nonenzymes, such as glutathione, which was significantly oxidized, in the antioxidant-reduction system of lymphocytes (55), while CAP treatment decreased the ratio of glutathione to glutathione disulfide (GSH/GSSG) and NADPH to NADP⁺ in cancer cells (54, 56). These observations suggest a potential defense mechanism that protects normal cells from exhibiting excessive ROS levels specifically by producing ROS scavenging enzymes and nonenzymes (Figure 2, 4).

Thus far, we have mainly summarized the oxidative damage mechanism of RONS derived from CAP on cells and discussed selective CAP effects on tumor cells. The active components of CAP are not only derived from RONS; charged particles, ultraviolet radiation and electromagnetic fields also exert effects (57). However, except for the mechanism of oxidative damage induced by RONS in cells, these mechanisms have not been elucidated. Therefore, herein we explore the most direct effects of RONS derived from CAP on cells, the known signaling cascades triggered by RONS, and their effects on cells.

As mentioned earlier, RONS derived from CAP causes lipid peroxidation of the tumor cell membrane, which leads to an increase in active substances, triggering signaling cascades inside and outside the cell. Calcium ions (Ca²⁺) are important second messengers involved in intracellular and extracellular signaling cascades and play key roles in cell apoptosis and senescence. There is a close association between calcium signaling and ROS signaling (58). One study showed that intracellular Ca²⁺ levels increased after direct and indirect CAP treatment of melanoma cells (59). They observed that CAP-induced Ca²⁺ influx was mainly derived from intracellular Ca²⁺ stores in endoplasmic reticulum (ER) and mitochondria. The specific mechanism was found to be related to the ryanodine receptor in the ER and mitochondrial permeability transition pores (mPTPs), and it was similar to the mechanism leading to increased intracellular Ca²⁺ levels induced by RONS. Notably, RONS regulate the cellular Ca²⁺ channel by modifying the sulfhydryl group of a cysteine residue. Ca²⁺ is released from the ER through ryanodine receptor and inositol 1,4,5-trisphosphate receptor mediation (60, 61), leading to ER stress and mitochondrial Ca²⁺ overload, oxidative stress, membrane depolarization and mPTP opening (62) by releasing cytochromec (63, 64). RONS induce intramolecular disulfide bond formation mediated by adenine nucleotide translocase, enhancing the sensitivity of mPTP to Ca²⁺ (65). Therefore, increased cytoplasmic Ca²⁺ realized through CAP is likely caused by RONS (59). Some studies have reported mitochondrial oxidation and membrane potential depolarization after CAP treatment, resulting in mitochondria-mediated apoptosis (55, 66), confirming the accuracy of the aforementioned mechanism (Figure 2, 5)

Moreover, it is clear that ROS can activate many different mitogen-activated protein kinase (MAPK) cascades, including stress kinase, c-Jun N-terminal kinase (JNK) and stress-activated protein kinase (SAPK) pathways (67). Among these kinases, extracellular regulated protein kinases (ERK) 1/2, JNK and p38 kinase have been the most extensively studied (68). Research has shown that plasma-derived RONS can induce the death of tumor cells via apoptosis and other forms of death by activating the ERK1/2, JNK and p38 MAPK signaling pathways (69, 70) (Figure 2, 6).

