Opportunities and Challenges of Small Molecule Inhibitors in Glioblastoma Treatment: Lessons Learned from Clinical Trials

Simple Summary Glioblastoma is the most common brain tumour, with a poor prognosis of about 15 months despite intensive treatment. Many trials have tested new drugs targeting specific genes, but none have succeeded, and treatments have not changed since 2005. This review explores why clinical trials with these drugs failed. It highlights the potential of combining different drugs to overcome resistance and suggests ways to improve future trials. The goal is to understand treatment failures and find new drug combinations to improve survival for GBM patients. Abstract Glioblastoma (GBM) is the most prevalent central nervous system tumour (CNS). Patients with GBM have a dismal prognosis of 15 months, despite an intensive treatment schedule consisting of surgery, chemoradiation and concurrent chemotherapy. In the last decades, many trials have been performed investigating small molecule inhibitors, which target specific genes involved in tumorigenesis. So far, these trials have been unsuccessful, and standard of care for GBM patients has remained the same since 2005. This review gives an overview of trials investigating small molecule inhibitors on their own, combined with chemotherapy or other small molecule inhibitors. We discuss possible resistance mechanisms in GBM, focussing on intra- and intertumoral heterogeneity, bypass mechanisms and the influence of the tumour microenvironment. Moreover, we emphasise how combining inhibitors can help overcome these resistance mechanisms. We also address strategies for improving trial outcomes through modifications to their design. In summary, this review aims to elucidate different resistance mechanisms against small molecule inhibitors, highlighting their significance in the search for novel therapeutic combinations to improve the overall survival of GBM patients.


Introduction
Glioblastoma (GBM) is the most common malignant primary brain tumour.The 2016 CNS WHO classification defines two groups: (1) isocitrate dehydrogenase (IDH)wildtype (wt) glioblastoma and (2) IDH-mutant (mt) grade IV astrocytoma [1].IDH-wt GBM is mainly diagnosed in adults aged 50 to 60, and patients have a median survival of 15 months [2,3], whereas IDH-mutant grade IV astrocytoma is mostly diagnosed between the ages of 35 and 45 and is the result of dedifferentiation from a low-grade astrocytoma [2,4].Patients with IDH-mt grade IV astrocytoma have a median survival of approximately 31 months [5].This review will focus on IDH-wt GBM.
After diagnosis, treatment consists of maximal surgical resection, followed by radiation and chemotherapy (temozolomide) [3,6].Even with this intensive multimodality treatment schedule, patients with GBM cannot be cured, and recurrence is inevitable for all patients [3,5].Despite multiple trials, no innovative life-prolonging treatments have been identified for patients with IDH-wt GBM to improve the current standard of care.
One of the main reasons for the failure of the clinical trials in GBM is inter-and intra-tumoral heterogeneity on the phenotypic, genotypic and transcriptional levels.The intertumoral heterogeneity of GBM tumours is marked by four subtypes Verhaak et al. identified by genomic analysis [7].First, the classical subtype is primarily characterised by alterations in epidermal growth factor receptor (EGFR).This subtype also shows a homozygous deletion spanning the Ink4a/ARF locus.Second, the pro-neural subtype is primarily characterised by amplification or activating the mutation of platelet-derived growth factor receptor A (PDGFRA) and point mutations in IDH1 and TP53.Third, the neural subtype is characterised by the expression of neuronal markers such as neurofilament light chain (NEFL), gamma-aminobutyric acid type A receptor subunit alpha1 (GABRA1), synaptotagmin 1 (SYT1) and potassium chloride transporter member 5 (SLC12A5).Finally, the mesenchymal subtype is primarily characterised by the loss of neurofibromatosis type I (NF1) [7].
