Only one pre-clinical study concerns the use of liposome-loaded hydrogels in the treatment of GBM. Indeed, Arai et al. have combined a thermoreversible gelling polymer with liposomes encapsulating doxorubicin (49). A sustained release of doxorubicin was observed (9, 24, 64 and 94% after 8, 14, 30 and 54 days, respectively) compared to non-encapsulated doxorubicin entrapped in the hydrogel (78, 94 and 100% after 2, 4 and 12 days, respectively). Intratumorally injected in a subcutaneous human GBM murine model (U-87 MG cell line), this hydrogel combined with liposomes encapsulating doxorubicin inhibited tumor growth up to 38 days after treatment, compared to 14 days with the hydrogels containing the non-encapsulated drug, showing promising avenue to have a long-term antitumor activity.

Kim et al. studied the association between multifunctional nanoparticles combining antitumor effect and a real-time biodistribution tracking: CoFe₂O₄/SN-38 system, and a poly (organophosphazene) hydrogel (50). Hydrophobic interactions between the surface of the nanoparticles and the poly (organophosphazene) ethoxy L-isoleucine allowed the reticulation and the hydrogel formation at 37°C. In vitro, the release of CoFe₂O₄/SN-38 was observed over 59 days. On a subcutaneous human GBM model (U-87 MG cell line), no tumor growth was observed when intratumorally treated with the CoFe₂O₄/SN-38-loaded hydrogel (0.8 and 1.2 mg/mL) compared to the non-treated group and the groups treated with SN-38 alone, CoFe₂O₄ nanoparticle-loaded hydrogel and CoFe₂O₄/SN-38-loaded hydrogel (0.4 mg/mL). Using an orthotopic human GBM model (U-87 MG cell line), the treated zones were visible by MRI and the CoFe₂O₄/SN-38-loaded hydrogels (3.6 mg/mL) demonstrated a higher therapeutic effect compared to SN-38 in solution.

Meenach et al. (51) have developed iron oxide nanoparticles dispersed in a paclitaxel impregnated hydrogel. When the hydrogel is exposed to an external alternative magnetic field, it emits heat (41–45°C) very quickly and causes hyperthermia: another strategy used to treat the tumor cells with inorganic nanoparticles. This heat did not expand the hydrogel, suggesting that it will not increase the intracranial pressure. Paclitaxel in vitro release studies have shown a total release in 22 days. When GBM cells (M059K cell line) were exposed to heat (43°C) and paclitaxel (10 and 50 μM), the cell viability decreased. However, no synergistic effect between both elements was observed.

Zhao et al. have studied poly (lactic-co-glycolic acid) nanoparticles encapsulating paclitaxel combined with a hydrogel reticulated by photopolymerization (52). Adding the polymer nanoparticles in the initial solution did not prevent the photopolymerization. Paclitaxel release from the hydrogel in phosphate buffered saline (PBS) is fast up to 8 h (11 ± 2%) and lasts for at least 7 days (29 ± 4%). Total paclitaxel release is supposed to happen after 4 weeks. In vivo short-term (2 months) and long-term (4 months) studies showed that non-loaded hydrogel did not induce apoptosis or increase microglia activation. The activity of paclitaxel, loaded in the polymeric nanoparticles themselves dispersed in the hydrogel, has been evaluated on a human GBM (U-87 MG cell line) murine resection model. Fifty percent of mice treated with this hydrogel are still alive 150 days after tumor cell inoculation whereas the median survival for paclitaxel-loaded polymeric nanoparticles alone (without hydrogel) or hydrogel alone (without paclitaxel-loaded polymeric nanoparticle) treated mice was 72 or 63.5 days, respectively. In another study, Zhao et al. added TMZ in the matrix of the paclitaxel-loaded hydrogel (53). A synergic effect of both drugs was observed in vitro. Using the same tumor resection model, the median survival for mice treated with TMZ-loaded hydrogel (without paclitaxel-loaded polymeric nanoparticle) or paclitaxel-loaded hydrogel (without TMZ) was 75 or 90 days, respectively, whereas the median survival of mice treated with both TMZ and paclitaxel-loaded hydrogel was still not defined after 110 days of experiment.

