Mebendazole (MBZ) is an antiparasitic drug with 40 year history of safe use in humans and meets many of the ideal features for a repurposed drug to treat CNS tumors [47, 48]. Like all benzimidazoles used for treating helminth infestations, MBZ inhibits parasite microtubule polymerization via high-affinity binding to a colchicine-sensitive site on $\alpha$-tubulin. This property allows MBZ to dysregulate microtubule-mediated transport of secretory vesicles, glucose uptake, and ATP formation [49]. Beside its microtubule destabilizing properties, MBZ exerts pleiotropic effects on a number of cancer-related pathways at clinically achievable concentrations, such as angiogenesis, apoptosis, cell cycle progression, epithelial-to-mesenchymal transition, metastatic spread, antitumor immune response, and protein kinase inhibition [50, 51, 52]. Its role as a blocker of VEGFR2 activity was discovered by molecular fit computations on 3671 FDA approved drugs across 2335 human protein crystal structures. This work allowed investigators to map new drug-target interactions and predict novel uses for this drug [53]. Notably, experimental validation demonstrated that MBZ inhibits VEGFR2 kinase activity as well as angiogenesis at doses comparable to its well-established antihelmintic effects. Proteo-chemometric computational methods, coupled with kinase assays, have identified other previously unrecognized targets of MBZ including ABL1, MAPK1/ERK2, MAPK14/p38a, BCR-ABL, and BRAF [54, 55]. A recent study that used computational cell cycle profiling to prioritize FDA-approved drugs with repurposing potential demonstrated that MBZ was one of 36 agents with the strongest cytotoxic effect of a panel of 884 drugs that also contained known anticancer agents [56]. Other unexpected anticancer activities of MBZ were discovered by in silico analyses followed by in vitro and/or in vivo validation, such as induction of DNA damage [56] and differentiation [57]. For example, MBZ is the second of the top 20 differentiation-promoting candidates in a library of 1235 drugs that were screened through a computational approach that leveraged drug-induced gene expression changes in the differentiation state of leukemia cells [58].

MBZ has been reported to efficiently suppress proliferation in cell lines from a wide range of cancers, including leukemia, breast, and colorectal cancer [59, 60, 61], while sparing noncancerous immortalized cells at the same dose. The growing interest in MBZ as a repurposed medicine against CNS tumors stems from its physicochemical properties of possessing no charge, lipophilicity, and relatively small size as these are features that suggest favorable neuropenetrance after systemic administration [62]. However, of the known three polymorphs of MBZ (i.e., polymorphs A, B, and C) only polymorph C only reaches therapeutically effective concentrations in murine intracranial allograft models and brain tissues, achieving a brain to plasma ratio of 0.8 [63].

The IC50s for cell viability of MBZ in brain tumor models varies from 0.1–0.3 $\mu$M in GBM cell lines [35, 64], up to 0.5 $\mu$M in non-neuronal cell lines and cultured DMG cells dosed with MBZ [65, 66], and that these dose ranges fall in a clinically attainable range [67]. Of interest, multiple single-agent screens of 2706 drugs targeting 860 distinct cellular mechanisms uncovered that MBZ was among the 371-potency-selected agents in analysis of six DMG cell culture models [66]. Seminal preclinical studies convincingly provided evidence that MBZ prolongs survival in intracranial glioma [35] and MB [68]. Moreover, MBZ was associated with reduced vascularity in tumor, but not in normal tissues [69]. MBZ also suppresses the assembly of the primary cilium [65], a microtubule-based structure that plays a fundamental role in signal transduction of sonic hedgehog (Shh) and other cancer-related pathways, and is essential for initiation and maintenance of MB allografts in mice [70]. A recent study leveraging in silico analysis of the hub genes of GBM and a connectivity map platform for drug repurposing, found that multiple azole compounds had potential anti-GBM activity [71]. However, only the benzimidazoles flubendazole, fenbendazole, and MBZ were proven to efficiently suppress DNA synthesis, cell migration and invasion, G2/M cell cycle arrest in in vitro validation experiments [72]. An in vitro drug screening on a panel of 19 cell lines derived from childhood solid tumors, including CNS tumors, interrogated drug candidates in a collection of approximately 3800 approved and investigational compounds [40]. In this study, MBZ and other benzimidazoles were identified among the 736 compounds with robust bioactive profiling, and were found to be effective against the majority of the cell lines in this panel.

MBZ acts synergistically with ionizing radiation (IR) and several chemotherapeutic agents in different preclinical models of CNS tumors. In human (GBM14) and murine (GL261) glioma cells, MBZ enhances DNA-damaging effects of IR by targeting interphase microtubules at doses significantly lower than that needed for inducing mitotic arrest [73]. Mechanistically, this radiosensitizing effect is linked with inhibition of DNA damage repair (DDR) signaling through interference with the cytoplasmic-nuclear trafficking of DDR proteins. MBZ has also been used in vitro and in vivo as an adjuvant to frontline chemotherapeutics for the clinical management of CNS tumors showing enhanced growth-inhibitory effects. Of interest, improved DDR response is a mechanism of radioresistance shared by glioma stem cells and many types of glioma cells, including pediatric glioma, and that this highlights the importance of further preclinical and clinical evaluation of MBZ in combination with radiotherapy (RT) [74].

