Polymeric micelles are prepared from spontaneously self-assembling amphiphilic block copolymers in aqueous medium. In general, these block copolymers consist of either two (hydrophilic-hydrophobic) or three (hydrophilic-hydrophobic-hydrophilic) segments. Assembly of the hydrophobic components of these copolymers creates a hydrophobic core that is separated from the aqueous environment by hydrophilic segments (Chen and Jiang, 2005). Hydrophobic interactions act as the principle driving force for micelle formation, but other intermolecular forces including hydrogen bonding, electrostatic interaction, and metal complexation have also been applied to increase stability (Harada and Kataoka, 1995; Kabanov et al., 1996, Kataoka et al., 1998; Nishiyama et al., 2003). Ionic copolymers may also be used, allowing for the formation of electrostatically stabilized polyion complex micelles (Harada and Kataoka, 1998, 2003; Kataoka et al., 1999) and polymer-metal complex micelles (Nishiyama et al., 2001, 2003).

Polyesters, poly(amino acids), and polyethers are commonly used as hydrophobic or ionic segments (Matsumura, 2008; Bae and Kataoka, 2009; Osada et al., 2009; Zhang et al., 2010; Gao et al., 2012; Tao et al., 2012). An advantage of polyesters, including poly(D,L-lactide) (PLA), and poly(glycolic acid) (PGA), and poly(ε-caprolactone) (PCL), is that they are FDA approved for biomedical applications owing to their biocompatibility and biodegradability (Gaucher et al., 2005). Poly(amino acids) such as poly(α,β-aspartic acid) (PAsp) and poly(L-lysine) (PLys) have also been extensively used to form the core of micelles via hydrophobic and electrostatic interactions (Lavasanifar et al., 2002; Matsumura et al., 2004). The variety of functional groups on poly(amino acids) facilitate numerous possibilities for drug-conjugation (Kataoka et al., 2001) as well as DNA incorporation into the core of micelles (Katayose and Kataoka, 1997).

The outer coating polymers are also essential components to stabilize the micellar structure in aqueous solution. Of these, polyethylene glycol (PEG) is most commonly used for the shell formation (Bae and Kataoka, 2009; Osada et al., 2009; Zhan et al., 2010; Shen et al., 2011; Gao et al., 2012). PEG is a non-toxic polymer with low immunogenicity that been widely used for prolonging the circulation time of drug-delivery systems (Verrecchia et al., 1995; Huang et al., 2012). It can prevent protein adsorption and minimize non-specific uptake of nanoparticles by the reticuloendothelial system in the body (Huang et al., 2012). Alternative biocompatible hydrophilic polymers to PEG include dextran, chitosan, and poly(ethylenimide) (PEI), which have also been used in the formation of the hydrophilic corona (Kwon and Kataoka, 1995; Qiu and Bae, 2007; Verma et al., 2012; Xie et al., 2012).

Other characteristics of polymeric micelles must also be considered when designing them for appropriate clinical use. Compared to surfactant micelles, polymeric micelles possess a lower critical micelle concentration (CMC), greater biocompatibility, and improved stability (Miller et al., 2012). The CMC of these systems is especially important to keep in mind. Below the CMC value, micelles begin to dissociate into monomers, decreasing the longevity of these particles in vivo. Especially in the context of targeting a brain tumor, these particles must remain intact for a sufficient amount of time in order to penetrate into and fully distribute within the tumor site. The CMC of a micelle depends to a significant extent on the copolymers used as well as conjugates incorporated within its structure, with values ranging from 10^-6 to 10^-7 M (Bae and Kataoka, 2009; Huang et al., 2012).

It is also critical to consider size and charge when designing micellar systems because the blood brain barrier (BBB) can regulate entry into the central nervous system (CNS) based on these characteristics. The BBB is comprised of tight junctions between endothelial cells, a number of transporters including efflux proteins (e.g. P-glycoprotein), as well as surrounding astrocytes that modulate endothelial function (Abbott and Romero, 1996). It acts to exclude many toxic and infectious agents, yet it remains a major barrier in the passage of chemotherapy agents into the CNS. The size of polymeric micelles typically ranges from about 10–100 nm depending on the composition and synthesis method (Huang et al., 2012). Particles even larger than this have been shown to penetrate the disrupted BBB and accumulate within tumor tissue by the enhanced permeability and retention (EPR) effect (Iyer et al., 2007). This is accomplished by diffusion of micelles out of circulation due to a tumor’s disrupted vasculature and accumulation within the tumor area due to poor drainage of interstitial fluid (Maeda, 2001). In terms of body clearance of these particles, micelles are large enough to prevent rapid renal clearance, which normally limits the effectiveness of smaller drug molecules less than 40 kDa (Greish et al., 2003).