A hydrogel without polymer, only based on lipid nanocapsules, has been recently developed and used as a local treatment against GBM (54–57). The hydrogel formation is only due to the association of lipid nanocapsules in suspension using an amphiphilic crosslinking agent: 4-(N)-lauroyl gemcitabine (GemC12), located at the oil/water interface of lipid nanocapsules, forming hydrogen bonds between the gemcitabine moieties and so inter-nanoparticular interactions (54). The advantage of this system is that hydrogel degradation corresponds to the release of GemC12-loaded lipid nanocapsules because no other component (natural or synthetic polymer, gelling agents, etc.) is present in the hydrogel, compared to conventional nanoparticle-loaded hydrogels (48). In artificial cerebrospinal fluid, 56 ± 9% of GemC12-loaded lipid nanocapsules are released in the first 48 h. The lipid nanocapsule release is then slower and sustained for at least a month (77 ± 8% after 30 days) (55). Another advantage of this system is that once all lipid nanocapsules are released corresponding to the complete dissolution of the hydrogel, no implant residue remains at the implantation site. In vivo, this hydrogel was well tolerated in the healthy mouse brain at short and long-term (2 and 6 months, respectively) (55, 56). In addition, when the hydrogel was injected in the resection cavity right after the GBM (U-87 MG cell line) resection, the median survival of mice treated increased significantly compared to the non-treated mice (62 and 35.5 days, respectively) and the apparition of recurrences was delayed (56). This antitumor activity was also observed on a rat GBM orthotopic resection model (57).

Regarding the preclinical studies described above, there is no doubt about the benefits that hydrogels and more particularly nanoparticle-loaded hydrogels can provide for the local post-resection therapy strategies against GBM. These systems can be tailored using natural matrix for the hydrogels, controlling their viscoelastic properties to be as close as possible to the brain characteristics, and the large expertise of the scientists makes nanoparticles more and more stable, better and better tolerated and can be developed on a large scale. A wide range of therapeutic molecules can thus be considered, and the combination of intrinsic properties of nanoparticles and drugs opens the way to new strategies. Nevertheless, the second limitation of Gliadel^® wafers is the lack of specificity of the treatment, and simple nanoparticle-loaded hydrogel also have this disadvantage. A common strategy used in nanovectorization is the combination of a targeting ligand on the surface of nanoparticles. The combination of these specific nanoparticles and the local administration inside a hydrogel seem quite easy to achieve but not yet tested. It is important here to review how or by which ligand to target GBM.

Several molecules are found overexpressed in GBM cells. Among these molecules are the chloride channels, particularly CIC-3. One of the most used molecules to target CIC-3 on the surface of GBM cells is chlorotoxin (CTX), a highly specific scorpion venom peptide. Surface functionalization of the nanoparticles with CTX has been used to increase the efficacy of therapeutic assets (63–65, 68, 69). For example, Yoo et al. (67) developed iron oxide nanoparticles as a vector of O6-methylguanine-DNA methyltransferase specific interfering RNA (siMGMT), one of the molecules responsible for tumor cell resistance to TMZ. These nanoparticles were functionalized on their surface with CTX (CTX-NP-siMGMT), which increased in vitro nanoparticle uptake by GBM cells (U-87 MG and T98G cell lines), compared to non-functionalized nanoparticles (NP-siMGMT). However, this absorption has not been tested on healthy cells. The addition of siMGMT promoted the suppression of expression and activity of O6-methylguanine-DNA methyltransferase in GBM cells (T98G cell line), leading to higher cell sensitization to TMZ. These effects were also observed when mice with orthotopic human GBM (T98G cell line) were intratumorally treated with CTX-NP-siMGMT. The tumor size was also reduced when mice were treated with CTX-NP, probably due to the anti-proliferative activity of CTX. Indeed, when binding to the chloride channels, CTX disturbs the chloride gradients essential for migration and the invasive character of GBM cells. In the same vein, Stephen et al. developed iron oxide nanoparticles functionalized with CTX to deliver 6-O-benzylguanine, a O6-methylguanine-DNA methyltransferase inhibitor, to GBM cells (after intratumoral administration in GBM orthotopic model) (66).