In addition, the increase in intracellular ROS can induce DNA damage (Figure 2, 7). The DNA damage caused by ROS includes single-strand breaks (SSBs) or double-strand breaks (DSBs) in DNA, oxidative damage, and DNA–protein crosslinking (DPC) (71–73). Some studies have evaluated cancer cell DNA strand breaks caused by CAP treatment by detecting the phosphorylated form of histone H2AX: γ H2AX (70, 74–76). The H2AX histone is a variant of histone H2A in mammalian cells, which has a specific serine glutamine motif in the carboxy terminus (C-terminus). PI3 kinase phosphorylates H2AX to generate γH2AX, which is the key response to DNA fragmentation (77). Therefore, γH2AX is considered by many scholars to be an important marker of DNA damage and repair. DSB activates ATM [a DNA damage response kinase in the PI3K-like protein kinase family (78)], which phosphorylates histone H2AX and activates P53-mediated signaling pathways and TP73 (a tumor suppressor protein), leading to apoptosis. Increases in ATM, P53 and TP73 have also been observed in several tumor cells (oral squamous cell carcinoma, lung adenocarcinoma, melanoma, etc.) treated directly or indirectly by CAP (74, 79, 80). In addition, studies have proven that CAP causes DNA strand breaks in cancer cells in a time-dependent manner, and the specific time is related to different CAP-generating devices (75, 81–83). Despite the support offered by the aforementioned studies, some recent studies have suggested that DNA strand breaks may not be direct effects of ROS produced by CAP but results of the cell death induced by CAP (84, 85). For example, Guo et al. studied the effect of CAP on DPC in bacteria, yeast and human cervical cancer cells (71). Plasma-derived RONS, including OH, ¹O₂, O²⁻ and ONOO-, reacted with proteins to form protein-free radicals, such as protein-C·, protein-O· and protein-OO·. Moreover, these protein-free radicals were unstable and reacted quickly with other functional groups to form alcohols, carbonyl groups and hydroperoxides. These products were converted into protein-free radicals before they reacted with DNA and RNA to form covalent cross-links, resulting in cell death. Moreover, direct irradiation through CAP led to the chemical modification of DNA bases, specifically, the production of 8-oxoguanine (8-oxoG) (86). The 8-oxoG DNA damage is more easily induced than strand breaks. However, due to the upregulation of intracellular DNA repair enzymes, this DNA damage may not be the main cause of cell death (87). Nevertheless, Cheng et al. demonstrated that 8-oxoG oxidative modification of histone mRNA followed by S-phase degradation and chromatin instability in the early stage of the cell cycle was a key mechanism of CAP-induced cell death in breast cancer cell lines (85).

The mechanisms of ROS-induced cell death have been extensively studied by many teams, but little research on the corresponding RNS mechanisms have been reported. As mentioned in the previous section, excessive NO derived from RNS may inhibit tumor formation. NO derived from RNS can interact with superoxide anions derived from NOX, resulting in the formation of nitrous peroxide. After protonation of peroxynitrite through the action of proton pumps located in the cell membrane, the resultant peroxynitrous acid may decompose, resulting in the formation of nitrogen dioxide and apoptosis-inducing hydroxyl radicals. This may cause inactivation of tumor cells. One study showed that long-term exposure of cells to NO and/or peroxynitrite may deplete the intracellular glutathione pool and make mitochondria more vulnerable to the harmful effects of ROS (88). The interaction between RNS and components of the electron transport chain can increase the production of reactive oxygen species in mitochondria, thus triggering the mechanism leading to cell death.

However, a recent study showed that known RONS, such as singlet oxygen, hydrogen peroxide, NO and nitrite/nitrate, are not the main causes of plasma-activated medium (PAM)-mediated selective cell death in the Hep3B and Huh7hepatocellular carcinoma cell lines. The results showed that other factors (such as negatively charged particles) were involved in the physiological effects of CAP and PAM (89). Currently, the effects of other active components produced by CAP (such as free charged particles and electric fields) on tumor cells have not been studied in-depth (Figure 2, 8). How these components change the intracellular environment and how the cellular response and regulatory mechanisms respond to these changes remain unclear. (90) Nevertheless, many studies have reported that nanosecond-pulsed electric fields can lead to contraction and death of tumor cells in cancer tissue (91–93). In summary, the active components of CAP and their specific mechanisms need to be further studied.

In cancer, activated fibroblasts (called cancer-associated fibroblasts, CAFs) are thought to play key roles in malignant progression. CAFs promote tumor development by secreting matrix metalloproteinases (MMPs) to initiate ECM remodeling and secrete a variety of cytokines and growth factors (95). A study showed that plasma exerted dual effects on fibroblasts; that is, short-term treatment promoted cell activity and collagen production, and long-term treatment inhibited cell activity and collagen production (96). Higher concentrations of CAP-derived ROS and longer plasma treatment can induce the senescence and necrosis of fibroblasts (97, 98).

The two main types of immune cells in the tumor immune microenvironment (TIME) are tumor-associated macrophages and T lymphocytes. Other cells, including B lymphocytes, natural killer cells (NK cells) and dendritic cells (DCs), are also involved in the immune escape of tumor cells. Recently, many studies have proven the effect of CAP on tumor-associated immune cells, and these studies are described in the following section.