Typically, the hydrophobic core of micelles serves as the site of therapeutic agent incorporation. As stated previously, these therapeutic agents include chemotherapy drugs, proteins, genes, and siRNA (Yoo et al., 2002; Oba et al., 2007; Heffernan and Murthy, 2009; Tong et al., 2010; Zhan et al., 2010, 2012a,b; Gao et al., 2012; Yin et al., 2013; Zheng et al., 2013) but can even include other nanoparticles such as superparamagnetic iron oxide (SPIO) particles (Nasongkla et al., 2006; Kessinger et al., 2011). The method of incorporation is dependent on the desired function and agent in question. For example, doxorubicin is incorporated non-covalently into the core of micelle in the SP1049C formulation used in clinical trials (Danson et al., 2004) whereas incorporation of cisplatin into NC-6004, another micelle formulation used in clinical trials, relies on the substation of Pt(II) atom from chloride to carboxylate in the side chain of poly(Glu). Nasongkla et al. (2006) was able to incorporate doxorubicin and SPIO nanoparticles into the core of micelles, both via non-covalent interactions. Genes can be incorporated into the core of micelles via electrostatic interaction with positively charge copolymer components such as PEI (Pathak et al., 2009; Zhan et al., 2012b). Micelles have even been coated with gold nanoparticles via shell-crosslinking, which could allow for the attachment of therapeutic or targeting agents to this surface of the particle in addition to loading within their core (Bae et al., 2006). Stimulus-sensitive linker molecules can also play a role in therapeutic agent incorporation but will be discussed later on in this review.

Genexol-PM is a micellar formulation composed of monomethoxy-PEG-block-poly(D,L-lactide) copolymer with paclitaxel loaded into its hydrophobic core (Gong et al., 2012). A major advantage of this formulation is that it does not contain Cremophor EL, a toxic surfactant normally used to solubilize paclitaxel in circulation in the clinical formulation of Taxol. There have been a number of Phase I and II clinical studies examining the use of Genexol-PM in solid tumor treatment. In a phase I clinical study, Kim et al. (2004) used Genexol-PM to treat patients with lung, colorectal, renal cell, breast, ovarian, and esophageal cancers. Out of 21 patients, 3 (14.3%) achieved partial responses and 6 (28.6%) maintained stable disease. Dose-limiting toxicities included neutropenia, sensory neuropathy, and myalgia (Kim et al., 2004). Lim et al. (2010) utilized a different dosing regimen (weekly delivery instead of once every 3 weeks) of Genexol-PM to treat patients with breast, head and neck, lung, and nasopharyngeal cancer. Out of 21 patients, 5 (23.8%) achieved partial responses and 9 (42.9%) maintained stable disease (Lim et al., 2010). A phase II clinical trial examining the efficacy of Genexol-PM in patients with metastatic breast cancer demonstrated that out of 41 patients, 5 (12.2%) achieved complete responses, 19 (46.3%) achieved partial responses, and 13 (31.7%) maintained stable disease (Lee et al., 2008). The median overall survival in this study was not reached, even with a median follow-up time of 17 months. It would have been interesting to see this formulation’s effect on brain metastasis in this disease context; however, one of the exclusion criteria in the study was CNS metastases (Lee et al., 2008). In another phase II clinical study, Kim et al. (2007) used a combination of Genexol-PM and cisplatin to treat advanced non-small cell lung cancer. Out of 69 patients, 26 (37.7%) achieved a partial response and 20 (29.0%) maintained stable disease. The study reported a median overall survival period of 21.7 months. Because this formulation lacks Cremophor EL, the authors pointed out that higher doses of paclitaxel could be achieved without an increase in toxicity (Kim et al., 2007). Genexol-PM was also tested in a phase II clinical trial against advanced pancreatic cancer (Saif et al., 2010). Median overall survival was 6.5 months for patients treated with a dose of 300 or 350 mg/m². For the 45 patients treated with this dose, 1 (2.2%) achieved a complete response, 2 (4.4%) achieved a partial response, and 24 (53.3%) maintained stable disease.

