As a rapidly developing and emerging discipline, physical plasma has been widely applied in medicine, electronics, industry, and military and daily life, and its applications in the medical sector have attracted much attention (Von Woedtke et al., 2019). Gas plasma was first proposed by the British physicist Crockes in 1879 as the ‘fourth state of matter’ (Hoffmann et al., 2013) and was named ‘plasma’ in 1929 by American scientists Langmuir and Tonks (Haertel et al., 2014). Cold atmospheric plasma (CAP), partially ionized gases produced at room temperature and atmospheric pressure, is composed of reactive oxygen and nitrogen species (RONS), such as hydroxyl radicals (OH•), singleton oxygen (O), superoxide (O^2–), nitric oxide (NO•), hydrogen peroxide (H₂O₂), ozone (O₃), nitrogen dioxide (NO₂), nitrogen trioxide (NO₃), dinitrogen tetroxide (N₂O₄), and nitrite in the form of anions or protons (OONO^–, ONOOH) (Davalli et al., 2018; Xiang et al., 2018; Dai et al., 2020a). In addition to these active substances, CAP also subjects cells to reactive macromolecules, electromagnetic fields, and ions (Tornin et al., 2021). One unique feature of CAP is its multimodality, i.e., its effect relies on a carefully designed cocktail of multiple reactive species that is not dependent on any single component and exhibits differential therapeutic outcomes depending on the dosing and ejection parameter configuration. In addition to biological alterations, CAP also causes chemical and physical alterations to cells by introducing mild radiation and thermal and electromagnetic effects.

The use of CAP in the medical sector date back to the mid-20^th century (Laroussi, 2018) for wound healing, blood coagulation, ulcer prevention, decontamination, and surface purification (Hong et al., 2009; Gay-Mimbrera et al., 2016). The anticancer potential of CAP was first reported in 2007 when treating melanoma (Yan et al., 2017c; Zhou et al., 2020). Since then, there have been consecutive reports of success using CAP to selectively induce programmed death events such as apoptosis (Xiang et al., 2018), necrosis (Hoffmann et al., 2013), autophagy (Liu et al., 2022) and immunogenic cell death (ICD) (Van Loenhout et al., 2019); to halt tumour cell migration (Wang et al., 2021); to rewire malignant cell metabolism; and to arrest cancer stemness (Dai et al., 2022), the processes of which involve both genetic and epigenetic alterations (Dai et al., 2020a; Gangemi et al., 2022). The anticancer efficacy of CAP has been demonstrated in approximately 30 tumours, including breast cancer (Xiang et al., 2018; Dai et al., 2022), glioma (Yang et al., 2021), pancreatic cancer (Van Loenhout et al., 2019), bladder cancer (Wang et al., 2021), prostate cancer (Hua et al., 2021), and liver cancer (Liu et al., 2022). In addition to using various in vitro and in vivo cell or animal models, the first clinical trial of CAP as a cancer therapy was approved by the U.S. Food & Drug Administration (FDA) in July 2019 and completed in April 2021 (NCT04267575) (Mark, 2019; Zhou et al., 2020), wherein 17 out of 20 recruited stage IV malignant cancer patients remained alive by the end of the trial, demonstrating the safety and efficacy of CAP as a new technology for cancer treatment.

Given our incremental understanding of the roles and complexity of redox modulation in cancer and health (Dai et al., 2019; Zhou et al., 2020), increasing attention has been paid to plasma-associated precision medicine. However, without comprehensive elucidation of the underlying molecular mechanism, the successful clinical translation of CAP is limited. Thus, identifying the underlying research gaps by reviewing the current research status of this field will be of fundamental importance before plasma medicine can truly benefit human oncology and cancer patients.

In this paper, by reviewing tumour types and stages in which the efficacy of CAP has been explored, the varied experimental models used, and the cancer hallmarks with which CAP has been shown to be capable of interfering, we systematically classify current CAP-based oncological studies into carefully designed logical frameworks and identify possible research directions worthy of intensive investigation towards leveraging the unique strengths of CAP for effective cancer treatment.

CAP as an oncotherapy classified by tumour type

Ever since the first report of the anticancer effect of CAP in 2007, its selectivity against cancer cells has been reported in approximately 30 types of tumours, with the most intensively investigated types being melanoma (Yan et al., 2021) and breast cancer (Xiang et al., 2018), which represent 16.5% and 14.5% of all studies, respectively (Table 1). The investigated tumour types are primarily distributed in eight systems: abdominal organs (22.5%), reproductive organs (22%), skin (20%), brain (15%), bone (9%), blood (6%), head (4.5%), and soft tissue (1%) (Fig. 1). Abdominal cancers represent a large proportion of solid tumours, and examination of their efficacy has attracted much attention. Cancers originating from the reproductive system are diverse, most of which are well characterized, offering hints to systematically decipher the molecular mechanism driving the anticancer properties of CAP. Many studies have explored the therapeutic potential of CAP in skin cancer, as CAP has long been used for skin care (Busco et al., 2020). Furthermore lesions on the skin can be easily reached even if the electromagnetic effect from CAP can only penetrate glass/polystyrene barriers as thick as 7 mm (Yan et al., 2021). Tumours residing in the craniocerebral system are difficult to treat and require new therapeutic approaches, to which CAP may be particularly well suited.