CD44 is a receptor located on the surface of cell membranes and is involved in cell-to-cell interactions (103). The expression of this glycoprotein is greater in several cancers, including GBM, compared to healthy tissues (104–107). In addition, studies have shown that this receptor is involved in the invasion of tumor cells (108, 109). Hyaluronic acid (HA), a major component of extracellular matrix, is a natural ligand of CD44 used to target cancer cells overexpressing the receptor thus improving its specificity and efficacy (110–113). Hayward et al. (70) have shown that surface-conjugated liposomes with HA specifically target human GBM cells (A172 and U-87 MG cell lines) compared to primary astrocytes and rat microglia cells. In addition, this system has improved the therapeutic efficacy of doxorubicin encapsulated in liposomes, while decreasing its absorption by astrocytes and microglia cells. Cohen et al. (71) also showed that lipid nanoparticles functionalized with HA could bind to GBM cells (U-87 MG cell line) but also to neurospheres of GBM cells from patients. These lipid nanoparticles charged with an interfering RNA directed against Polo-Like Kinase 1 (siPLK1), a kinase involved in cell cycle regulation, significantly reduced the expression of PLK1 mRNA in U-87 MG cells compared to lipid nanoparticles charged with siPLK1 but without HA on their surface. In addition, this expression decrease was correlated with higher cell toxicity. Finally, in vivo studies using an orthotopic human GBM model (U-87 MG cell line) showed that intratumoral treatment with lipid nanoparticles combined with HA and siPLK1 increased the mouse median survival compared to saline and lipid nanoparticles combined with HA and siLuciferase (not determined after 95 days compared to 33 and 34.5 days, respectively).

Epidermal growth factor receptor (EGFR) is part of the tyrosine kinase receptor family. This transmembrane glycoprotein is found overexpressed in about 40% of primary GBM (114, 115), and is involved in cell proliferation and survival (116, 117). The most common mutation of the EGFR gene is variant III (EGFRvIII), constitutively activated, and accounts for up to 60% of EGFR amplifications in primary GBM (118). This characteristic makes it a preferential target (119), and many laboratories developed nanomedicines using anti-EGFR and anti-EGFRvIII antibodies in combination with many types of nanoparticles to preferentially target these receptors. For example, Jamali et al. (72) conjugated a monoclonal antibody anti-EGFRvIII to PLGA nanoparticles encapsulating curcumin used as a photosensitizer for the development of photodynamic therapy on GBM cells. Others also used cetuximab, an antibody recognizing EGFR and EGFRvIII and inhibiting their action, or other types of ligands, such as an EGF peptide, in combination with iron oxide (73–75), gold (78), polymeric (77), solid lipid (76) and human ferritin nanoparticles (79).

Other molecules are also found deregulated in GBM cells such as the integrin αvβ3 receptor, the α2 receptor of IL-13 or the fibroblast growth factor-inducible 14 (Fn14). The integrin αvβ3 receptor is a cell adhesion molecule that plays a role in cell propagation, migration, survival, proliferation, and differentiation (120). It is found overexpressed in glioma cells and newly formed vessels, thus also an interesting target for the development of nanomedicines (121). The cyclic peptide RGD was used in combination with several types of nanoparticles (human ferritin ones or mesoporous silica) to target this receptor in the treatment of GBM (79, 80). In another study, Jiang et al. (81) functionalized selenium nanoparticles with polysaccharides from the Gracilaria lemaneiformis alga which have a strong affinity for the integrin αvβ3 receptor. Interleukin 13 (IL-13) is a cytokine involved in the regulation of immune responses and microenvironment. In most cells, IL-13 binds with low affinity to the α1 receptor (IL-13Rα1) which then pairs to the α receptor of interleukin 4 to form a heterodimer. This complex then enables the activation of STAT6 signaling pathways. In some healthy cells and in cancer cells, IL-13 can bind with strong affinity to the α2 receptor (IL-13Rα2). IL-13Rα2 is a decoy receptor that sequesters IL-13, allowing tumor cells to escape apoptosis (122). This receptor is found overexpressed in 75% of GBM (123). Kim et al. (82) grafted doxorubicin-loaded nanoparticles to the T lymphocyte surface with a mutant version of IL-13. Wang et al. (83) functionalized the surface of PLGA nanoparticles with the Pep-1 peptide, specifically recognizing IL-13Rα2. Fn14 is a member of the tumor necrosis factor receptor family. Its expression level in the healthy brain is low whereas it is found more important in 70 to 85% of GBM (124). Moreover, the overexpression of Fn14 was found in both primary tumor and tumor cells infiltrated in healthy tissue (124, 125), whether newly diagnosed or recurrent (126, 127). These data make Fn14 an optimal surface molecule for the development of targeting therapy using nanoparticles associated with anti-Fn14 antibodies for the treatment of GBM (84, 85).