In addition to this intertumoral heterogeneity, Patel et al. found that one genotypic subtype does not characterise a GBM tumour, but rather, several subtypes occur in one tumour [8].Additionally, intra-tumoral heterogeneity is further supported by identifying four main cellular states in which tumour cells in GBM exist: neural progenitor-like (NP-like), oligodendrocyte-progenitor-like (OPC-like), astrocyte-like (AC-like) and mesenchymal-like (MES-like) states [9].These states are influenced by the tumour microenvironment, which has been shown to have a spatially heterogeneous immune compartment based upon the immune cell subtype and activation levels [9].Alterations in cyclin-dependent kinase 4 (CDK4), EGFR, PDGFRA and NF1 are associated with the relative frequency of cells in each state, which varies between GBM samples [9].This inter-and intra-tumoral heterogeneity in GBM may be the reason why "one shoe fits all" conventional treatment options do not achieve long-term responses in GBM [8][9][10].
Achieving a uniform drug distribution in GBM is complex due to the intact bloodbrain barrier (BBB) in some areas.The BBB, primarily composed of endothelial cells forming tight junctions, restricts drug transport from the capillaries to the brain [11,12].Only drugs with a molecular mass under 500 Da and high lipid solubility can pass the BBB [13].Consequently, all large molecule agents, such as gene therapeutics and monoclonal antibodies, and over 98% of small molecule inhibitors cannot cross the BBB [13].Efflux transporters further complicate drug delivery by expelling chemotherapy and small molecule inhibitors at the BBB [14,15].
Small molecule inhibitors might offer a potential solution to the inter-and intratumoral heterogeneity in GBM by targeting the drivers of tumorigenesis [16].They can bind to various intracellular and extracellular targets [17,18].Small molecule inhibitors can either act as enzyme inhibitors, reducing enzyme activity, or as receptor antagonists, countering receptor effects [17,18].These compounds can function as multi-kinase inhibitors, targeting a wide range of the human kinome, or as selective inhibitors, targeting a specific component of a signalling pathway [19].By targeting different signalling pathways and cellular processes involved in cancer, small molecule inhibitors are therefore an interesting treatment option for various types of cancer [19].In BRAF-mutated metastatic melanoma and EGFRmutated advanced non-small cell lung cancer (NSCLC), small molecule inhibitors targeting BRAF or EGFR, respectively, significantly increased the overall survival and have become the standard of care in these patients [20][21][22].
Nonetheless, despite different trials with small molecule inhibitors in patients with GBM, none has led to an improvement in the standard of care for these patients.
Therefore, we will review trials involving small molecule inhibitors in GBM patients and discuss possible resistance mechanisms, contributing to their lack of efficacy.Furthermore, we will propose potential combinatorial therapies based on the molecular characteristics of individual tumours.

Clinical Trials with Small Molecule Inhibitors in GBM 2.1. Mono-Target Small Molecule Inhibitors
Clinical trials in GBM with small molecule inhibitors specifically inhibiting one specific target are listed in Table 1.
Only a few mono-target small molecule inhibitors evaluated in phase I GBM trials have progressed to phase II trials, mostly with disappointing results (Table 1).However, most of these trials are biomarker naïve and unable to identify "on-target" effects of the specific drug, monitor treatment response or identify resistance mechanisms.This challenge arises since re-sampling of the tumour during treatment is not feasible due to its location, which could damage eloquent brain areas and result in a subsequent loss of function.Furthermore, the potential of liquid biopsy in GBM remains unexplored, as it faces several challenges, such as the extremely low concentrations of biomarkers in the blood and their short half-life, both of which complicate detection [23].So far, no prognostic or predictive biomarkers have been identified in the blood of GBM patients [23,24].Further studies are necessary to assess the sensitivity and specificity of liquid biopsies in GBM.
The mono-target compounds that have shown some clinical benefit, defined as an improvement in progression-free survival (PFS) and/or overall survival (OS) compared to the standard of care, in (a subset of) GBM patients included in these trials, will be discussed in more detail.
One of the most frequently altered genes in GBM is the epidermal growth factor receptor (EGFR).EGFR is a tyrosine kinase receptor amplified and constitutively activated in 57% of all GBM patients [4].In fifty percent of the tumours with EGFR amplification, activating EGFR gene rearrangements occur, with the most common extracellular domain mutation being EGFRvIII [25,26].This mutation leads to a deletion of exons 2-7 of the EGFR gene and renders the mutant receptor incapable of binding any known ligand.Despite this, EGFRvIII displays low-level ligand independent constitutive signalling that is amplified by reduced internalisation and downregulation.