Genexol-PM is not the only clinical micelle formulation containing paclitaxel. NK105 is a core-shell micelle composed of PEG and poly(aspartic acid) modified with 4-phenyl-1-butanol to increase its hydrophobicity (Gong et al., 2012). In a preclinical study, Hamaguchi et al. (2005) investigated treatment efficacy in a colorectal cancer xenograft mouse model. NK105 exerted superior anti-tumor activity as compared to free paclitaxel in nude mice transplanted with HT-29 colon cancer cells, and a 25-fold higher tumor area under the curve (AUC) for NK105 was observed compared to free paclitaxel (Hamaguchi et al., 2005). In a phase I clinical study, Hamaguchi et al. (2007) looked at treatment with NK105 in pancreatic, bile duct, and colon cancer. Out of 19 patients, 6 (31.6%) patients were found to have stable disease, and a partial response was seen in 1 patient with metastatic pancreatic cancer and 1 patient with metastatic stomach cancer (10.5% in total). AUC and total clearance rate of NK105 at 150 mg/m² were ~32-fold larger and 72-fold lower, respectively, than for Genexol-PM at a dose of 300 mg/m², suggesting NK105 is more stable in circulation (Hamaguchi et al., 2007). NK105 was also used to treat patients with advanced or recurrent gastric cancer in a phase II clinical trial (Kato et al., 2012). Out of 56 patients evaluable for efficacy, 2 (3.6%) achieved complete responses, 12 (21.4%) achieved partial responses, and 17 (30.4%) maintained stable disease. Median overall survival was 14.4 months. A phase III clinical trial using NK105 to combat breast cancer was begun in July 2012, but no updates have been presented thus far (NCT01644890).

Besides paclitaxel, other drugs have also been incorporated into micelles for clinical purposes. NC-6004 (Nanoplatin^TM) is a polymeric micelle comprised of PEG and poly(glutamic acid) with incorporated cisplatin, which is normally cleared rapidly via renal excretion after systemic administration, leading to nephrotoxicity (Gong et al., 2012). NC-6004 led to significant anti-tumor effects in a mouse model of colon adenocarcinoma 26, human gastric cancer (MKN-45) bearing mice, and in HT29 oxaliplatin-resistant bearing mice (Nishiyama et al., 2003; Uchino et al., 2005; Alami et al., 2006). In a phase I clinical trial, NC-6004 was used in the treatment of several solid tumors including lung, colon, hepatic, pancreatic, renal, melanoma, and esophageal cancers (Plummer et al., 2011). Out of 17 patients, 7 (41.2%) achieved stable disease. Renal impairment was still observed at high doses of the NC-6004, but in general, toxicities were less severe and less frequent when compared with cisplatin.

An additional micelle platform for the delivery of a platinum-based compound is NC-4016, a formulation consisting of PEG and a coordinate complex of poly amino acid and 1,2-diaminocyclohexane platinum (II) (Gong et al., 2012). A phase I clinical trial was started in March 2009 (Gong et al., 2012), but no updates have been published as of yet.

Doxorubicin has also been encapsulated into micelles for use in patients with solid tumors. SP1049C, a polymeric micelle consisting of poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) block copolymer has been developed for this purpose (Gong et al., 2012). In a phase I clinical trial, SP1049C was used in the treatment of several solid tumors including colorectal, esophageal, lung, ovarian, kidney, and hepatic cancers in addition to soft-tissue sarcoma, mesothelioma, neuroblastoma, cholangiocarcinoma, and Ewing’s sarcoma (Danson et al., 2004). About 21 patients were evaluable for response, but no patients demonstrated complete or partial responses; 8 patients (31%) had stable disease. A phase II clinical study of SP1049C was conducted in patients with advanced adenocarcinoma of the esophagus and gastroesophageal junction (Valle et al., 2011). Out of 19 patients evaluable for response, 9 (47.4%) had partial responses and 8 (42.1%) had minor responses or stable disease. Median overall survival was 10 months, and neutropenia was found to be the principal toxicity of the compound. Another micelle formulation encapsulating doxorubicin is NK911, which consists of PEG-poly(Asp) block copolymers conjugated to Doxorubicin (Gong et al., 2012). In a phase I clinical study, NK911 was used to treat several other solid tumors including pancreatic, colorectal, esophageal, gall bladder, and stomach cancer as well as leiomyosarcoma (Matsumura et al., 2004). Out of 23 patients, 8 (34.7%) exhibited stable disease and 1 (4.3%) demonstrated a partial response. As with SP1049C, neutropenia was the primary hematologic toxicity.

Another micelle system for clinical use may be just on the horizon. Takahashi et al. (2013) reported the use of NC-6300, an epirubicin-incorporated micelle, in mice bearing human hepatocellular carcinoma xenograft tumors. At doses of 10 and 15 mg/kg, NC-6300 led to significant survival improvements when compared to both control and epirubicin treated mice. The micellar formulation also appeared to decrease cardiotoxicity normally caused by epirubicin (Takahashi et al., 2013).