Figure 1. Cancer types and proportions in which cold atmospheric plasma (CAP) is being explored as an oncotherapy. As of 14 December 2022, the field of plasma oncology encompasses eight broad categories of cancer: abdominal organs (45 studies), reproductive organs (44 studies), skin (40 studies), brain (30 studies), bone (18 studies), blood (12 studies), head (9 studies), and soft tissue (2 studies). Breast cancer (‘reproductive organs’), melanoma (‘skin’), glioma (‘brain’), and osteosarcoma (‘bone’) were the most commonly investigated (Table 1).

Among the abdominal tumours that have been investigated (i.e., pancreatic cancer (Van Loenhout et al., 2019), lung cancer (Wang et al., 2022b), colorectal cancer (Schneider et al., 2018a), colon cancer (Huang et al., 2022), bladder cancer (Wang et al., 2021), hepatocellular carcinoma (Wang et al., 2022), cholangiocarcinoma (Vaquero et al., 2020), and oesophageal cancer (Estarabadi et al., 2021)), pancreatic cancer, one of the most aggressive solid tumours, has been the best studied (10 out of 45, Table 1). For tumours originating from the reproductive system (i.e., breast cancer (Xiang et al., 2018; Dai et al., 2022), prostate cancer (Hua et al., 2021), cervical cancer (Kim et al., 2009), ovarian cancer (Koensgen et al., 2017)), breast cancer are the most commonly studied (29 out of 44, Table 1). Among the different breast cancer subtypes, triple-negative breast cancers (TNBC, without surface expression of characteristic receptors, i.e., oestrogen receptor, progesterone receptor, human epidermal growth factor receptor 2) exhibit the greatest cancer stemness and lack effective therapeutic options (Dai et al., 2016). CAP has been shown to be capable of selectively triggering apoptosis (Xiang et al., 2018) and inhibit migration (Wang et al., 2021) in TNBCs, effects that were both attributed to the ability of CAP to arrest cancer stemness (Dai et al., 2022). Tumours thus far examined in the skin system primarily include melanoma (Kim et al., 2009; Yan et al., 2021), cutaneous squamous cell carcinoma (Wang et al., 2019) and basal cell carcinoma (Yang et al., 2020b), with melanoma being the most frequently reported (33 out of 40, Table 1). For example, by combining CAP with gold nanoparticles bound to anti-FAK antibodies, five times more melanoma cells were killed than by using CAP alone (Kim et al., 2009). Among cancers in the craniocerebral system with demonstrated responsive to CAP treatment (i.e., glioma (Van Loenhout et al., 2021), vestibular schwannoma (Yoon et al., 2018), neuroblastoma (Walk et al., 2013), and retinoblastoma (Silva-Teixeira et al., 2021)), glioma is best studied (24 out of 30, Table 1). CAP has been synergized with auranofin to trigger cell death and immunogenic responses in glioblastoma and has been combined with gold nanoparticles to treat glioblastoma, where an overall 30% increase in cell death was achieved compared with CAP alone (Cheng et al., 2014a).

CAP as oncotherapy classified by experimental models

Most investigations on the safety and efficacy of CAP as a selective approach against cancer cells rely on in vitro studies and/or small in vivo animal models, with relatively little effort devoted to other models such as organoids, clinical case studies or trials. Currently, no large animals, such as primates or pigs, have been used to assess the onco-therapeutic efficacy and safety of CAP.

The in vitro antitumour efficacy of CAP has been examined using petri dishes or porous plates for almost 30 types of cancer cells, including brain cancer, skin cancer, breast cancer, colorectal cancer, lung cancer, cervical cancer, leukaemia, liver cancer and head and neck cancer (Keidar et al., 2011; Kaushik et al., 2013; Wang et al., 2013; Zhao et al., 2013; Bavelloni et al., 2015; Xiang et al., 2018). In vitro studies have been extensively used to evaluate the selectivity of CAP against tumour cells and to explore its mechanism of action. For instance, CAP-treated mouse neuroblastoma cell cultures exhibit reduced metabolic activity and increased apoptosis (Walk et al., 2013). Lung cancer cells and mouse melanoma cells exhibit dose-dependent cytopenia in response to CAP (Keidar et al., 2011). Melanoma exhibited significantly increased cellular mortality compared with normal keratinocytes after being treated with CAP, suggesting the role of CAP-imposed RONS in inducing cellular lipid peroxidation and DNA damage that pushes cancer cells past the apoptotic threshold sooner than their healthy peers (Zucker et al., 2012). In studies using 17 cell lines covering 7 tumour types (lung cancer, liver cancer, colon cancer, adenocarcinoma, melanoma, oral carcinoma, and uterine sarcoma) and 2 types of normal cells (i.e., adipose-derived stem cells and lung fibroblasts), CAP selectively induced the apoptosis of p53-deficient cancer cells without harming their healthy peers, suggesting the role of CAP in accelerating the genome instability of malignant cells (Ma et al., 2014). A more in-depth investigation reported the role of AQP3 in mediating cellular entry of some CAP-imposed reactive species and its competition with FOXO1 in sharing the same E3 ubiquitin ligase for K48-ubiquitination, where FOXO1 was responsible for fostering the stemness of cancer cells (Dai et al., 2022).