The GBM cancer stem cells (CSC), resistant to radio and chemotherapy are tumor-initiating cells, also partly responsible for the recurrences observed in patients after the implementation of standard treatments (128). Therefore, the targeting and eradication of CSC could lead to a better management of GBM. GBM CSC can express biomarkers associated with neural stem cells (129–132), and these biomarkers have been used as targets in several studies to improve drug delivery (87–90).

Among the markers that can be associated with CSC, CD133 is one of the most studied in brain tumors (133). This membrane glycoprotein plays a role in cell differentiation and the epithelial-mesenchymal transition. In the 2000's, studies have shown that cells with the marker CD133, isolated from human brain tumors, can reproduce the original tumor in immunocompromised mice (134, 135). Furthermore, recurrences of GBM often have a higher percentage of CD133-positive cells compared to tumor cells before treatment. A large proportion of CD133-positive cells is correlated with lower patient survival (136). These cells with a high capacity to form tumors have therefore become targets for the treatment of GBM (86, 87). However, not all GBM CSC express CD133. Subsequent studies have shown that tumors can successfully develop from GBM CSC without the CD133 marker in xenograft models (137, 138). Therefore, the identification of GBM CSC cannot be based solely on the expression of CD133. CD133-positive cells often co-express nestin, a protein from one of the cytoskeletal components (intermediate filaments). The expression of nestin is found in neural and progenitor stem cells but also in several types of cancer, including GBM (134, 135, 139–141). Increased expression of nestin is associated with high-grade gliomas and a low patient survival rate (132, 141).

To compensate for this low survival, several teams were interested in the treatment of GBM CSC. For example, Prabhu et al. (88) developed TMZ-charged iron oxide nanoparticles functionalized with an anti-nestin antibody to fight GBM CSC while sparing healthy tissue. In addition, Gonçalves et al. (89, 90) developed gold nanorods (AuNR) combined with a peptide specifically recognizing nestin (NesPEG-AuNR). These AuNR generate heat when irradiated by a laser emitting in the near infrared and cause localized cellular damage. In vitro internalization studies using cells expressing or not nestin, cultured in single layer (2D culture) or spheroids (3D culture) showed that NesPEG-AuNR were mainly internalized by nestin-positive cells and not by nestin-negative cells. Internalization of AuNR involves energy-dependent mechanisms, including endocytosis mediated by caveolin. Photothermal treatments of NesPEG-AuNR resulted in selective elimination of nestin-positive cells by cell apoptosis, while nestin-negative cells remained viable. The results also indicated that in the presence of AuNR, nestin-positive cells as spheroids are more resistant to photothermal treatments than when they are cultured in a monolayer, indicating that the 3D model is closer to in vivo models than the 2D model.

Tumor-associated myeloid cells (TAMCs) are a heterogeneous population of myeloid cells from hematopoietic precursors. They include tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs). TAMCs are massively recruited at the GBM level, reaching 30 to 50% of the tumor mass (142–144). These cells are the main cause of immunosuppression in GBM (60, 145, 146). They can strongly inhibit innate and adaptive immunity (147–149). They suppress effective immune cell function by several pathways, especially by depriving lymphocytes from their essential nutriments, generating oxidative stress and triggering the recruitment of regulatory T cells (147). As a result, TAMCs have recently been recognized as an attractive therapeutic target to decrease immunosuppression in GBM in the hope of maximizing the effectiveness of anti-tumor therapies.