The combination of temozolomide (TMZ) with MBZ exerted a greater anti-proliferative effect than TMZ alone in patient-derived cultures and GBM cell lines, although this combination failed to significantly prolong survival in orthotopic glioma mouse models when compared to MBZ monotherapy [35]. In MB lines, MBZ overcomes resistance to the Shh inhibitor vismodegib mediated by mutant smoothened (SMO) receptor, and exerts an additive inhibition of canonical Shh signaling in combination with vismodegib [65]. Although these studies suggest potential combinatorial efficacy effects from MBZ, caution must be taken when including MBZ in treatment protocols since antagonistic effects of MBZ with different therapeutics have been reported by collateral activation of other pathways such as NF-kB [75] and the MEK-ERK [76] pathways.

The first studies investigating anti-tumor activity and safety of MBZ in the clinic were two case reports on individual patients with metastatic adrenocortical carcinoma [77] and metastatic colon cancer [78]. After failure of all treatment approaches, MBZ monotherapy was administered at the standard antihelmintic dose of 100 mg twice daily. This dosing showed a favorable safety profile accompanied by long-term disease control in both cases. Data on larger numbers of cancer patients confirmed an acceptable toxicity profile, and overall long-term safety of MBZ, even at much higher doses (up to 4 g/day), although antitumor efficacy was not consistently reported [79, 80]. All of the ten patients with refractory advanced gastrointestinal cancer studied developed rapid progressive disease while on single-agent MBZ at individualized doses targeted to a serum concentration of 300 ng/mL [79]. Another study used MBZ as an adjuvant therapy in association with FOLFOX4 and bevacizumab in 20 patients with metastatic colorectal carcinoma. Here the investigators reported significant improvement of overall response rate (ORR) and elevation of progression free survival (PFS) with respect to the control group of 20 patients treated with FOLFOX4 and bevacizumab [80]. However, the study failed to show significant variation in one year OS between the two arms.

Two safety trials were designed to determine the highest tolerable dose of MBZ in an adjuvant setting using frontline drugs for recurrent [81] or newly diagnosed adult high-grade glioma (HGG) patients (NCT01729260 [82]). Both trials reported good drug tolerability, although the small number of patients enrolled in these trials preclude drawing conclusions concerning MBZ efficacy in the treatment of glioma (Table 1, Ref. [82]). Pharmacokinetic analysis documented dose-related plasma levels of MBZ with large inter-patient variability, suggesting that administration of 75–100 mg/kg/day for future trials would provide safety, acceptable toxicity, and adequate plasma concentrations to reach those that result in antineoplastic activity in vitro [82]. However, a Phase II trial of MBZ (1600 mg thrice daily) with TMZ or lomustine in 88 patients with recurrent GBM failed to show clinical benefit with OS being no better than historical controls [83].

In the pediatric setting, a pilot trial (NCT02274987) has examined the feasibility of personalized therapeutic interventions in 15 children with newly diagnosed DMG using molecularly targeted therapy with up to four FDA approved drugs. These drugs were selected by an integrated approach leveraging genetic profiling with whole exome sequencing (WES), RNA sequencing, and in silico drug-gene matching [84]. Although the study was not designed to assess therapeutic efficacy, no survival benefit was reported in patients (n = 8) who followed treatment recommendation vs patients (n = 7) who underwent standard treatment. In this study, MBZ was selected to target PDGFRA copy number gain or PDGFRA overexpression and was administered in combination with other agents to 6 children.

A dose escalation clinical testing of oral MBZ as a monotherapy has recently been completed in pediatric patients with no longer responsive HGG or MB (NCT02644291). The study was aimed at analyzing adverse events and 2-year OS as a secondary endpoint. A phase I, II trial (NCT01837862) studying MBZ in combination with frontline treatments for gliomas is currently enrolling pediatric patients with low-grade glioma (LGG) and HGG, including diffuse intrinsic pontine glioma (DIPG), to find out the maximum tolerated dose (MTD) and 3-year clinical benefits. However, no records of completed clinical trials in children are available at this time.

Further clinical experimentation on MBZ should be formulated to address critical questions such as poor aqueous solubility and systemic availability of the agent, poor gastrointestinal absorption, and extensive first pass metabolism. Each of these variables result in large inter-patient pharmacokinetic variability and hamper achieving adequate therapeutic concentrations in vivo. These shortcomings may be overcome by nanocarrier-based formulation, prodrug design, and solid dispersion approach to increase bioavailability and reduce pill burden [46].