Organoids, self-organized three-dimensional (3D) tissue cultures derived from stem cells, can be crafted to replicate the complexity of an organ. Although commercially available organoid models are limited, they have been used in several studies as 3D experimental models to examine the anticancer efficacy of CAP. For instance, CAP demonstrated selective toxicity against bladder cancer cells in patient-derived preclinical organoid models (Gelbrich et al., 2023) and reduced the spheroid area and halted epithelial-mesenchymal transition (EMT) in patient-derived TNBC and bladder cancer organoids (Wang et al., 2021). Additionally, cytotoxicity, spheroid shrinkage and cell migration were monitored using 3D glioblastoma spheroid tumours after CAP treatment, where the impact of short-lived species was emphasized (Privat-Maldonado et al., 2018).

In vivo animal studies have largely relied on mouse models, including in situ models, subcutaneous xenograft models, peritoneal models and intracranial models (Chen et al., 2017). In vivo small animal assays have primarily been used to confirm the onco-therapeutic efficacy of CAP and the perturbed signalling networks identified using in vitro studies. After hypothesizing that CAP exhibits selective toxicity towards cancer cells, researchers used in vivo murine models of subcutaneous bladder tumours and melanoma to demonstrate that CAP can suppress tumour growth, enhance tumour shrinkage, and improve mouse survival (Keidar et al., 2011). Using a mouse neuroblastoma model, CAP was shown to be capable of ablating tumours in vivo, where the median survival duration was prolonged from 15 to 28 days (p<0.001) (Walk et al., 2013). In addition to arresting tumour growth, CAP reduced the migration of TNBCs in vivo, as demonstrated by enhanced expression of E-cadherin, a suppressive marker of metastasis (Zhou et al., 2020). Several studies have also explored the synergistic effects of CAP with therapeutics such as traditional chemotherapies and targeted agents using animal models. For instance, CAP exhibited enhanced selective cytotoxicity against melanoma when combined with dacarbazine (Alimohammadi et al., 2020) or doxorubicin (Pefani-Antimisiari et al., 2021) in vivo. A combinatorial strategy taking advantage of CAP and paclitaxel resulted in enhanced death and a higher frequency of DNA fragmentation than using paclitaxel alone in breast cancer (Mihai et al., 2022).

Several clinical case studies and clinical trials have been launched to prove the feasibility of CAP in cancer treatment. CAP has been used as a catheter-guided scalpel for surgery, with a significant postoperative curative effect reported for head and neck cancers. Specifically, 12 patients with head and neck cancers were recruited, with superficial partial tumour remission, healing of infected ulcerations, and decreased pain medication requirements after CAP treatment (Metelmann et al., 2015). Later, 6 patients with locally advanced squamous cell carcinoma of the oropharynx suffering from open infected ulcerations were recruited, and partial tumour remission in response to CAP was reported for 2 patients; the efficacy of CAP was attributed to CAP-induced ICD (Metelmann et al., 2018). Cold plasma scalpel CHCPS functioned as an adjunct to intraoperative radiotherapy in removing intraperitoneal sarcoma (Isbary et al., 2010). The Canady scalpel, a tool of CAP established for cancer treatment, was coupled with surgery to treat a 33-year-old with rare recurrent peritoneal sarcoma in 2019 (2020). In a clinical trial examining the efficacy of CAP against actinic keratotic lesions (NCT02759900), 19 out of 24 patients exhibited remission of their lesions, thereby decreasing the development of squamous cell carcinoma (the transformation rate of intraepidermal keratinocytic dysplasia to invasive squamous cell carcinoma is approximately 20% within 10-25 years) (Friedman et al., 2017; Wirtz et al., 2018). The first clinical trial using CAP as an oncotherapy was initiated in 2019 and completed in 2021 and involved 20 stage IV late-stage cancer patients (NCT04267575) (Mark, 2019; Zhou et al., 2020).

CAP as an oncotherapy classified by treatment approaches

CAP can be delivered to cells through both direct and indirect approaches. Direct treatment refers to exposing cells or tumours directly to plasmas ejected from various plasma sources. Indirect strategies typically treat objects with plasma-activated medium (PAM), such as plasma-activated normal saline (PSW) or plasma-activated PBS solution (PAPBS) (Chen et al., 2017) (Fig. 2).