TAMCs have recently been shown to express the programmed death-ligand 1 (PD-L1) more strongly than other immune cells or tumor cells (150, 151). This advantage was used by Zhang et al. (91) to develop surface functionalized lipid nanoparticles with anti-PD-L1 antibody. Ex vivo, the nanoparticles combined with the antibody were able to target TAMCs, to be internalized and to accumulate at the level of lysosomes. These functionalized nanoparticles encapsulating dinaciclib, a cyclin-5-dependent kinase inhibitor, reduced the viability of TAMCs in a dose-dependent manner. In addition, they reduced the recycling of PD-L1 on the surface of TAMCs by directing the ligand to the lysosome. In an orthotopic in vivo model of murine GBM (GL261 cell line), 24 h after injection at different locations in the brain, the nanoparticles functionalized with the anti-PD-L1 antibody encapsulating dinaciclib were retained in the tumor and more particularly, co-located with TAMCs. In addition, the median survival was significantly higher in mice treated with radiotherapy combined with these nanoparticles compared to mice treated with saline solution, radiotherapy alone and nanoparticles alone. These results were also observed when mice were intranasally treated. Finally, these nanoparticles were able to target TAMCs from GBM patients, validating the relevance of this approach on a human model.

Monocytes recruited in cancer tissues differentiate into macrophages that can be activated into two different phenotypes (type 1 or 2) in response to signals from their microenvironment. Type 1 macrophages (M1) coordinate the development of an adverse inflammatory microenvironment for cancer cells and play a central role in initiating and maintaining antitumoral immunity. On the contrary, type 2 macrophages (M2) suppress anti-tumor immunity and coordinate the remodeling of the microenvironment, making it favorable to survival, growth and tumor progression. Many studies suggest that TAMs are primarily M2-type in GBM (152, 153). Strategies to reduce the number of TAMs in the tumor or modulate their phenotype have shown strong potential for the treatment of GBM (154, 155).

Li et al. used the intertwined relationships between TAMs and GBM cells to modify the M2 phenotype of TAMs (92). TAMs with an M2 phenotype, loaded with nanodiamonds combined with doxorubicin (Nano-DOX) were able to deliver the Nano-DOX to GBM cells in vitro and in vivo, causing damage to these cells. The altered GBM cells emitted molecular patterns associated with this damage, modifying the M2 phenotype of the TAMs to a M1 phenotype and thus reducing the tumor size of a human orthotopic GBM model (U-87 MG cell line). Chemokine ligand 2 (CCL2) is involved in differentiation and survival of TAMs. Cho et al. used a CCL2 inhibitor (NOX-E36) to suppress their recruitment and study the effect of combined therapy with bevacizumab, an antibody directed against the vascular endothelium growth factor (93). Inhibition of CCL2 blocked macrophage recruitment and angiogenesis, resulting in decreased tumor and blood volumes in an orthotopic human GBM model (U-87 MG cell line) expressing CCL2. In addition, the median survival of mice treated with NOX-E36 combined with bevacizumab is greater than that of mice treated with bevacizumab alone (32 and 22 days, respectively). This study shows that CCL2 inhibition can play an important role in increasing the effectiveness of anti-angiogenic treatment in GBM by inhibiting the recruitment of TAMs.

The extracellular matrix (ECM) consists of a complex network of fibronectins, collagens, chondroitins, laminins, glycoproteins, heparin sulfate, tenascins and proteoglycans (156). These molecules play a role in migration, differentiation and inflammation but they also participate in the process of invasion and metastasis of malignant cells in the host tissue (157). Some of these molecules are found at high levels in ECM of tumor tissue. This feature has been used to target and improve drug delivery in the GBM. In this next section, only the most studied proteins for the treatment of GBM in combination with NP have been described.

ECM is characterized by the presence of matrix metalloproteinases (MMP), enzymes part of the gelatinase family, which play a key role in tumor progression and metastasis. In GBM, MMP-2 and MMP-9 are overexpressed. They degrade ECM and contribute to the angiogenic and invasive potential of glioma cells. These characteristics make the MMP attractive targets for the development of GBM therapy (63, 158). Agarwal et al. (63) developed morusin (MOR)-loaded PLGA nanoparticles, a naturally occurring chemotherapy active ingredient, and surface-conjugated with CTX (PLGA-MOR-CTX). The anti-cancer potential of PLGA-MOR-CTX nanoparticles was evaluated in vitro in human GBM cells (U-87 MG and GI-1 cell lines). PLGA-MOR-CTX nanoparticle treatment resulted in nanoparticle accumulation in GBM cells, mediated by MMP-2 targeting. In addition, important cytotoxicity parameters such as reactive oxygen species generation, increased caspase activity, cytoskeletal destabilization, and inhibition of MMP-2 activity were observed in GBM cells following treatment with PLGA–MOR–CTX nanoparticles. These results, combined with the non-toxicity of healthy human neuronal cells (HCN-1A cell line), highlight the specific therapeutic potential of this strategy for the treatment of GBM. Other teams have also targeted MMP-2 to deliver therapeutic nanoparticles in GBM (68). For example, Locatelli et al. (69) formulated multifunctional nanocomposites consisting of polymeric nanoparticle containing two cytotoxic agents: alisertib, an inhibitor of Aurora A kinase, and silver nanoparticles conjugated with CTX. In addition, Bruun et al. (94) used lipid nanoparticles loaded with a siRNA and functionalized with angiopep-2 and a lipopeptide capable of targeting MMP to target ECM.