Valproic acid (VPA) is a short branched-chain fatty acid with a long-established use in humans as a mainstream antiepileptic medicine. Significant effort on understanding the embryotoxic effects of in-uterus exposure to VPA [94] led to the discovery of its anti-proliferative and differentiation-inducing capabilities in transformed cells of neural origin both in culture [95] and in vivo [96] in the 1990’s, fueling investigation of VPA as an alternate cancer-fighting therapeutic. Since these pioneering studies, evidence has gradually accumulated regarding the antineoplastic activity of VPA in several model systems, and this topic has been the focus of numerous reviews [97, 98, 99]. The mechanisms that underpin the antiepileptic and antitumoral activity of VPA are not clearly defined, but are very likely distinct in nature [98, 100]. VPA anticonvulsant action is clearly related to the blocking of voltage-dependent sodium channels and to increased brain concentrations of the inhibitory neurotransmitter gamma-aminobutyric acid (GABA) through VPA-induced modulation of enzymes involved in GABA metabolism. Effective VPA concentrations for these actions are 0.5–1 mM in in vitro competitive assays and in cultured hippocampal neurons [101].

VPA is also a relatively weak inhibitor of class I histone deacetylases (HDAC). IC50s for VPA in in vitro HDAC assays are within, or close to, its therapeutic range (0.35–0.7 mM in serum) and result in dose-dependent histone hyperacetylation [100, 102]. As an epigenetic modeler of chromatin structure, VPA is involved in oncogenic silencing and recruitment of tumor suppressive transcription factors [103] that ultimately trigger apoptosis, autophagy, or differentiation, with concurrent inhibition of cell-cycle progression, invasion, and tumor angiogenesis [97, 98, 99]. Mechanistically, VPA modulates several distinct cancer-related pathways, including MAPK, $\beta$-catenin, and GSK-3 signaling pathways.

Several seminal reviews in the field have thoroughly examined the cancer-related effects of VPA in various glioma models [99, 103, 104]. Generally, IC50s of VPA at 72 h are far above the upper limit of clinically therapeutic concentration (1 mM) when tested in the majority of glioma cell lines [105, 106, 107, 108]. However, 1 mM VPA is able to induce a time-dependent, oxidative stress-related, autophagy as well as activation of MAPK signaling, H3 and H4 histone acetylation, upregulation of the cell-cycle inhibitor p21, differentiation, and inhibition of VEGF secretion in glioma cells and HUVEC tube formation [105, 109, 110]. Higher doses of VPA (5–20 mM) are needed to reduce stress-related molecules such as paraoxonase 2 (PON2), an endogenous free-radical scavenging enzyme, that increases the apoptosis inducing reactive oxygen species (ROS) [107]. This in vitro VPA activity translates to in vivo glioma models where VPA inhibits tumor growth and angiogenesis while inducing differentiation and apoptosis [107, 110, 111].

Few studies are available concerning the activity of VPA in preclinical models of pediatric CNS tumors. In medulloblastoma cell lines, VPA at clinically safe concentrations (0.6 and 1 mM) necessitate prolonged exposure (14 days and longer) to induce potent growth inhibition, apoptosis, and differentiation, although detectable H3 and H4 hyperacetylation are noted as earlier as 3 days after drug addition [112, 113]. In both heterotopic and orthotopic MB xenografts, VPA treatment significantly suppresses tumor growth and angiogenesis, while enhancing histone hyperacetylation, apoptosis, and differentiation.

In search for more effective treatment strategies against DMG, VPA was included in a chemical screen of 83 of the most promising therapeutic agents identified in preclinical models [114]. However, it was not effective when tested in a panel of seven patient-derived DMG lines, possibly because the maximum dose used was 10 $\mu$M and the readout used was cell viability at 72 h. Indeed, in a subsequent study that focused on dose-dependent effects of VPA in H3K27M mutant DMG, IC50s ranged between 2.96 mM and 5.1 mM [115]. In this study, VPA antiproliferative effects were associated with dose-dependent augmentation of histone acetylation and apoptosis in tumor cells, whereas at the highest VPA dose tested (5 mM) minimal toxicity to rat hippocampal neuronal and glial cells was observed.

The HDAC-inhibitory activity of VPA, together with its radio and chemo-sensitizing potential [115, 116, 117, 118] and the limited side-effects in humans, implicate VPA as a potential neo-adjuvant therapeutic useful in multimodal treatment approaches. Currently, more than 80 trials are evaluating the safety and clinical benefits of VPA in combination with standard of care, or investigational therapies, in a vast array of cancers, including adult and pediatric CNS tumors (Table 2, Ref. [119, 120, 121, 122]).

Results obtained, to date, concerning the efficacy of VPA treatment show conflicting significance in glioma patients. A number of retrospective analyses have indicated VPA has moderate activity in newly diagnosed GBM patients treated with standard of care TMZ-based chemotherapy and radiotherapy, linking improved outcomes to the radio and chemo-sensitizing properties of VPA [123, 124, 125]. Common drawbacks for the majority of the studies reported are the small sample sizes, the historical bias, and the paucity of information on VPA administration protocols that did not assure these studies achieved anticancer concentrations in vivo with the dosing regimens used. Indeed, a pooled analysis of prospective clinical trials in a much larger cohort of patients (n = 1869) with newly diagnosed GBM did not uncover any association between use of the antiepileptic levetiracetam, or VPA, as add-on drugs to standard RT + TMZ and improvement of either PFS or OS [126]. However, it is very likely that patients received typical anti-seizure prophylaxis at doses of 5–10 mg/kg/day, although no information is specifically provided on VPA doses used. In another retrospective analysis of 359 patients with gliomas, VPA use positively correlated with a survival benefit in the GMB (WHO IV) group, whereas in Grade II/III gliomas VPA it was associated with a decrease in PFS and more rapid malignant progression. Conversely, higher doses of VPA (up to 25 mg/kg/day) added to concomitant RT + TMZ therapy (NCT00302159) resulted in an improvement of OS of patients with HGG over historical controls (Median OS of 29.6 months vs OS of 8.6–19.3 months previously reported in other studies) [119], and was associated with little late toxicity in a follow-up study [120]. A Phase II trial is currently under way to assess the efficacy of VPA plus sorafenib and sildenafil for the treatment of recurrent HGG (NCT01817751).