Figure 2. Diagram summarizing differences between direct CAP injection and indirect PAM treatment.

For direct treatment, there are three main ways to source CAP (Bernhardt et al., 2019): dielectric barrier discharge (DBD), plasma jet, and plasma torch (Dai et al., 2018). The most striking difference between DBD and plasma jet or plasma torch lies in that the tissues or cells being treated act as electrodes in DBD but not in the other two approaches (Yan et al., 2017c). In addition, plasmas ejected from DBD and jet/torch differ in shape and range. Although plasma ejected by jet/torch is slender, has no fixed boundary, has a small range of action (Xu et al., 2021) and can be used for medical applications, DBD is advantageous in handling 3D objects and covers large surface areas (Corbella et al., 2021). It is noteworthy that the plasma jet exhibited greater efficacy in triggering the death of acute myeloid leukaemia cells than did DBD under the same conditions (Xu et al., 2021).

Among these strategies, DBD was the earliest approach and was proposed by Siemens and his colleagues in 1857 (Hoffmann et al., 2013). As a commercial example of DBD, the PlasmaDerm^® series (Fig. 3A) has been widely used to promote chronic wound healing and treat chronic inflammatory diseases (Bernhardt et al., 2019; Wandke et al., 2022). Koinuma et al. (1992) developed the first cold plasma jet in 1992. Plasma jets have been widely applied for medical purposes (Stoffels et al., 2002; Laroussi and Lu, 2005; Laroussi et al., 2006; Breathnach et al., 2018), with kINPen^® being the most widely used (Isbary et al., 2013b) (Fig. 3B). Microwave-driven plasma torches have also been launched for disease treatment, such as the use of microPlaSterβ^® (Fig. 3C) in wound healing (Arndt et al., 2013). Several miniature plasma jets, such as invivoPen and µCAP, have been established and used in cancer treatment (Chen et al., 2017; Zhou et al., 2020) (Fig. 3D, 3E). In addition, the activity of CAP can be delivered in liquid form, i.e., PAM, which can be easily stored, transported, and administered, largely expanding its application scenarios (Sersenova et al., 2021).

Figure 3. Representative cold atmospheric plasma (CAP) sources. Images and schematic presentations of (A) single electrode DBD, PlasmaDerm^® VU-2010 (CINOGY System GmbH plasma technology, Duderstadt, Germany); (B) plasma jet, kINPen^® (INP Greifswald/neoplas GmbH, Greifswald, Germany); (C) plasma torch, microPlaSterβ^® (Adtec Plasma Technology Co. Ltd., London, UK); and plasma sources for in vivo treatment, i.e., (D) invivoPen and (E) µCAP plasma generator. (A) was reproduced with permission from Bernhardt et al. (2019) under a Creative Commons CC-BY licence; (B) was reproduced with permission from Breathnach et al. (2018) under a Creative Commons CC-BY licence; (C) was reproduced with permission from Arndt et al. (2013a) under a Creative Commons CC-BY licence; (D) was reproduced with permission from Zhou et al. (2020) under a Creative Commons CC-BY licence; and (E) was reproduced with permission from Chen et al. (2017) under a Creative Commons CC-BY licence.

Direct ejection: In principle, direct CAP ejection exhibits anticancer efficacy comparable to that of indirect PAM treatment despite the observation that direct CAP treatment slightly raised the temperature of cells by approximately 1-2°C (Keidar et al., 2011).

Most CAP sources require direct cell or tissue exposure to deliver their active species and have limited penetration. Thus, direct CAP treatment has been largely used in in vitro assays, for skin cancers in vivo, and combined with surgery in clinical studies. By applying a convenient and portable ambient air-fed device, namely, aCAP, for local postoperative cancer treatment targeting residual tumour cells in the surgical cavity, CAP induced ICD and antitumour immunity by releasing tumour-associated antigens (TAAs) in situ. Immature dendritic cells (DCs) phagocytose TAAs and process them into peptides during the migration of TAAs to the tumour-draining lymph nodes. Here, mature DCs present antigenic peptides to T cells, producing cytotoxic T cells that inhibit tumour recurrence and suppress tumour growth. Treating residual tumour cells in the operative cavity with aCAP after surgical resection of 4T1 breast tumours significantly increased the survival of mice compared with the control group (Chen et al., 2021). Direct CAP treatment of melanoma reduced tumour burden and prolonged mouse survival by enhancing the cancer-immunity cycle (Lin et al., 2022). The Canady scalpel has been successful in treating patients with pancreatic cancer or rare recurrent peritoneal sarcoma (2020) and had been used in conjunction with surgery in a clinical trial involving 20 late-stage malignant cancer patients (NCT04267575) (Mark, 2019; Zhou et al., 2020).