The small molecule inhibitors erlotinib and gefitinib, both binding to the kinase domain of EGFR and inhibiting phosphorylation and activation, have been assessed in multiple trials (Table 1).Overall, the response rates were low, and overall survival did not improve significantly.Moreover, EGFRvIII/EGFR amplification did not correlate with a better outcome compared to EGFR wildtype [27][28][29][30].
However, in a small subset of patients, a stabilisation or even decrease in tumour volume was identified when treated with erlotinib.Haas-Kogan et al. showed, based upon a tissue analysis, that erlotinib has a significantly better effect on tumours expressing high levels of EGFR and low levels of phosphorylated protein kinase B (AKT) compared to tumours with low EGFR levels and high levels of phosphorylated AKT [31,32].This indicates that activation might be a resistance mechanism of GBM to EGFR inhibitors.This mechanism was also suggested in a trial investigating gefitinib, which showed a significantly better PFS and OS in patients with de novo GBM with EGFR gene alterations combined with wildtype phosphatase and tensin homolog (PTEN) (inactivation of the AKT pathway) compared to patients with wildtype EGFR and altered PTEN (activation of the AKT pathway) [33].Similarly, in EGFR-mutated non-small cell lung cancer (NSCLC), PTEN loss is known as a resistance mechanism to EGFR inhibitors [34,35].
A phase II trial also found that erlotinib treatment in patients with recurrent GBM resulted in improved six-month progression-free survival (PFS-6) and comparable median OS compared to historical values for patients undergoing treatment with irinotecan [36].Nonetheless, because of the small number of responses, no conclusions could be drawn from the molecular subgroup analyses.Based on the limited efficacy of erlotinib and gefitinib in the total, unstratified GBM study population, these compounds are currently not approved for the standard of care in GBM patients.In addition to the trials mentioned above, other trials have been conducted investigating mono-target small molecule inhibitors but were not able to show an effect on tumour size and therefore did not improve overall survival for GBM patients.These trials investigated the effects of buparlisib (phosphoinositide 3-kinase (PI3K) inhibitor), capmatinib (mesenchymal epithelial transition (c-MET) inhibitor), periforsine (AKT inhibitor), deforolimus (mechanistic target of rapamycin (mTOR) inhibitor), PF-06840003 (indoleamine 2,3-dioxygenase-1 (IDO-1) inhibitor), GSK2256098 (focal adhesion kinase (FAK-kinase) inhibitor), adavosertib (Wee-1 inhibitor), pegdinetanib (VEGFR-2 inhibitor), navtemadlin (mouse double minute 2 homolog (MDM2) inhibitor), tipifarnib (FTase subunit ß inhibitor), AXL1717 (insulin-like growth factor 1 receptor (IGF-1R) inhibitor) and selinexor (exportin 1 (XPO-1) inhibitor).More details on these trials can be found in Table 1.
In summary, mono-target small molecule inhibitors showed poor efficacy in GBM, with only a few trials demonstrating modest efficacy in a specific subgroup.An explanation for the ineffectiveness of mono-target small molecule inhibitors is the clonal selection of resistant tumour cells due to intra-tumour heterogeneity [8,9].Investigation of primary GBM tumour samples has revealed that multiple receptor tyrosine kinases (RTKs) are activated within a single tumour sample [8,10,37].Moreover, malignant cells in a GBM tumour exhibit plasticity, and a single cell is able to generate all four states [9].This plasticity of the cells leads to changes in distribution of these states after targeting genetic drivers [9].By inhibiting only one genetic driver-for example, EGFR-only cells harbouring alterations in EGFR will be targeted, while the other tumour cells remain unaffected, leading to sub-clonal selection and, eventually, to recurrence.