In the pediatric setting, the first report concerning the clinical efficacy of VPA is a single-case study on a 10-year-old boy with a radio and chemoresistant GBM, treated with a VPA monotherapy to reach a plasma trough level greater than 1 mM [127]. This is the threshold concentration associated with an anti-cancer effect and is 2 to 3-fold above concentrations commonly obtained in children with epilepsy. Although a complete remission was documented after 10 months on VPA, drug-related drowsiness led to discontinuation of VPA and tumor recurrence after 16 months. In two subsequent underpowered retrospective analyses in heavily pretreated pediatric patients with CNS tumors of different histologies, including EPN [128] and DIPG [129], prophylactic valproate treatment, although it was well tolerated, resulted in no, or barely, statistically significant survival benefits. A phase I study (NCT00107458) found dose-limiting somnolence at drug exposures required to maintain biologically relevant threshold levels, whereas VPA administered to reach trough concentrations up to 0.7 mM showed acceptable toxicity [121]. H3 and H4 hyper-acetylation in peripheral blood mononuclear cells (PBMCs), as surrogate markers of HDAC inhibition, were observed independently of VPA dose in half of the patients. Based on these findings, a multi-institution, phase 2 clinical trial of radiation and VPA, followed by maintenance VPA and bevacizumab (NCT00879437) was conducted in 38 children with glioma, specifically 20 with DIPG and 18 with HGG [122]. The treatment strategy was generally well tolerated, but no improvement in EFS or OS when compared with historical values was observed, although anecdotally encouraging tumor responses were observed.

More recently, VPA has been used in combinations with other drugs in pilot studies focused on evaluating personalized treatments for DIPG patients to specifically target FOSB-overexpressing tumors (4/15 cases) [84] or as a therapeutic backbone at a plasma trough levels of 0.5–0.7 mM (6/9 cases) [130]. Although no significant improvement of OS was observed in either study in the personalized treatment cohort compared to controls, 2 long-term survivors ($>$2 years) out of 9 patients were reported in one study, one of whom received a multidrug regimen that also included VPA [130].

The only trial so far to advance to a phase III study is under way in 49 centers (International cooperative Trial, NCT03243461) and is enrolling children and adolescents with HGG to evaluate the efficacy of VPA as a neo-adjuvant to TMZ and RT with respect to historical controls.

Overall, although VPA treatment has not been reported to yield satisfactory results in a number of studies, some patients have responded well to VPA when used in combination with other treatments. Therefore, further prospective studies are warranted to better understand potential clinical benefits of VPA in the treatment of CNS tumors.

Given the potential relevance of DRD2 in cancer, other DRD2 modulators can be potentially be repurposed for cancer therapy. The highly brain-penetrant antipsychotic haloperidol is FDA-approved for treating schizophrenia, Tourette syndrome, severe behavioral disorders, and hyperactivity in children [138]. Haloperidol has also shown promise in preclinical models of various cancer types, making it a good candidate for repurposing in the treatment of pediatric CNS tumors. Beside antagonizing DRD2 with an IC50 of approximately 30 nM, haloperidol also exerts anti-adrenergic (a1), anti-histaminic (H1), and anti-cholinergic action. However, haloperidol is associated with various side-effects including extrapyramidal syndromes [139]. In addition, it has an antagonist effect on Sigma-1 receptors that have been implicated in VEGF synthesis and cell proliferation [140, 141].

As an antineoplastic agent, haloperidol is an inducer of autophagy, apoptosis, and cell cycle arrest as well as being an inhibitor of MAPK activity [142]. In different GBM cell lines, haloperidol suppresses cell proliferation with IC50s ranging between 20 and 40 $\mu$M, and, moreover, induces a significant increase in apoptosis [141]. However, the effective concentrations in vitro are approximately 3 orders of magnitude greater than therapeutic plasma concentrations (2–50 nM) in both adult [141] and pediatric [143] populations. These values are significantly higher than that required for haloperidol inhibition of DRD2, suggesting that mechanisms other than DRD2 antagonism underpin the antiproliferative effects of haloperidol. Other studies have reported that DRD2 agonists increase formation, proliferation and invasiveness of GBM tumorspheres, whereas molecular knockdown or pharmacological inhibition by DRD2 antagonists, including haloperidol, decreased spheroid formation at concentrations close to those clinically attainable [144, 145]. Cotreatment of cells with TMZ and haloperidol results in a synergistic suppression of cell growth, an effect linked to TMZ-mediated upregulation of DRD2 expression [141, 146].