Despite the demonstrated feasibility of CAP as an onco-therapy, direct CAP exposure can only cause cell death in the upper 3 to 5 cell layers (Chen et al., 2017). A novel plasma jet capable of ejecting CAP within tissues, namely, invivoPen, has been established and exhibits efficacy similar to PAM in treating TNBC in mice, suggesting the possibility of achieving minimally invasive cancer surgery (Zhou et al., 2020). Traditional CAP devices are too large for treating intracranial cancers. A micro-CAP device (μCAP) was invented to deliver plasma through an intracranial endoscopic tube to target glioblastoma in mice. The approach exhibited improved therapeutic outcomes compared with indirect approaches with a reduced gas flow rate and plasma jet size (Chen et al., 2017).

Indirect treatment: Unlike direct CAP treatment, which requires direct contact with cells/tissues and is restricted by limited depth penetration, the reactive species of CAP can be stored in the form of medium (i.e., PAM) and preserved for up to a week. Thus, PAM has been widely applied in CAP-based cancer investigations, especially in treating tumours residing in deep tissues, intraperitoneal tumours and metastatic tumours (Chen et al., 2017; Kaushik et al., 2018; Keidar et al., 2018). However, the efficacy of PAM may be reduced due to the nutrients it contains. Foetal bovine serum (FBS) in whole cell medium and pyruvate in Dulbecco’s modified Eagle’s medium (DMEM) can reduce the anticancer properties of PAM (Yan et al., 2016b). In addition, reactive species in PAM degrade over time due to the reaction of CAP substances with cell medium components (primarily cysteine and methionine), which can be significantly mitigated by the addition of 3-nitro-L-tyrosine or sequestration at –80°C (Yan et al., 2016a). The incubation time and dilution of PAM during treatment also affect the efficacy of PAM (Chen et al., 2017; Kaushik et al., 2018; Zhou et al., 2020). Many parameters may affect the efficacy of PAM, complicating its dosing. For instance, the use of larger wells in multiwell plates, shorter distances between the plasma source and the medium, and smaller media volumes have been reported to promote the anticancer properties of PAM, as the anticancer effect is determined by the amount of PAM acting on individual cells rather than the therapeutic dose of PAM as a whole (Yan et al., 2015; Jo et al., 2022a).

PAPBS and PAW are plasma-treated buffer solutions and differ from PAM in lacking the essential nutrients cells required to survive. However, PAPBS and PAW only affect cells for short periods of time (i.e., several hours), which has substantially restricted their application. For example, unlike PAM, which has demonstrated significant anticancer effects on a variety of cancer cells, PAW has no obvious cytotoxic effect on glioblastoma (Chen et al., 2017). This finding underlines the importance of establishing novel technologies capable of restoring the activity of PAM. Among various RONS, hydrogen peroxide has been proposed to play the critical role in the selectivity of CAP against cancer cells (Dai et al., 2022). Encouragingly, plasma-activated hydrogel (PAH), prepared by directly dissolving acrylamide, polymerizing acrylamide in PAW, and supplementing the mixture with crosslinking agents, initiators, and polymerization inhibitors, has been shown to prolong the lifespan of hydroxyl radicals for up to 140 days (Liu et al., 2019). However, the anticancer effect of CAP is complemented by other reactive species (Bauer et al., 2019; Dai et al., 2022), including singlet oxygen, which has been reported to play a leading role in selectively inducing cancer cell apoptosis (Sersenova et al., 2021). Effective strategies for preserving the activities of these species remain unknown.

CAP as an oncotherapy classified by molecular mechanisms

Both direct and indirect treatments can selectively induce the death of cancer cells by elevating cellular RONS levels. One prevailing explanation suggests that cancer cells exhibit higher cellular RONS levels and a more fragile antioxidant system than normal cells, rendering cancer cells more sensitive to redox stress and thus closer to the death threshold (Gorrini et al., 2013; Dai et al., 2022). Another theory attributes the selectivity of CAP against cancer cells to the lower levels of cholesterols and/or more aquaporins in cancer cell membranes, which are associated with increased RONS intracellular flow (Van der Paal et al., 2017).

With these conceptual explanations of the anticancer efficacy of CAP, it is necessary to identify the specific pathways and cancer hallmarks they target to promote the establishment of a CAP-based precision oncological system. CAP is known to interfere with all of the 10 well-recognized cancer hallmarks (Senga and Grose, 2021). These hallmarks can be groups into five traits: selective triggering of programmed death, halting tumour angiogenesis and metastasis, rewiring tumour metabolism, boosting immunity and suppressing tumour-promotive inflammation, and accelerating cancer cell genome instability (Fig. 4).