Unfortunately, these previously mentioned micelle formulations have not been applied to patients with GBM nor has there been any substantial preclinical testing in appropriate glioma animal models. However, NK012, a micelle composed of PEG-poly(Glu) block copolymer with covalently bound SN-38, has shown promise in the treatment of malignant gliomas. SN-38 is an active metabolite of CPT-11 (Irinotecan), a topoisomerase I inhibitor (Hsiang and Liu, 1988; Kawato et al., 1991; Gong et al., 2012). Kuroda et al. (2009) compared NK012 versus CPT-11 treatment in a U87MG xenograft mouse model. In vitro studies demonstrated that NK012 was 34 to 444-fold more potent than CPT-11 as tested in five different human glioma cell lines. In vivo studies demonstrated that NK012-treated (30 mg/kg/day) mice lived for a significantly longer time period when compared to both control (p = 0.001) and CPT-11-treated (66.7 mg/kg/day; p = 0.0014) mice (Kuroda et al., 2009). This group further expanded upon the study by examining the efficacy of NK012 ± bevacizumab (Kuroda et al., 2010). NK012 monotherapy (30 mg/kg/day) led to greater survival improvements in mice bearing U87MG orthotopic intracranial tumors when compared to any dosing method of CPT-11 in combination with bevacizumab (40 or 66.7 mg/kg/day CPT-11 + 5 mg/kg/day bevacizumab; p < 0.05). No difference was observed between NK012 mice and those treated with NK012 and bevacizumab (Kuroda et al., 2010). In vivo bioluminescence studies of these experiments are displayed in Figure 1. In terms of clinical trials, a phase I clinical study of NK012 for treatment of colorectal, pancreatic, and esophageal cancers as well as small cell, carcinoid, and non-small cell lung cancers was conducted (Hamaguchi et al., 2010). Out of 23 patients that were evaluable for response, 2 patients (8.7%) had partial responses and 5 patients (21.7%) maintained stable disease. In another phase I trial, Burris et al. (2008) reported that out of 16 evaluable patients treated with NK012, 2 (12.5%) were reported to have partial responses and 10 (62.5%) maintained stable disease. As with the other micelle formulations however, no micelle-drug combinations have been used in clinical studies to target GBM or other forms of brain cancer.

The major releasing mechanisms that fall under this category are pH-sensitive and reduction-responsive release. It is known that the tumor microenvironment is slightly acidic with a pH of about 6.5–7.2 (Webb et al., 2011; Du et al., 2013). In addition to the microenvironment, acidic compartments within cells such as endosomes and lysosomes have a pH around 4.5–5.5 (Fehrenbacher and Jäättelä, 2005; Du et al., 2013). The low pH in such compartments can be a powerful tool for enabling extensive drug release, especially for particles that are taken up via the endocytic pathway. For example, micelles composed of pH-sensitive copolymer such as PEG-poly(L-histidine) or pH-labile hydrazone linkers are stable during the circulation in the blood; yet, after accumulation in the tumor, they can be dissembled into copolymers and release encapsulated drugs (Gaucher et al., 2005). Different pH-sensitive conjugation linkers have been developed for micelle systems including acid labile ortho esters (Tang et al., 2011), hydrazone bonds (Bae et al., 2003; Lee et al., 2012; Xiao et al., 2012), cis-aconityl bonds (Yoo et al., 2002), and acetal bonds (Gillies and Frechet, 2003). Yoo et al. (2002) compared doxorubicin release from micelles after using either a hydrazone or cis-aconityl bond for drug conjugation. For micelles possessing cis-aconityl bonds, less than 10% release was observed at pH 7 by 24 h whereas at a pH of 5, roughly 50% of the drug was released. For micelles possessing a hydrazone linkage for drug incorporation, about 30% drug release was observed by 16 days at pH 7 whereas close to 100% of doxorubicin was released at a pH of 5 (Yoo et al., 2002).

Another intrinsic release mechanism that has been investigated involves reduction-mediated release of therapeutic agents from disulfide-cross-linked micelles. Micelles possessing this crosslinking are more stable in circulation but upon internalized into a cell and exposure to high levels of glutathione in this environment, the stability of the system is disrupted, facilitating drug release (Heffernan and Murthy, 2009; Xu et al., 2009). Abdullah Al et al. (2011) described thiolated pluronic micelles with cores formed by disulfide bonds of functionalized Pluronic F127, a PEO-PPO-PEO triblock copolymer. At increasing concentrations of the reducing agent dithiothreitol (DTT), increasing paclitaxel release was observed. Heffernan and Murthy (2009) developed micelles composed of PEG-PLL block copolymer that were modified with cross-linkable dithiopyridine groups. These micelles were used to deliver proteins such as antigen ovalbumin and catalase as well as CpG-DNA. The proteins mentioned were modified with dithiopyridine moieties as well to allow for tethering to the core of micelles (Heffernan and Murthy, 2009). Xu et al. (2009) reported the synthesis of reduction-sensitive cross-linked micelles for triggered release of doxorubicin. When no DTT was present, only ~10% doxorubicin release was observed in cross-linked micelles at 10 h. However, at 10 mM DTT exposure, ~75% doxorubicin release was observed by 9 h (Xu et al., 2009).