In the process of tumor formation, a tumor-specific vascular system emerges, and the newly formed blood vessels provide oxygen and nutrients to the tumor, thereby supporting tumor progression and providing a channel for tumor cell metastasis (99). Among that of the tumor cells involved in angiogenesis, the role of tumor endothelial cells (TECs) is very important. Vascular endothelial growth factors (VEGFs) secreted by tumor cells loosen tight junctions between endothelial cells (100), resulting in vascular infiltration, and excess TECs, to which other cells associated with normal vascular structures cannot readily adhere (99, 101). The intercellular gaps formed by these mechanisms allow macromolecules and tumor cells to pass freely into growing vessels (94). One study showed that endothelial cells exhibit higher levels of double-strand DNA damage than keratinocytes or fibroblasts after CAP irradiation (102). Moreover, higher levels of ROS derived from CAP can induce cell cycle arrest and decrease cell viability and thus reduce DNA damage in vascular endothelial cells (103). These findings support the supposition that CAP clearance of tumor endothelial cells may help control tumor progression.

Other cells such as adipose-derived stem cells (AMSCs) promote angiogenesis, antiapoptosis, proliferation and pluripotent differentiation, which are usually associated with tumor initiation and metastasis (104). However, to date, research on the effect of plasma on AMSCs in a tumor background is very limited.

In addition to the aforementioned cells, the TME is composed of collagen, elastin, fibronectin, glycoproteins and proteoglycans, collectively referred to as the ECM (14). Due to the overgrowth of tumor cells in the TME, the ECM in tumors usually shows excessive collagen deposition (105). CAP can inhibit tumor progression by oxidizing or destroying the structure of ECM (Figure 2, 1). Two studies on the clinical application of CAP in the treatment of head and neck cancer revealed that connective tissue proliferation induced by CAP after wound treatment indicated an increase in collagen deposition (106, 107). This outcome may be related to the aforementioned effect of CAP on fibroblasts. Another study showed that CAP inhibited the excessive synthesis of collagen in keloid fibroblasts (KFs) (108), and the activated fibroblasts may have similar characteristics with CAFs (109). An in vitro study on the effect of CAP on dentin surfaces and extracted type I collagen fibrils showed that CAP can destroy the structure of collagen (110). However, it remains to be seen whether CAP has any effect on the structure of tumor-related collagen fibers. Integrin is an adhesion molecule on the cell surface that plays a vital role in the adhesion, migration and invasion of tumor cells (111). The inhibitory effect of CAP on the adhesion, migration and invasion of tumor cells has been observed. In addition, after CAP treatment, the expression of integrin α2, integrin α4 and focal adhesion kinase (FAK) on the surface of melanoma cells has been found to be inhibited (112). Similarly, another study showed that CAP treatment inhibited the expression of integrin β1 and integrin αv in mouse skin cancer cells (113). These outcomes indicate that CAP may inhibit the adhesion, migration and invasion of tumor cells by inhibiting the expression of integrins. In addition, ROS can destroy or oxidize other ECM factors, such as hyaluronic acid (HA, an ECM proteoglycan) and fibronectin (114, 115). However, the effect of CAP on these proteins has not been extensively studied.

Plasma affects the communication between cells and between cells and the ECM. Some untreated cells outside the circumferential area of diffused ROS derived from CAP-treated cells can be damaged by plasma treatment. This outcome can be explained by the bystander effect and the abscopal effect (14). The bystander effect causes CAP-treated cells to transmit signals to untreated neighboring cells to induce biological changes. The effect of CAP on signaling pathways has been reported, and these damage signals can be transmitted through communication junctions [ion channels (59, 116), extracellular vesicles (117–119), gap junctions (120), and connexins (120–122)], occluding junctions [tight junctions (123), claudins (124), and occludins (124)] and anchoring junctions [adherens (122, 124–126) and desmosomes (127)]. As the current research on the effect of CAP on the malignant cell communication is not sufficient, the cellular communication references mentioned above include the study of malignant cells and non-malignant cells. In their review (14), Privat et al. described in detail the bystander effect mechanism in CAP-treated cells. Moreover, the abscopal effect enables CAP-treated cells to initiate responses in cells far from the treated area; that is, it is a systemic response involving the immune system (128). Plasma has been shown to inhibit the growth of melanoma tumors in distant untreated areas in mice, indicating that plasma therapy affects the immune response (129). The immune-related response induced by CAP therapy applied to tumors is described in the next section.