An inverse relationship between DRD2 and EGFR has been observed in a portion of clinical GBM samples in publicly available datasets [147]. Moreover, high EGFR expression is associated with chemoresistance to two different DRD2 antagonists, haloperidol and ONC201, in a panel of GBM patient-derived xenografts (PDXs) tested both in vitro and in vivo. These findings indicate that constitutive activation of EGFR signaling, through ectopic overexpression of a EGFRvIII mutant, confers resistance to DRD2 antagonists. Notably, a statistically significant inverse correlation between EGFR expression and ONC201 efficacy was reported in a phase II GBM trial (n = 15), suggesting that EGFR can serve as a chemo-predictive biomarker of sensitivity to DRD2 inhibition.

In contrast, others have reported haloperidol possesses pro-angiogeneticas well as neuroprotective effects. For example, chronic exposure of up to 72 h to relatively low concentrations of haloperidol (5 $\mu$M) induces upregulation of expression and secretion of BDNF and VEGF from the human GBM cell line T98-G [148, 149].

Many of the anticancer effects caused by haloperidol in vitro can only be observed at high concentrations, which are also likely to promote off-target receptor engagement. Therefore, further preclinical investigation on the potential repurposing of haloperidol is warranted.

MET is a veteran, first-line therapy for patients with Type 2 Diabetes (T2D), shows a very favorable safety profile, simplicity of administration [161, 162], and high blood brain barrier penetrability [163]. The first hints to an anticancer activity associated with MET stemmed from epidemiological studies that indicated an inverse relationship between treatment of diabetic patients with MET at standard clinical doses and the incidence of cancer [164, 165] or overall cancer mortality rates [166]. A recent systematic review and meta-analysis of cancer outcomes in MET users demonstrated that MET could be a useful adjuvant agent in patients with colorectal and prostate cancer [166]. By contrast, several studies failed to show an improved survival in diabetic women with breast cancer as a result of MET treatment [161]. Conclusive evidence in support of the use of MET as an add-on drug in anticancer therapy is still lacking, perhaps because most epidemiological studies are retrospective and present methodological limitations. This view strongly advocates for randomized controlled trials (RCTs) to investigate dose, duration, and efficacy of MET treatment in cancer patients [166].

The antineoplastic properties of MET are mediated by both a direct action on cancer cells and indirect effects on the TME [167, 168]. Primarily, MET affects tumor cell metabolism by the downregulation of the mTOR pathway [169], and the blockade of mitochondrial complex 1 that results in decreased ATP production and a consequent energetic stress [170]. It has been shown that MET exerts dual and opposing roles on mitochondrial respiration, stimulating mitochondrial biogenesis at pharmacological doses, or reducing it through decreasing adenine nucleotide levels at supra-pharmacological levels [171]. Indirect effects of MET on host metabolism are regulated by AMPK-dependent inhibition of hepatic gluconeogenesis and subsequent reductions in blood glucose and insulin. In addition, MET affects the TME by virtue of its anti-inflammatory [166] and anti-angiogenetic properties [167].

Preclinical studies conducted in vitro and in vivo have shown antineoplastic MET activity in models of different types of cancer [151, 168]. In glioma models, MET inhibits cell proliferation and invasiveness, induces apoptosis, autophagy, and GBM stem cell differentiation [167, 168]. However, the doses used in these preclinical investigations of up to more than 10,000 mg/L, are generally far above the plasma levels achievable in the clinic (0.465–2.5 mg/L) [172]. Recently, exosome-mediated co-delivery of MET and cPLA2 siRNA to target mitochondrial and phospholipid metabolism has shown promise in treatment of GBM PDXs and this combination exerted a significant increase in survival [173]. Therefore, potential use of MET as an add-on drug in glioma therapy, also in virtue of its radio-sensitizing [174, 175] and chemotherapy-sensitizing effects [176, 177, 178] warrants further preclinical evaluation.

In PFA tumors and cell lines with EZHIP overexpression, an enhanced glycolytic phenotype is associated with H3K27 acetylation enrichment at hexokinase-2, pyruvate dehydrogenase, and AMPKa-2 [156]. MET treatment reduces EZHIP expression while increasing H3K27me3, suppressing TCA cycle metabolism, and limiting tumor growth in vitro and in patient-derived animal models as well as sensitizing cells to the antitumoral effects of the HDAC inhibitor panobinostat [156]. Interestingly, sensitivity to MET appears to be dependent on EZHIP expression and shows reduced sensitivity in neural stem cells expressing catalytically inactive EZHIP. Together these data suggest that EZHIP can epigenetically rewire metabolic pathways in therapeutically-targetable PFA tumors. A possible limitation of the study is that MET exerts growth impairment of EPN cells in cultures and mouse models at supra-physiological concentrations that are generally thought to be unachievable in patients.

In DIPG models, dual targeting of mitochondrial function by a combination of MET and dichloroacetate synergistically suppresses cell proliferation through ROS-mediated DNA damage and apoptosis, and promotes radiosensitization both in vitro and in vivo [159], similarly to what has been reported in GBM models [179].

