Brain tumors, especially malignant gliomas, belong to the most aggressive human cancers. Despite numerous advances in neurosurgical operative techniques, adjuvant chemotherapy and radiotherapy, the prognosis for patients remains very unfavorable.1, 2, 3 Even the chemotherapeutic drugs most effective in glioblastoma, such as nitrosoureas, platinum compounds and temozolomide, increase the survival time of patients only marginally.4, 5
Features responsible for the aggressive character of glioma include rapid proliferation, diffuse growth and invasion into distant brain areas in addition to extensive cerebral edema and high levels of angiogenesis. Glioblastomas develop a distinct neovasculature that is highly permeable to macromolecules and small particles. However, disruption of the blood-brain barrier (BBB) remains a local event, which is evident in the tumor core but absent at its growing margins. Accordingly, therapeutic drug levels have been found in necrotic tumor areas, while in peritumoral regions the drug levels were markedly lower or nondetectable.6 Confirming the importance of the BBB, improved survival of patients with malignant gliomas was reported when systemic chemotherapy was applied after hyperosmotic or chemical disruption of the BBB.7
Therefore, the development of a strategy allowing drug delivery across the BBB is of prime importance and would offer a possibility to use highly active antitumor agents that are usually not employable in the CNS due to effective entrance block by the BBB. The paramount importance of a targeted drug delivery was recently acknowledged by the NIH Brain Tumor Progress Review Group.8 The feasibility of this approach to CNS chemotherapy was demonstrated by a number of authors who showed that brain delivery of anticancer drugs such as the anthracyclines could be achieved using sterically stabilized colloidal carriers such as liposomes or solid lipid nanoparticles.9, 10, 11, 12, 13, 14
One of the most likely candidates for CNS chemotherapy, doxorubicin, has proven to be effective in glioblastoma cell lines15 but lacks the ability to cross the BBB, being a P-glycoprotein substrate. However, a significant increase in survival rate was achieved in patients with malignant gliomas treated with intratumoral injections of doxorubicin,16 while clinical trials demonstrated that after i.v. injection, doxorubicin did not reach cytotoxic levels in glioma tissue due to delivery problems.17
Some types of liposomes are capable of crossing the BBB.18, 19, 20 However, this capacity is dependent on their size and structure.18 Whereas the large vesicles, which enter the brain only via breaches of the BBB, are retained in glial tumors, small vesicles, which may cross the BBB unrestrictedly, are not specifically enriched in tumors and confer a risk of uncontrolled embolization.18, 19 In addition, the application of liposome-bound doxorubicin enabled only short-term relief to glioblastoma patients.20 Therefore, the search for more efficient carriers to transport anticancer drugs to the brain must continue.
We have recently demonstrated that polysorbate 80-coated poly(butyl cyanoacrylate) nanoparticles (NPs) enabled an improved distribution of doxorubicin21 as well as of loperamide,22 tubocurarine,23 the hexapeptide dalargin24 and the NMDA receptor antagonist MRZ 2/57625 into the brain after i.v. administration. Under these conditions, very high doxorubicin concentrations of 5 μg/g could be obtained in the brain following i.v. injection of doxorubicin bound to polysorbate 80-coated nanoparticles, whereas all control preparations, including doxorubicin in saline, doxorubicin in saline plus polysorbate 80 and doxorubicin bound to nanoparticles without polysorbate coating, did not achieve any detectable brain levels (detection level, 0.1 μg/g).21 Earlier studies also revealed enhanced permeability and retention effects that could augment the nanoparticle-assisted doxorubicin transport to the brain.26, 27, 28, 29
In consequence, the objective of the present study is the evaluation of doxorubicin-loaded nanoparticles for the chemotherapeutic treatment of glioblastomas. The employed animal system, the 101/8 glioblastoma in rats, is similar to human glioblastomas with respect to morphology and biologic behavior such as development of diffuse spread, vascular proliferation and necrosis. In addition, no peripheral metastases were observed. This tumor was intracerebrally implanted into Wistar rats and then treated with different doxorubicin formulations. A statistically significant increase in survival time of the glioblastoma-bearing rats treated with doxorubicin bound to polysorbate-coated nanoparticles compared to the controls and to doxorubicin solution was observed. More than 20% of the animals showed a long-term remission.
DAB, diaminobenzidine; DOX, doxorubicin; GFAP, glial fibrillary acidic protein; IST, increase of survival time; LDL, low-density lipoproteins; NMDA, N-methyl-D-aspartate; NP, nanoparticle; PAS, periodic acid Schiff's reagent; PS, polysorbate 80; TUNEL, terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling.
MATERIAL AND METHODS
Preparation and characterization of nanoparticulate formulation
Butylcyanoacrylate (BCA) was provided by Sichel-Werke (Hannover, Germany); doxorubicin was a generous gift from Sicor (Rho, Italy). All other reagents were obtained as follows: dextran 70,000, mannitol and Evans Blue from Sigma (Deisenhofen, Germany); NaOH, HCl and hematoxylin from Merck (Darmstadt, Germany); polysorbate 80 (Tween 80) from ICI Chemicals (Essen, Germany); pentobarbital from Merial (Hellbergmoos, Germany).