Additionally, most trials include unstratified GBM patient populations.Consequently, by not taking biomarkers or tumour heterogeneity into account, GBM tumours are included that are already intrinsically resistant to small molecule inhibitors because of their molecular characteristics.
Therefore, multitarget small molecule inhibitors might be more effective in tackling the inter-and intra-tumoral heterogeneity than mono-target small molecule inhibitors and prevent sub-clonal selection [9].

Multitarget Small Molecule Inhibitors
Clinical trials with small molecule inhibitors targeting multiple targets are presented in Table 2.
Remarkably, most trials that showed a promising result after treatment with a multitarget small molecule inhibitor involved small molecule inhibitors targeting vascular endothelial growth factor (VEGF/VEGFR).Since GBM exhibit extensive vascularity, VEGF/VEGFR is an important target [38].
Different small molecule inhibitors have been designed to target VEGF/VEGFR.For example, in a randomised phase II trial comparing axitinib, which, apart from VEGFR, also targets PDGFRß, and c-Kit, with physicians' best choice of therapy in patients with recurrent GBM, axitinib monotherapy was found to increase the 6-month PFS and overall response rate but not OS [39].
Another trial compared treatment with axitinib alone to axitinib combined with lomustine in patients with recurrent GBM.Axitinib improved the response rate and progressionfree survival in this population compared to historical controls, but the overall survival of these patients treated with axitinib alone did not improve [40].
A phase II trial investigating the effects of cediranib, a pan-VEGFR inhibitor also targeting FLT1/4, c-Kit and PDGFRβ, showed promising results in patients with recurrent GBM [41].Since its results were comparable to historical controls, the inhibitor proceeded to a phase III trial.However, the primary endpoint of the trial (PFS prolongation) when comparing cediranib as a monotherapy and in combination with lomustine was not met.Subgroup analysis investigating the effect of the VEGFR baseline levels showed no significant effect after cediranib monotherapy compared to lomustine monotherapy [42].Hence, cediranib has not been approved for GBM.Regorafenib, which inhibits VEGFR, PDGFR, c-Kit, c-RET, Raf-1, FGFR and Abl, showed in a randomised, controlled, phase II trial a significant improvement in overall survival in patients with recurrent GBM compared to lomustine [43].Regorafenib therefore continues to be investigated in a phase III trial, the results of which are awaited (NCT03970447).One explanation for regorafenib's potential as an inhibitor lies in its role as a multi-kinase inhibitor, targeting not only VEGFR but also targeting kinases involved in oncogenesis and the tumour microenvironment [44].
Although fusion genes were discovered a long time ago, their significance and role in oncogenesis was not recognised until recently [45,46].Currently, different tyrosine kinase inhibitors (TKI) are investigated in trials to target fusion genes in recurrent GBM patients [47].In the context of GBM, fusion genes are present in approximately 30-50% of all GBM patients, and druggable fusion genes are found in approximately 4% of these patients [48,49].Multiple fusion genes have been identified, with the most prevalent involving MET, EGFR, FGFR, NTRK, RET and ROS [47,50].Currently, several TKIs are investigated in trials to target these fusion genes in recurrent GBM [47].Infigratinib (FGFR 1/2/3 inhibitor) showed limited efficacy in the entire recurrent glioma patients with the FGFR alterations population but demonstrated a durable response in patients with FGFR1 or FGFR3 point mutations and those with FGFR3-TACC3 fusions [51].More research is necessary to clarify the efficacy of TKIs targeting fusion genes in a biomarker-driven GBM patient cohort.
In addition to the trials mentioned above, many trials have been conducted investigating multitarget small molecule inhibitors, but none were able to improve OS (Table 2).
While there are no phase III trials with mono-target small molecule inhibitors, there have been two phase III trials with multitarget small molecule inhibitors.However, none have shown an improvement in overall survival in a phase III trial so far.

Small Molecule Inhibitors Combined with Chemotherapy/Bevacizumab
An overview of all the trials involving small molecule inhibitors combined with chemotherapy or bevacizumab can be found in Table 3.