MET is currently under investigation in clinical trials of a large array of human malignancies, with more than 60 studies currently active. Importantly, the maximal daily dose of oral MET administered in the majority of these trials is between 1000 and 2000 mg/day (median 1700 mg) and that this is within the established dosing range used to treat patients with T2D [180].

A number of studies have addressed how MET influences progression of CNS tumors, however, discrepancies in these data warrant further observation. In a matched case-control analysis, a negative relationship was observed between long-term and poorly controlled diabetes and the risk of glioma, that this effect was stronger in men and older individuals, and that antidiabetic treatment was unrelated to risk [181]. One of the first studies to examine the anti-tumor activity of MET in primary GBM reported a statistically significant longer PFS in diabetic patients with MET therapy when compared to other diabetic patients (10.13 vs 4.67 months; p = 0.043) and other patients without diabetes (10.13 vs 6.7 months; p = 0.018) [182]. However, a seminal retrospective analysis on 1093 glioma patients found a favorable relationship between MET intake and OS and PFS in patients with grade III gliomas, but not grade IV gliomas [183]. Conversely, MET as monotherapy, or in combination with other drugs, yielded no improvement in life expectancy [184] in a pooled analysis of MET use and outcomes in 1731 individuals with newly diagnosed GBM from three randomized trials.

A dozen clinical trials focused on the use of MET in patients with CNS tumors are in the recruiting phase (Table 3, Ref. [185, 186, 187, 188]). A phase I factorial study (NCT01430351) examined the tolerability of TMZ in various combinations with three repurposed agents, specifically, MET, mefloquine, and memantine after completion of standard of care for GBM. Although an encouraging 2-year survival rate of 43% was observed, the trial was not sufficiently powered to evaluate the efficacy of interventions [185]. The clinical benefits of MET as an add-on agent to radiotherapy and chemotherapy has been assessed in a limited number of Phase II clinical trials in patients with newly diagnosed or progressive GBM. Based on encouraging preclinical results demonstrating that the combination of MET and TMZ yields superior antitumoral activity in GBM cells and tumors over monotherapy using any single agent [177], the efficacy and safety of low dose TMZ with or without MET were compared in patients with non-responsive GBM (NCT03243851). However, no results are currently available for this study. The addition of MET to TMZ and hypofractionated, accelerated radiotherapy (NCT02780024; M-HARTT STUDY) for the treatment of GBM patients was well tolerated and safe, and favorable outcomes were observed in those patients with low methylation levels of MGMT [186].

An ongoing multicentric Phase II study (NCT04945148; OPTIMUM) is aimed at evaluating the effect of MET to target OXPHOS in association with the standard first-line treatment with RT and TMZ in participants with GBM with wild-type IDH. This tailored approach is based on the observation that GBM with wild-type IDH present a unique and homogenous energetic metabolism that is specifically dependent on OXPHOS, and likely suppressed by MET, an oral inhibitor of mitochondrial complex 1. Newly diagnosed wild-type IDH GBM patients with the OXPHOS+ signature will be enrolled in the trial, with the major outcomes to be measured being PFS, OS and ORR.

Two ongoing pilot Phase II clinical studies (NCT04691960; NCT05183204) intend to evaluate the feasibility and activity of MET and a ketogenic diet (KD), a high fat, low carbohydrate diet, in the treatment of HGG. Because glucose is thought to be a driver of tumor growth, lowering blood glucose levels by KD and MET are hypothesized to have therapeutic potential in GBM. A previous safety study (NCT02149459) [187] found that combined metabolic therapy and RT in adult patients with WHO III and WHO IV HGG was well tolerated.

To our knowledge, no work on the cancer-related properties of MET in pediatric patients with CNS tumors has been published. In a pilot precision medicine study by Mueller and colleagues [84], MET was used in a 25 month old boy with H3 wild-type anaplastic astrocytoma to target IGF1R overexpression, in combination with MBZ, etoposide and carboplatin. This child progressed at 27.4 months from initiation of on-study therapy.

There is only one active trial on MET in children with primary CNS tumors and other solid tumors (NCT01528046). This study is aimed at analyzing the safety of MET as an adjuvant agent in a backbone regimen of vincristine, irinotecan, and TMZ, and as a secondary endpoint, MET antitumor activity.

Two other trials are focused on examining the potential of MET to promote brain repair in children treated with cranial radiation for MB. This study is based on preclinical observations in rodents where MET promotes neurogenesis and prevents brain injury by fostering recovery of sensory-motor and cognitive function [188]. In mice with transient middle cerebral artery occlusion, MET rescue of BBB disruption is mechanistically associated with a decrease in neutrophil infiltration, cytokine expression, endothelial injury, and an enhancement of neurobehavioral outcomes [189]. Beside stroke therapy, these MET effects may also counteract the pro-inflammatory and pro-angiogenetic microenvironment that promotes neuro-oncogenesis. Indeed, the use of MET is associated with better performance than placebo in tests of declarative and working memory in 24 long-term survivors of pediatric MB (NCT02040376) [188]. Based on these encouraging results, a multicenter phase III trial (NCT05230758) is currently investigating the effects of MET on cognitive function and the brain in an estimated population of 140 children treated for CNS tumors.