Figure 4. Effects of cold atmospheric plasma (CAP) on cancer hallmarks. The 10 cancer hallmarks (Senga and Grose, 2021) can be summarized into five traits, i.e., selective triggering of programmed death events in cancer cells, halting of tumour angiogenesis and metastasis, rewiring of tumour metabolism, boosting of immunity and suppression of tumour-promoting inflammation, and accelerated cancer cell genome instability. In response to reactive oxygen and nitrogen species (RONS) induced by CAP, the genome stability of cancer cells is compromised, which accelerates the mutation rate. As the genome integrity of cancer cells is associated with cancer stemness and drug resistance, targeting the integrity of transformed cells often leads to the killing of cancer stem cells and enhanced sensitivity of cancer cells to oncoagents. Cellular RONS levels continue to increase with cell chaos as a result of accelerated genome mutation that, once exceeding the death threshold as sensed by p53 (e.g.), may trigger various types of programmed cell death events. On the other hand, mitochondria, in response to enhanced cellular RONS, may trigger death programs by releasing, e.g., cytochrome C, which activates caspases. CAP can reshape cancer cell metabolism, including that of carbohydrates, nucleic acids, lipids, and amino acids, by redirecting pyruvate from increasing lactate production to enhancing efficient energy production, enabling β-oxidation and disturbing amino acid metabolism. In addition, CAP can promote tumour antigen release, stimulate immunogenic cell death (ICD), enhance tumour antigen presentation by antigen presentation cells, and enhance the sensitivity of tumour cells to immune therapies. Finally, CAP can arrest cancer angiogenesis and metastasis by halting the epithelial-mesenchymal transition (EMT).

Programmed cell death: CAP is capable of selectively inducing a wide panel of programmed cell death events in cancer cells, including apoptosis, cell cycle arrest, autophagic cell death, ferroptosis, pyroptosis, necroptosis, and ICD. For example, CAP induced apoptosis in human colon and lung cancer cells (Wang et al., 2022b), selectively induced G₀/G₁ cell cycle arrest in androgen receptor (AR)-independent prostate cancer cells (Hua et al., 2021), induced the expression of autophagy-related genes in melanoma cells (Alimohammadi et al., 2020), triggered human lung cancer cell ferroptosis by enhancing intracellular and lipid ROS levels (Jo et al., 2022a), induced necroptosis in vestibular schwannoma cells (Yoon et al., 2018), initiated ICD in colorectal cancers in vivo (Lin et al., 2018), and led to pyroptosis in a dose-dependent manner in lung, gastric, and hepatoma carcinoma cells (Yang et al., 2020a).

With an increased understanding of cell death programs, novel cell death events continue to be identified. Cuproptosis, a copper-triggered modality of mitochondrial cell death (Wang et al., 2022c), has recently attracted much attention. Given the significant overlap between cuproptosis and ferroptosis with respect to gene signatures and molecular mechanisms, it is highly possible that CAP also triggers cuproptosis, which warrants intensive investigation.

Tumour angiogenesis and metastasis: CAP plays an inhibitory role in tumour angiogenesis by selectively inhibiting VEGFA expression and altering the tumour microenvironment (TME) in vivo. For instance, CAP significantly reduced intratumoural vascular density in hepatocellular carcinoma and weakened microcirculation by causing vascular occlusion in the tumour-associated vascular system established using a hepatocellular carcinoma model (Kugler et al., 2022). It is worth noting that while inhibiting tumour angiogenesis, CAP can improve wound angiogenesis by activating angiogenesis-related healthy cells such as desquamated cells, fibroblasts and endothelial cells (Dai et al., 2020a).

CAP halts cancer cell migration and invasion by interfering with the epithelial mesenchymal transition (EMT). Specifically, although CAP has been shown to be capable of inducing epithelial cell apoptosis (Dezest et al., 2017), it preferentially inhibited the growth of mesenchymally shifted carcinoma cells without harming their epithelial peers in bladder and breast cancer organoid models (Wang et al., 2021). CAP, by synergizing with temozolomide, reduced glioblastoma cell migration by increasing the cell surface expression of αvβ3 and αvβ5 integrins (Gjika et al., 2020). Additionally, CAP inhibited the migration of myeloma cells by decreasing the secretion of MMP-2 and MMP-9, both of which are key metalloproteinases for degrading the extracellular matrix (Xu et al., 2016).

Tumour metabolism: CAP can rewire the metabolism of cancer cells by primarily perturbing carbohydrate metabolism, lipid metabolism, amino acid metabolism and energy metabolism.

Cancer cells are characterized by aerobic glycolysis, wherein glucose is converted into pyruvate by glucose transporters and glycolytic enzymes to produce sufficient biomass and adenosine triphosphate (ATP) for sustained cell proliferation. During this process, excess lactate is exported to the TME, which fosters an acidic environment favouring accelerated tumour growth. CAP can neutralize this acidic environment by inhibiting several major enzymes of glycolysis, including hexokinase (HK), phosphofructokinase-1 (PFK-1), pyruvate kinase (PK), and lactate dehydrogenase (LDH), to reduce the production of lactate (Guo et al., 2021). Once glucose enters cells, it is first catalysed by HK to generate glucose 6-phosphate, followed by catalysis by PFK-1, PK, LDH and other enzymes to produce energy. CAP blocks the glycolytic pathway by inhibiting hexokinase phosphorylation in blood cancer cells, which subsequently decreased cellular ATP levels and inhibited cancer cell proliferation (Kaushik et al., 2015). CAP downregulated glucose transporters (GLUTs) by suppressing PI3K/AKT signalling in glioblastoma cells (Makinoshima et al., 2014) and suppressed intracellular expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a classical glycolytic enzyme (Lackmann et al., 2013).