Of all these trials, only two showed an increase in PFS and OS [52,53].In the first trial, recurrent GBM patients with VEGF overexpression and EGFRvIII mutation showed an increase in the response rate and PFS after treatment with erlotinib combined with bevacizumab.However, since the trial only included four patients, no conclusions could be drawn [52].In addition, other trials involving erlotinib combined with bevacizumab have also not shown an increase in PFS or OS [54,55].However, these other trials were not biomarker-driven.
In the second trial, anlotinib, a multitarget small molecule inhibitor with similar targets as regorafenib, showed promising results in a phase II single-arm trial in twentyone recurrent GBM patients in combinations with temozolomide with an increase in the response rate, PFS (7.3 months (95% CI 4.9-9.7))and median OS (16.9 months (95% CI 7.8-26.0))compared to historical controls [53].
Additionally, different trials have investigated the effects of different small molecule inhibitors combined with chemotherapy or bevacizumab.Lomustine has been investigated together with buparlisib (PI3K inhibitor), but this combination did not improve OS in 18 recurrent patients [56].The combination of hydroxyurea with imatinib and vatalanib (VEGFR/FLT inhibitors) also did not show a positive effect on the OS of these patients [57,58].Trials investigating the combination of irinotecan with CT-322 (VEGFR-2 inhibitor) or sunitinib (VEGFR/PDGFR inhibitor) were unsuccessful as well [59,60].Other small molecule inhibitors have been investigated in combination with temozolomide or bevacizumab, but almost all of these trials could not show an increase in OS (Table 3).

Combinations of Small Molecule Inhibitors
A method to target intra-tumoral heterogeneity is to target multiple kinases in the same tumour.Only a limited number of trials have investigated the combinatorial treatments of different small molecule inhibitors.These trials are listed in Table 4.
A trial combining gefitinib and cediranib, targeting EGFR and VEGFR, showed a trend towards an improved response rate in recurrent GBM patients in a phase II trial [61].Unfortunately, this trial was discontinued prematurely, and no statistically significant analysis of the PFS and OS was performed.
Most of the trials combined different small molecule inhibitors with a mTOR inhibitor.The PI3K/Akt/mTOR pathway is frequently altered in GBM, making it an interesting target [64].As discussed above, targeting this pathway is not effective as a monotherapy (Tables 1 and 2).Combinations with erlotinib or gefitinib and sirolimus or temsirolimus (mTOR inhibitor) and temsirolimus with perifosine (AKT inhibitor) or sorafenib (Raf/VEGFR/PDGFR inhibitor) have been investigated.None of these trials, however, showed an increase in the PFS, OS or response rate (Table 4).This is most likely due to the fact that the majority of PI3K/mTOR inhibitors are unable to pass the BBB and achieve adequate concentrations at the tumour site [65,66].
In addition, researchers have investigated combinations of different (multitarget) small molecule inhibitors, targeting various pathways, none of which showed an increase in OS in GBM patients (Table 4).

Intrinsic Versus Acquired Resistance
The discouraging results of trials with small molecule inhibitors in patients with GBM underscore the complexity of developing a treatment schedule for these patients.Multiomics analysis has unravelled the complexity of proteomic, phosphorylation, metabolic, lipidic and immunogenic alterations and their influence on GBM subtypes and survival; yet, despite this progress, a uniformly present and druggable driver oncogene across all GBM tumour remains elusive, making a "one show fits all" treatment targeting all GBM subclones unavailable for these patients [4,[7][8][9]67,68].
Extensive research has been conducted investigating the effects of mono-target small molecule inhibitors (Table 1), but most of these trials did not show an improvement in PFS and/or OS.One of the possible reasons for failure of these trials is the lack of patient stratification.In two studies, subgroup analyses demonstrated that patients with amplified EGFR and inactivation of the AKT pathway responded better to EGFR inhibitors than those without [31][32][33].Additionally, the investigation of primary GBM tumour samples has revealed that multiple RTKs are activated within a single tumour sample [8,10,37].The inhibition of only one genetic driver will lead to the clonal selection of resistant tumour cells and, eventually, recurrence.