Tenascin C (TNC) is a glycoprotein of ECM overexpressed during normal tissue repair and in many malignant tumors. It plays an important role in tumor progression including angiogenesis, proliferation and cell migration, making it an attractive target for GBM therapy (159–161). Doxorubicin-loaded liposomes, modified with sulfatid known to bind to TNC, have been used to improve efficacy and reduce the side effects of free doxorubicin (95). Biodistribution, therapeutic efficacy and systemic toxicity of the liposomes were evaluated in an in vivo human GBM xenograft model (U-118 MG cell line). The median survival was greater in GBM-bearing mice treated with liposomes compared to free doxorubicin and saline (93, 61 and 45 days, respectively). Kang et al. adopted a different strategy by targeting both TNC and neuropilin-1 (NRP-1), a transmembrane protein overexpressed in newly formed tumor cells and blood vessels (96). The FHK peptide targeting TNC was coupled with the tLyp-1 peptide known to increase penetration of nanosystems into tumor cells via NRP-1 (97–99). PLA nanoparticles loaded with paclitaxel were functionalized with this FHK/tLyp-1 peptide (FHK/tLyp-1-NP-PTX). In vitro, the functionalization of the nanoparticles allowed their internalization in U-87 MG and HUVEC cells (two-dimensional culture), but also their deep penetration into GBM spheroids. In addition, in vivo, real-time imagery showed an accumulation of functionalized nanoparticles in GBM using an orthotopic mouse model (U-87 MG cell line). It has been shown that this accumulation is mediated by TNC and that transport through cancer cells is governed by NRP-1. Finally, the median survival of the mice was greater when treated with FHK/tLyp-1-NP-PTX compared to FHK-NP-PTX, tLyp-1-NP-PTX, NP-PTX and PTX (59 compared to 37.5, 33.5, 25.5 and 16 days, respectively). These results suggest that the combination of the two peptides provides an interesting tool both to target GBM and to treat it using a synergistic mechanism. The targeting of TNC or one of its isoforms in combination or not with other molecules to improve drug delivery was also explored (100, 101).

Fibronectin (FN) is a glycoprotein that plays a role in cell adhesion, migration, growth and differentiation. This protein can bind to several molecules including fibrin. Fibrin-FN complexes play a role in coagulation (162). The significant presence of fibrin-FN complexes is found in many invasive tumors, including GBM (163). These complexes play an important role in survival, the proliferation and invasion of cancer cells, making it another attractive target for the treatment of GBM (164). Wang et al. (83) functionalized PLGA nanoparticle surface with the Pep-1 and CREKA peptides (Pep-1/CREKA-NP). Pep-1 peptide can pass through the BBB and penetrate GBM cells by targeting the IL-13Rα2, overexpressed on the plasma membrane of GBM cells (123). CREKA peptide works as an anchor by binding to fibrin-FN complexes in the ECM of tumor cells (102). The functionalization of the nanoparticles with both CREKA and Pep-1 increased the nanoparticle retention in the GBM tissue (U-87 MG cell line) and the distribution of the therapeutic agent in cancer cells. In fact, Pep-1/CREKA-NP penetrated deeper into GBM compared to nanoparticles combined with any of the peptides (Pep-1-NP or CREKA-NP). In addition, the median survival of mice with orthotopic human GBM (U-87 MG cell line) treated with Pep-1/CREKA-NP was higher than those treated with Pep-1-NP, CREKA-NP, NP-PTX, PTX and a saline solution (61 compared to 53, 55, 47, 43 and 36 days, respectively).