Enzyme-mediated activation is also an attractive mechanism for therapeutic activation of micelles. Gu et al. (2013) developed a different strategy for glioma targeting by designing micelles modified with a cell penetrating peptide (low molecular weight protamine, LMWP) that could be specifically activated when MMP-2 and 9 were present in the environment. MMP-2 and MMP-9 are overexpressed by glioma cells as well as the tumor vasculature (Forsyth et al., 1999). The positive charges on the LMWP were masked by a polyanionic peptide with an MMP-2/9-cleavable peptide linker sequence (PLGLAG). This activatable LWMP (ALMWP) was conjugated to micelles, allowing for delivery of paclitaxel. In mice bearing intracranial C6 glioma tumors, ALWMP micelles carrying paclitaxel led to significantly longer survival times when compared to Taxol (p < 0.01) as well as LMWP-modified micelles (p < 0.05). In vivo fluorescence imaging after micelle administration and quantification of intratumor accumulation of paclitaxel is displayed in Figure 2.

Dual-responsive micelles that can release doxorubicin upon exposure to low pH and a reductive environment have already been investigated (Chen et al., 2013) and may allow for even more release specificity and increased stability of particles in circulation. These micelles, composed of PEG-SS-poly(2,4,6-trimethoxybenzylidene-pentaerythritol carbonate) block copolymer, only allowed for ~24.5% doxorubicin release by 21 h at pH 7.4. However, upon exposure to both 10 mM GSH and pH 5.0, 94.2% doxorubicin release could be achieved.

External triggers such as ultrasonography (US), heat, and light are alternative approaches for mediating micellar drug release (Marin et al., 2002; Rapoport, 2004). An important advantage of external triggers is that the drug release may be controlled locally and in a time-dependent manner, which can improve drug uptake into target tissues while minimizing systemic toxicity. One method used to externally mediate release is US, a non-invasive imaging technique. Yin et al. (2013) designed ultrasound-sensitive nanobubbles for the delivery of siRNA targeting sirtuin 2 (SIRT2), an anti-apoptotic gene. Nanobubbles were composed of a hetero-assembly of siRNA-loaded polymeric micelles and liposomes. Positively charged micelles were first loaded with siRNA, which were then loaded onto negatively charged gas-cored liposomes. Upon exposure to low-frequency US, siRNA-loaded micelles could be released from their electrostatic interaction with liposomes. In vivo experiments demonstrated that mice with subcutaneous C6 glioma tumors displayed both smaller tumor volume and improved survival when treated with this system in combination with low frequency US (Yin et al., 2013).

Local heating of a tumor can be achieved by various methods including continuous wave ultrasound (Rapoport, 2007) as well as other hyperthermia-inducing instruments. Thus, thermosenstive crosslinking may also increase the efficiency of drug delivery to a particular site. Several studies have already demonstrated thermosensitive micelles incorporating therapeutic agents (Soga et al., 2005; Bae et al., 2006; Yang et al., 2007; Prabaharan et al., 2009; Talelli et al., 2011; Shi et al., 2013). For example, Yang et al. (2007) described micelles composed of a novel thermosensitive poly(N-isopropylacrylamide-co-acrylamide)-b-poly(D,L-lactide) copolymer which was stable up to 41°C. Docetaxel release at 43°C was ~90% by 70 h versus ~50% release at 37°C (Yang et al., 2007). Micelles have even been combined with other particle types to achieve thermosensitive volume changes that could allow for transient release of incorporated agents (Bae et al., 2006).

Light-mediated release may also be a valuable mechanism for drug unloading. Goodwin et al. (2005) demonstrated that micelles with incorporated 2-diazo-1,2-naphthoquinones (DNQ) could undergo dissociation in response to ultraviolet and infrared light. Recently, diazonaphthoquinone-cored amphiphiles assembled from Janus-type poly(amido amine) dendrimers responding to near-infrared light were described (Sun et al., 2012). Release of doxorubicin nearly doubled after 10 min of exposure to 808 nm laser irradiation compared to non-irradiated particles.