In addition to inter-and intra-tumoral heterogeneity and the clonal selection of resistant tumour cells, other mechanisms may contribute to the primary resistance to small molecule inhibitors in GBM.First, while the majority of small molecule inhibitors primarily target the kinase domain of a receptor, the situation differs in GBM, where mutations may not always be confined to this domain.An illustration of this variability is observed in the case of the EGFR receptor, which most prevalent mutation, the EGFRvIII mutation, is situated in the extracellular domain of the EGFR receptor, resulting in the constitutive activation of non-mutated kinase domains [25,26].Third-generation EGFR inhibitors, like Osimertinib, used in EGFR-mutated non-small cell lung cancer, specifically targeting the mutated kinase domains, are ineffective in EGFR-amplified or EGFRvIII-mutated GBM [25,26].
Lastly, GBM tumours can bypass the effect of a single small molecule inhibitor by activating other pathways.These bypass signalling pathways could be another explanation for treatment resistance, which supports the need for combination therapies.
In GBM, ERBB3, IGFR1R and TGFBR2 were positively correlated with PDGFR, and combining a PDGFR inhibitor with either an ERBB3 inhibitor or an IGF1R inhibitor resulted in a more significant reduction in tumour growth than each inhibitor alone [69].
Moreover, the upregulation of c-MET has been found after EGFR monotherapy in glioblastoma stem cells (GSCs) in vitro, causing long-term self-renewal ability in these cells [70].The inhibition of both MET and EGFR resulted in overcoming the resistance to an EGFR inhibitor (gefitinib) in a preclinical mouse model [71].Further research, however, found that, after EGFR and c-MET inhibition, ERK was reactivated via a NF-kB-dependent feedback loop that led to the activation of FGFRs.This bypass resistance mechanism could be overcome in vivo through simultaneously inhibiting FGFR [72].More research has demonstrated that the expression of EGFRvIII induces the transactivation of c-Jun N-terminal kinase isoform 2 (JNK2) in GBM cells, which, in turn, activates the hepatocyte growth factor (HGF)/c-MET signalling pathway [73].
These preclinical results demonstrate crosstalk between different RTKs but also highlight that the inhibition of three or more RTKs is essential to overcome this resistance mechanism.While this approach is feasible in preclinical settings, using a combination of three or more mono-target small molecule inhibitors in patients could lead to severe toxicity, hindering clinical application.Consequently, targeting downstream feedback loops may circumvent the need for multiple inhibitors, thereby reducing multi-drug toxicity while maintaining a high efficacy in overcoming treatment resistance in GBM.
Crosstalk between the PI3K/mTOR pathway and MEK/ERK pathway is one of these feedback loops.Inactivation of the PI3K/mTOR pathway activates the MEK/ERK pathway and vice versa, thereby bypassing the effect of either PI3K/mTOR inhibitors or MEK/ERK inhibitors [74,75].The combinatorial treatment of a MEK inhibitor with a dual PI3K/mTOR inhibitor was therefore found as a synergistic combination treatment in vitro [74].However, the synergistic efficacy in vivo was minimal, most likely due to subtherapeutic dosing due to dose-limiting toxicity and poor tumour penetrance [76].Trials exploring combinations of the MAPK pathway and PI3K/mTOR pathway have proven unsuccessful due to their low efficacy because of low tumour penetrance and high toxicity (Table 4) [77][78][79].Therefore, novel brain-penetrant PI3K and mTOR inhibitors, such as paxalisib, are of interest and are currently being evaluated.
Even though most trials involving multitarget small molecule inhibitors have been ineffective (Table 2), a few inhibitors tend to show some promising results, all targeting VEGFR.Regorafenib and anlotinib both showed an improvement in OS, but in both trials, the treatment group exhibited more favourable prognostic factors, potentially leading to an overestimation of the observed improvement in OS [39,40,43,53].