Another strategy adopted by cancer cells to enhance glycolysis and uncontrolled proliferation is the suppression of pentose phosphate signalling, wherein glucose is converted to nucleotides instead of lactate and energy. CAP can modify the metabolic profiles of U251SP glioblastoma cells by inhibiting the glycolytic pathway and enhancing pentose phosphate signalling (Kurake et al., 2019).

Lipids are key components of cells and the plasma membrane, which form an impermeable barrier and plays an important role in numerous cellular processes, such as cell growth, proliferation, differentiation, and signalling (Pan et al., 2021). CAP inhibited β-alanine metabolism in myeloma cells, leading to inhibition of fatty acids (FAs) synthesis by coenzyme A (CoA), as well as disrupted energy and material metabolism in cancer cells (Xu et al., 2018b).

Glutamate is an essential nutrient that sustains uncontrolled tumour cell growth and is produced through glutamine metabolism. CAP blocked the nutritional source of leukaemic cells by inhibiting the activity of glutaminase, thereby decreasing glutamate production (Xu et al., 2019b, 2021).

Tumour-promoting inflammation and immunity: DCs and macrophages in the innate immune system play essential roles in the balance between inflammation and anti-inflammation. The primarily role of DCs is to present TAAs to T cells to induce the production of tumour antigen-specific cytotoxic T cells (CTLs), which leads to activation of antitumour immunogenic programs. CAP synergizes with auranofin in activating DCs and initiating the antitumour immune response in glioblastoma cells (Van Loenhout et al., 2021). Additionally, CAP stimulated the differentiation of monocytes into macrophages (Kaushik et al., 2019) and primed macrophages towards the M1 state (exhibiting an antitumour phenotype) to promote an antitumour microenvironment by increasing the expression of cytokines associated with the M1 phenotype, including TNFα, TNFsf1 and Il1β. Cancer models used to explore these phenomena include human cholangiocarcinoma and glioma (Kaushik et al., 2019; Vaquero et al., 2020; Duchesne et al., 2021).

In adaptive immunity, CAP acted as a unique RONS delivery system to stimulate the immune system and promote an anticancer immune response (Bekeschus et al., 2017; Lin et al., 2019; Duchesne et al., 2021; Gangemi et al., 2022; Min et al., 2022). CAP reversed the downregulated immune response in glioblastoma multiforme (Almeida et al., 2019) and promoted the tumour recruitment of immune cells by releasing immune stimulation signals in the TME of pancreatic cancer cells (Lin et al., 2018; Almeida et al., 2019; Van Loenhout et al., 2019). CAP synergized with photodynamic therapy (PDT) in inducing greater apoptotic cell death in human papillomavirus (HPV)-positive cervical cancers than using PDT alone (Kim et al., 2009).

Genome stability: CAP sabotages the genome stability of cancer cells by causing DNA damage and interfering with the DNA repair system. Accelerated genome instability ultimately triggers death program due to rapid accumulation of damage that is intolerable by cells. CAP can induce DNA damage events in cancer cancers, especially double-strand breaks (DSBs), during the early stage of CAP treatment, which is marked by specific phosphorylation of threonine 139 on the H2A histone family member X (γ-H2AX). Recruitment of ataxia telangiectasia mutation (ATM) at the DSB threonine phosphorylation sites leads to subsequent p53 phosphorylation, DNA damage response complex formation, p21 activation and cell cycle arrest (Joerger and Fersht, 2008). In addition to cell cycle arrest, p53 phosphorylation activates proapoptotic factors such as Bax, Puma and Noxa, resulting in the release of cytochrome C into the cytoplasm and its binding with apoptotic protease factor-1 (Apaf-1). The subsequent formation of apoptotic bodies promotes cleavage-mediated activation of caspase-9 and caspase-3/7, and, ultimately, the initiation of apoptosis (Yan et al., 2017c; Chen et al., 2022). In addition to directly inducing DNA damage, CAP can impose endoplasmic reticulum (ER) redox stress, which indirectly endangers genome stability. During ER stress, Bax and Bak in the ER allow Ca²⁺ to be released from the ER into the cytoplasm, leading to m-Calpain and subsequent caspase-12 activation. The downstream sequential activation of caspase family members promotes the release of cytochrome C into the cytoplasm, apoptotic body formation, inhibition of anti-apoptotic factors and activation of pro-apoptotic proteins such as Bim, Bax, and Bak (Dai et al., 2021).