Celecoxib is a potent selective cyclooxygenase-2 (COX-2) inhibitor that functions by blocking the COX-2/prostaglandin E2 (PGE2) signal axis [201]. COX-2 catalyzes the rate-limiting step in the conversion of arachidonic acid to PGE2, that has been found to promote tumor development by exerting pro-survival, pro-angiogenetic, and proliferative effects as well as maintaining an inflammatory state that helps to escape immunity [202]. COX-2 is overexpressed in many tumors, is found in greater abundance in HGG than either LGG or normal brain tissue, and possesses a reverse relationship with patient prognosis [203].

In addition to its action on the immune/inflammatory and metabolism TMEs [45], celecoxib has emerged as a potent antineoplastic in different malignancies, including glioma [190, 204]. Its anti-proliferative and anti-metastatic properties are linked to its ability to affect signaling pathways crucial in oncogenesis such as AKT, NF-$\kappa$B, STAT3, and MMP. In addition, it also acts in a COX-2 independent manner as a “mitocan” and “pro-oxidant agent” in environmental contexts of low oxygen or glucose [201, 205]. As a mitocan, celecoxib impairs mitochondrial biogenesis in tumor cells. As a “pro-oxidant agent”, it inhibits the OXPHOS by disturbing electron transfer within the respiratory cycle, which leads to reactive oxygen species (ROS) overload and triggering mitochondrial ROS-induced apoptosis [201, 206]. Celecoxib also targets cancer stem cells by down-regulating WNT signaling, inhibiting EMT, and decreasing angiogenetic and growth promoting factors such as VEGF, FGF, and MMPs that restrain angiogenesis, metastasis, and stemness [201, 205, 207]. Computational proteo-chemometric analysis found that celecoxib ranked among top hits for binding to CDH11 [53], an adhesion protein that, in glioma, regulates cell migration and survival in vivo [208].

The therapeutic dose of celecoxib used to suppress inflammation is 300 mg, with plasma levels of approximately 2–7 $\mu$M, whereas that for cancer this does is 400 mg twice a day [209]. Half-maximal inhibitory concentrations for COX-2 are in the submicromolar range (0.29–0.87 $\mu$M) [201], whereas the IC50 values of celecoxib as an anticancer agent range between 20 and 70 $\mu$M across all tumor cell lines tested.

In preclinical models of glioma, celecoxib has proved to be the most potent selective COX-2 inhibitor to impair cell proliferation in vitro, although the high effective doses of 25–100 $\mu$M suggest COX-2-independent pathways [210, 211, 212]. A variety of mechanisms underlie the anti-proliferative effects of celecoxib, including proteosomal degradation of survivin [211], p53-dependent G1 cell cycle arrest [213], potent reduction of WNT/$\beta$-catenin/Tcf [212, 214] and STAT3 signaling pathways [215], and downregulation of the CCL2/CCR2 and CXCL10/CXCR3 signaling axes [216]. In addition, celecoxib reduces self-renewal capacity and increases apoptosis in GBM [217] and medulloblastoma-derived stem cells [215]. Recent evidence implicates COX-2 as a target of miRNA-128 through direct binding to the 3${}^{\prime }$UTR of the COX-2 transcript [218]. Additionally, a miR-128 inhibitor significantly increased COX-2 mRNA expression and proliferation of glioma cells.

There is scarce literature on celecoxib as single agent in pediatric CNS tumor models, and the available evidence is mostly restricted to MB, where celecoxib hampers tumor growth in vitro and in vivo [215, 219]. Of note, in the work of Lin and coworkers conducted in DIPG, celecoxib was found to be among the 371-potency-selected agents uncovered in a panel of 2706 approved and investigational drugs [66].

Celecoxib has been included in a large number of clinical trials in CNS tumors (Table 4, Ref. [220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231]) due to a favorable side effect profile, antiangiogenic properties, tumor sensitization to chemotherapy, targeted therapy, and radiation therapy [215, 232, 233].

In the adult setting, a phase II clinical trial (NCT00047294) enrolling patients with newly diagnosed GBM examined the addition of metronomic celecoxib and thalidomide as antiangiogenic agents to adjuvant, post-irradiation TMZ. Although the three-drug combination was well tolerated, no significant survival advantage was reported [220]. The outcome in terms of PFS and OS was examined in the adjuvant setting following chemoradiation in GBM patients (NCT00112502). In this factorial study, dose-dense (dd)-TMZ was used as the backbone for combinations with one or more 3 chemotherapy agents that included celecoxib. Although the multiple cytostatic agent regimen was safe and generally well tolerated in combination with dd-TMZ [221], a follow-up study failed to demonstrate superiority of any of the combinations vs dd-TMZ alone [222].