Moreover, anlotinib was more effective than regorafenib, which could be explained by the fact that anlotinib was combined with TMZ in GBM.Preclinical research has shown that anlotinib combined with TMZ decreased GBM growth more than anlotinib alone via inhibition of the JAK2/STAT3/VEGFA pathway [84].This illustrates how the concurrent administration of chemotherapy with, in particular, an angiogenesis inhibitor can enhance the efficacy.However, a combination of bevacizumab with lomustine in recurrent GBM failed to show a survival advantage in a phase III trial [85].
In other types of solid cancer, such as colorectal carcinoma, the inhibition of bevacizumab/VEGFR also failed to show an improvement in OS [86].This indicates that targeting angiogenesis alone merely slows tumour growth and the tumour can circumvent the inhibition, which results in an increase in PFS but not in OS.
Different combinatorial treatments with small molecule inhibitors have been investigated, most of them with disappointing results (Table 4).One trial investigating the combination of dabrafenib with trametinib in recurrent BRAF V600E-mutated GBM patients showed an increase in OS [63].Dabrafenib combined with trametinib is currently considered the standard of care in progressive BRAF V600E-mutated CNS tumours [87].Most likely, this combination is successful, since, by adding a MEK inhibitor, the MAPK pathway is inhibited at two positions.

Influence of the Tumour Microenvironment
The likelihood of a combination treatment exclusively targeting tumour cells curing GBM is low, since the tumour microenvironment (TME) also plays a significant role in therapy resistance within GBM.
The TME in a GBM tumour contains not only tumour cells but also non-neoplastic cells, including immune cells, vascular cells and other glial cells [88].GBMs are considered as immunogenic cold tumours due to low numbers of tumour-infiltrating lymphocytes (TILs) and other immune effector cell types [89].
One of the critical immunosuppressive factors in the GBM microenvironment is hypoxia, which suppresses the antitumour functions of immune cells that are able to infiltrate into the GBM microenvironment.Vice versa, functions of immunosuppressive cells, such M2 macrophages, are enhanced by hypoxia [90,91].
Trials investigating the currently available immune checkpoint inhibitors (ICIs) in newly diagnosed and recurrent GBM have not demonstrated an improvement in overall survival [92][93][94].A key factor limiting the effectiveness of these inhibitors is the very low number of TILs in the GBM microenvironment [89,95].
Consequently, future research must focus on targeting the GBM-specific immune environment, which is characterised by immunosuppressive tumour-associated macrophages (TAM).To achieve this, it is imperative to elucidate and understand the interplay between the GBM tumour and TAMs.Additionally, more information is required on how inhibitors might be able to shift the TME to become immune-permissive.The combination of Lenvatinib with anti-PD-1 antibody and cabozantinib with a poxviral-based cancer vaccine caused a more permissive TME with a decrease in TAMs and activation of the type-I interferon response when investigated in other types of cancer [96,97].
Preclinical research in GBM has revealed that combining an EGFR inhibitor with a mTOR inhibitor can modulate the TME by downregulating immunosuppressive chemokines and inhibiting tumour-promoting macrophage infiltration [98].Thus, small molecule inhibitors could possibly be used with a novel intent to create a shift towards an immunepermissive TME.This approach might lead towards innovative combinations of small molecule inhibitors with immunotherapy, potentially preventing the clonal expansion of resistant tumour clones.Further research is needed to explore whether such combinations of immunotherapy with small molecule inhibitors will be effective and overcome treatment resistance in GBM.

Clinical Trial Design
In addition to the reasons mentioned above, the disappointing results observed in trials involving small molecule inhibitors for glioblastoma patients could be attributed to several other factors related to the trial design.
The BBB makes it difficult for different drugs to arrive in effective concentrations at the tumour site in brain cancer [12,13].The aim of phase 0 trials is to investigate the biological effect of a study drug in human patients by using a pharmacologically active but non-therapeutic dosage [99].However, in brain tumours, a higher systemic dose will be necessary to detect study drug levels in the CNS with greater risks of related side effects [100].Different techniques to open the BBB, such as focused ultrasound (FUS), are being investigated.These techniques might be a way to enhance drug delivery at the tumour site without the need to increase the systemic dose [101][102][103].