Given that CAP can promote genome instability, it is possible that CAP can arrest cancer stemness by altering the cellular antioxidant system. Although cancer stem cells (CSCs) have relatively lower redox levels relative to bulk tumour cells and noncancer cells, CSCs exhibit greater sensitivity to CAP treatment due to altered antioxidant homeostasis as a result of impeded genome stability (Dai et al., 2022). This theory has been examined in TNBC cells, which harbour a greater proportion of CSCs than other breast cancer subtypes. In TNBC cells, CAP treatment resulted in significant reductions in the levels of FOXO1 (antioxidant protein marker), ALDH1 (a marker of breast cancer stemness) and IL6 (a protein associated with cancer stemness) (Zhou et al., 2020; Sersenova et al., 2021; Dai et al., 2022).

Drug resistance is a common issue in oncology and is associated with the persistence of residual CSCs after surgery or drug therapy. Several experiments have demonstrated the ability of CAP to restore the drug sensitivity of cancer cells. For example, CAP rewired the resistance of 73% of breast cancer cells to paclitaxel by altering the expression of 49 genes, including Bcl2l13, Hoxa9, Dagla, and Ceacam1 (Park et al., 2019). CAP treatment also sensitized breast cancer cells to tamoxifen by altering the mRNA levels of Ma1 and Hoxc6 (Lee et al., 2017).

With our increasing understanding of the unique effects of CAP on cancer cells, CAP has emerged as an effective tool for preventing cancer metastasis and drug resistance by targeting cancer stemness (Dai et al., 2022). Since 2007, when the tumour selectivity of CAP was established, its effectiveness in selectively targeting cancer cells without harming their healthy peers has been demonstrated in diverse types of cancers (Xiang et al., 2018; Van Loenhout et al., 2019; Hua et al., 2021; Wang et al., 2021; Yang et al., 2021; Dai et al., 2022; Liu et al., 2022). These studies collectively suggest that CAP possess significant potential for clinical translation as a precision oncological tool, alone or by synergizing with other treatment modalities. For instance, CAP significantly inhibited the growth of head and neck cancers without obvious side effects in 2015 (Schuster et al., 2016), prolonged the lifespan of a 75-year-old patient with late-stage pancreatic cancer for 2 additional years in 2016, and altered the course of a 33-year-old patient with rare relapsed incurable peritoneal sarcoma in 2019 (2020). Despite these encouraging clinical efforts (Dai et al., 2018), several issues remain and require more focused attention.

Although a large panel of cancer types are responsive to CAP treatment, no systematic investigation have been conducted to identify features sensitizing cells to CAP treatment, nor have intensive efforts been made to establish a dosing system capable of arresting cancer cells based on their molecular characteristics. The lack of such a system may be due to different parameter configurations and the types of gases used to generating plasma in different studies, which results in differential combinations of reactive species and dosing intensities and renders these results difficult to compare. This system is of paramount importance before CAP can be accurately translated into clinical practice, as the outcome of CAP is dose-dependent (Dai et al., 2020b). A bridge between various CAP-initiated cell death programs and intrinsic cellular characteristics is required to achieve desirable therapeutic outcomes.

Current studies on CAP largely rely on in vitro assays. The establishment of organoids makes it possible to investigate the 3D behaviour of cells in response to CAP using an organ-mimicking system without resorting to in vivo animal experiments. In addition, organoid models make it possible to screen agents capable of improving the therapeutic efficacy of CAP, largely enhancing the ability to identify molecules that are likely to synergize with CAP. However, commercially available organoid tumour models remain limited, and relatively few studies have relied on organoids for CAP-associated investigations, both of which deserve intensified focus.

Direct plasma treatment, although superior to indirect approaches in cancer treatment, has difficulty reaching deep lesions. Although devices such as μCAP and invivoPen have been developed, investigations in this area are far from sufficient to resolve all issues. Indirect plasma treatment, although adding a great degree of flexibility to gas plasma administration, faces the obstacle of preserving the activity of PAM. I The finding that hydroxyl radicals, one of the leading reactive species in promoting cancer cell death (Dai et al., 2022), can be preserved in the form of PAHs for as long as 140 days is highly encouraging (Liu et al., 2019). However, the multimodal characteristic of CAP does not rely on any single component but requires the aid of other species, making the preservation of other reactive species equally important. This may represent a pivotal issue hampering the use of CAP in the medical sector.

Most studies on the anticancer mechanism of CAP focus on various types of death programs, migration and metastasis. With increasing evidence concerning the benefits of immunotherapy, greater attention has shifted to CAP-associated cancer immunity. However, relatively little has been devoted to deciphering the role of CAP in suppressing tumour angiogenesis, reshaping cancer metabolism, and accelerating cancer genome instability. The elucidation of these effects may deepen our understanding of the functionality of redox modulation in cancer initiation and progression. Additionally, CAP is known to trigger multiple death programs in cancer, the precise form of which varies with the components and dosing of CAP. As novel forms of programmed cell death and molecular mechanisms continue to emerge, it is interesting to examine whether CAP can potentiate these programs or possibly promote as of yet undescribed death events or pathways that are unique to CAP or redox modulation.

The authors declare no competing interest.