The combination of celecoxib with 6-thioguanine and capecitabine plus lomustine or TMZ (NCT00504660) [223] failed to achieve any marked improvement in PFS or OS over earlier studies on recurrent GBM. Conversely, this multidrug regimen proved to be promising and well tolerated for patients with high-grade anaplastic glioma. These patients fared better than those treated only with an alkylating agent, although the number of patients did not reach the preset statistical certainty. In this group of patients, PFS-12 was 44% (95% CI 29–67%) vs 21–33% of the most effective therapies in combined analysis of 16 previous studies.

Celecoxib has also been included in the polypharmacological regimen CUSP9v3, that uses nine repurposed non-oncological BBB-penetrating drugs in combination with metronomic TMZ in recurrent/progressive GBM (NCT02770378) [224, 225]. Preclinical investigation showed that the 9-drug combination exhibited marked cytocidal, anti-proliferative, and antimigratory actions in GBM stem cells and cell lines [209, 224]. It should be noted that seven of the drugs in this multi-agent strategy, including celecoxib, exerted limited anti-tumor effects within the dose range covering clinically achievable concentrations [225]. In a previous pilot study on 9 GBM patients, the CUSP9v3 approach was shown to be generally safe and well-tolerated, showing potential positive effects that require replication in larger trials [225].

In pediatric CNS tumors, different trials have reported a low toxicity profile of various multi-agent regimens that combine chemotherapy with various ancillary agents, including celecoxib, to both target tumor cells and modulate the TME [234, 235]. A pilot study, NCT00165451, tested the feasibility of a four-agent antiangiogenic chemotherapy regimen with continuous oral celecoxib in heavily pretreated children with recurrent or progressive cancers, including EPN and DIPG [226]. This drug combination was well tolerated and associated with a prolonged disease-free status, mostly in patients with EPN, although the study was not powered for disease-specific efficacy evaluation. A similar five-drug antiangiogenic chemotherapy regimen that also included metronomic fenofibrate, was trialed on 97 children with hematologic and solid tumors, including HGG, LGG, EPN and MB [236]. This trial confirmed acceptable toxicity in the majority of patients and noted clinical benefit only in EPN and LGG. In a phase II pilot study, NCT01756989, the combination of an antiangiogenic triple medication consisting of thalidomide, celecoxib, and etoposide was used in a metronomic fashion after local RT with topotecan in 8 children with inoperable DIPG (ANGIOCOMB protocol) [237]. This protocol was easily administered and well tolerated and associated with improved quality of life (QOL) for the patients over historical controls (eight patients with DIPG, who received RT only). In a larger cohort of patients (n = 41) the ANGIOCOMB protocol achieved a 12 and 24 month OS of 61% and 17%, with a prolonged QOL [227]. However, these results were not confirmed in another group of 17 children with refractory or high-risk malignancies where Grade III-V adverse events occurred in 76% of the patients [228].

Phase II multicentric trials focusing on the clinical efficacy of combinatorial regimens that also include celecoxib have been completed, or in the recruiting phase, in pediatric patients with recurrent/refractory solid and CNS tumors. The trial NCT01285817 examined the safety profile and anti-neoplastic efficacy of the SFCE METRO-01 regimen, based on a combination of cyclophosphamide, methotrexate, vinblastine, and continuous celecoxib. The cohort included 44 children with extra-cerebral tumors [229] and 29 children with CNS tumors, including EPN, HGG, and LGG [230]. Confirming a previous toxicity profile indicated in a pilot study in 16 children [234], the regimen was well tolerated, although it proved to be active only in patients with LGG with a potential clinical benefit in patients with neuroblastomas. The trial NCT01356290 (Metronomic and Targeted Anti-angiogenesis Therapy for Children With Recurrent/Progressive Medulloblastoma, Ependymoma and ATRT; MEMMAT), focused on testing the survival benefits of the antiangiogenic multidrug combination bevacizumab, thalidomide, celecoxib, fenofibrate, etoposide, and cyclophosphamide that was augmented with alternating courses of intraventricular etoposide and cytarabine in patients with recurrent EPN, MB or atypical teratoid/rhabdoid tumor (AT/RT). The results of this study are currently being evaluated. This trial was based on preliminary encouraging results reported in a cohort of 16 children with recurrent embryonal tumors [235], where treatment with this drug combination showed to be particularly beneficial in patients with MB, with both OS and EFS after 12 months of 85.7 $\pm$ 13%, vs the historical data of only 23–30%. By contrast, all patients with CNS PNET had died by 12 months. Very recently, a long-term follow-up of 29 patients with recurrent medulloblastoma prospectively treated according to the MEMMAT strategy between 2006 and 2016 has been published [238]. This study documented improved EFS and OS with respect to previously published series, with 9/29 patients still alive as of 07/2022. An ongoing phase II trial, NCT02574728, is assessing the response to treatment of daily celecoxib and sirolimus, a potent immunosuppressive agent, in combination with the common chemotherapy drugs etoposide, cyclophosphamide in patients with solid tumors, including EPN. A preliminary Phase I study, NCT01331135, aimed at determining the MTD, toxicities, and response to this treatment in a small cohort of children and young adults found that the combination was well tolerated [231]. Moreover, half of the patients experienced stable disease, some for extended durations, and one patient showed a partial response.