The advantage of the study design of phase 0 trials is that it allows for the examination of different patient samples over time, such as blood, cerebrospinal fluid and tumour tissue, hereby giving information on pharmacokinetics and -dynamics, BBB passage and target inhibition [99,100,104].On the one hand, this information will be crucial to eliminate drugs with low tumour penetration or target inhibition and prevent unnecessary burdens on patients in phase I and II trials.On the other hand, newly developed drugs that are able to pass the BBB and inhibit the desired target effectively can be taken into phase I and II trials to investigate their effect on tumour growth, PFS and OS.
Additionally, the absence or limited use of molecular markers in patient selection might contribute to the failure of these trials, given the considerable intertumoral heterogeneity.Therefore, to find an effective drug in GBM patients, it may be necessary to select patients based on the molecular characteristics of the tumour.However, to date, only a few biomarker-driven trials have been performed using specific molecular markers of the parental tumour as an inclusion criterion.Moreover, in CNS tumours, the predictive significance of these mutations remains insufficiently studied [87].
Whole genome sequencing (WGS) is gaining prominence in oncological care.Compared to commonly used DNA diagnostic approaches, WGS offers a high sensitivity and accuracy, and all important mutation types may be consistently identified in a single assay [105].The importance of WGS is highlighted in a study wherein 62% of the patients with different metastatic cancers had mutations that could be used for further treatment or inclusion in trials after using WGS [106].
The first biomarker-driven trial was the IMPACT study, where patients with advanced metastatic cancer that received tumour genetic profiling and were evaluated for investigational therapy were included.In total, the overall response rate (ORR) was 27% for the patients treated with matched targeted therapy and 5% for patients with non-matched therapy, emphasising the potential of biomarker-driven studies [107].Hereafter, other biomarker-driven trials followed, such as the SHIVA trial, IMPACT-II and NCI-MATCH, to investigate the benefits of using genetic alterations to guide matched targeted therapies, aiming to enhance the ORR [108][109][110].
Given the absence of FDA approval for the use of all targeted drugs in all types of cancers, the TAPUR trial investigated the effects of FDA-approved drugs outside their approved indications in different cancer types based on pre-specified genomic targets [111].Encouragingly, most treatment arms have shown success, underscoring the potential of this approach [111].Similarly, the ongoing DRUP (Drug Rediscovery Protocol, NCT02925234) study, an ongoing biomarker-driven, multi-basket trial in the Netherlands, allocates patients to different treatments based on the mutational status of their tumours, achieving an overall clinical benefit rate of 33% for both rare and non-rare cancers [112].
These results highlight the importance of performing molecular tests to guarantee that all cancer patients have an equal chance of receiving the right treatment.
By identifying actionable mutations and enabling access to a broader spectrum of treatment options, studies like DRUP aim to offer renewed hope and extended survival prospects for patients with refractory GBM.

Conclusions and Future Perspectives
Although small molecule inhibitors have become part of the standard of care in various types of cancer, the overall results of clinical trials in GBM patients have been disappointing.The main factor contributing to this poor performance is the pervasive inter-and intratumoral heterogeneity characteristics of GBM.This tumour heterogeneity undermines the feasibility of a universal therapeutic approach, emphasising the inefficacy of a "one-size fits all" drug for these patients.Moreover, various aspects of the TME can counteract the effectiveness of small molecule inhibitors.Consequently, a need arises for further research directed towards identifying drug combinations that can target tumour clones that drive tumour relapse and are capable of modulating the complex dynamics of the TME.
In addition, more phase 0 trials will be needed to investigate drug penetration and target inhibition at the tumour site of GBM of small molecule inhibitors.Furthermore, future trials should focus on patient stratification based on molecular tumour markers to achieve the best possible results and identify drugs and drug combinations for certain patient subpopulations.

Table 1 .
Clinical trials involving mono-target small molecule inhibitors.

Table 3 .
Clinical trials investigating the combination of small molecule inhibitors with the standard of care.

Table 4 .
Clinical trials investigating combinations of small molecule inhibitors.