Particles were manufactured by anionic polymerization.21 One percent of butyl-cyanoacrylate was added to a 1% dextran solution in 0.001 N HCl under constant stirring. After 40 min, doxorubicin was added to the mixture to obtain a final doxorubicin concentration of 0.25%. After 2.5 hr, the mixture was neutralized with NaOH, filtered through a G1 glass filter (Schott, Mainz, Germany) and freeze-dried after addition of 3% mannitol as a cryoprotector. The particle size was measured by photon correlation spectroscopy (Brookhaven Instruments, Holtsville, CA); an average diameter of 270 ± 20 nm was found. The amount of free drug in the preparation was assessed spectrophotometrically (U-3000, Hitachi, Tokyo, Japan) after particle separation by centrifugation (150,000g for 45 min). It was shown that 70% of the drug in suspension was associated with NPs. In order to obtain polysorbate 80-coated particles (DOX-NP+PS), 1% polysorbate 80 (PS) was added to the suspension, and the resulting preparation was incubated for 30 min under stirring and used within 2 hr. Empty NPs were synthesized using the same technique with an average diameter of 250 ± 30 nm.
Tumor model system
For this study, we used an experimental system based on intracranially implanted 101/8 glioblastoma in rat. This tumor was initially produced by local injection of an α-dimethylbenzanthracene (DMBA) pellet into a Wistar rat cerebellum and maintained by continuous passages by intracerebral implantation.30 The 101/8 glioblastoma was extensively employed in a variety of studies, especially for the investigation of biology and pathology of glia and glial tumors, as well as for experimental chemotherapy of brain malignancies.16, 30 For long-term storage, the tumor tissue was kept at −196°C and was propagated by injection into the brains of rats. The tumor has a stable monomorphous structure and shows the characteristic histologic picture of aggressive glioblastoma with fast diffuse growth in the brain parenchyma and a rather low tendency toward necrosis. The transplantability of the tumor was about 100%, yielding a predictable symptom-free life span after inoculation. The transplantation of 101/8 glioblastoma in the present study was performed using fresh tumor tissue. This technique was chosen to preserve the major typical features of the parent tumor, especially its antigenic structure and differentiation.
Animals and tumor inoculation
One hundred fifty-one adult male Wistar rats weighing 200–250 g (Charles River, Wilmington, MA, and Stolbovaya, Moscow, Russia) were acclimatized for 1 week and caged in groups of 5. They were fed ad libitum with standard laboratory food and water. For tumor implantation, animals were deeply anesthetized by intraperitoneal injections of pentobarbital (50 mg/kg) and then placed in a stereotactic device (Leitz, Wetzlar, Germany). Through a midline sagittal incision, a burr hole of 1.5 mm in diameter was made with a dental drill at a point 2 mm posterior to the right coronal suture and 2 mm lateral to the sagittal midline. Tumor material (approximately 106 cells) from the frozen stock was introduced into a tuberculin syringe (B. Braun, Melsungen, Germany) linked to a 21 gauge needle. The tip was placed 4 mm below the bone surface and the tumor tissue was injected into the bottom of the right lateral ventricle. The scalp incision was sewn or closed with glue (Turbo 2000 Kleber Universal, Boldt, Wermelskirchen, Germany). After development of pronounced clinical signs of the disease (usually day 14), the animals were sacrificed by carbon dioxide asphyxiation, then decapitated. The brain was immediately removed. The tumor was excised and chopped with a scalpel; a tumor implant (5 mg) was inoculated into the brain of new experimental animals as described above. The appropriate coordinates were confirmed and the technique was refined by repeated pilot experiments.
The animal experiments were performed in accordance to the Russian Guidelines for Animal Experiments and authorized by the Russian Ministry of Health (1045-73 and 52-F3-24.04.95) or the German Tierschutzgesetz and the Allgemeine Verwaltungsvorschrift zur Durchführung des Tierschutzgesetzes and were authorized by the Regierungspräsident Darmstadt (II 25.3-19 c 20/15-F 31/10).
Assessment of BBB disruption
In order to ascertain proper surgical procedures, the BBB integrity after tumor implantation was assessed by i.v. injections of Evans Blue solution (2%, 2 ml/kg) 30 min after implantation.
Tumor-bearing animals were randomly divided into 6 groups (n = 5–10) and received one of the following formulations: untreated control, blank NP coated with polysorbate 80 (NP+PS), doxorubicin in saline (DOX), doxorubicin in 1% polysorbate solution (DOX+PS), doxorubicin bound to NP (DOX-NP), or doxorubicin bound to NP coated with polysorbate 80 (DOX-NP+PS). These preparations were injected i.v. into the tail vein using the following dose regimens: 2.5 mg/kg and 1.5 mg/kg on days 2, 5 and 8. The experiments with the dosage of 3 × 1.5 mg/kg were run in triplicates (2 independent runs in Moscow and another in Frankfurt) and with the dosage of 2.5 mg/kg in duplicates (2 in Moscow). Since the outcome was statistically comparable in all 3 or in the 2 runs, respectively, the data were pooled for the 3 or 2 respective runs. Antitumor efficacy was estimated by increase of median survival time (IST; Table I) as compared to control (ISTC, %) and to DOX (ISTD, %). Results are also shown by Kaplan-Meier plots (Figs. 1 and 2).
Table I. Statistical Analysis of Antitumor Efficacy of Doxorubicin Bound to Nanoparticles Coated with Polysorbate 80 in Rats with Intracranially Implanted 101/8 Glioblastoma and Long-term Survivors
Explorative meta-analysis was performed by means of pairwise log-rank tests.31 Significance levels for individual tests were adjusted according to Holm32 in order to keep a multiple significance level of 5% per dose regimen. Evaluated data are displayed graphically using Kaplan-Meier plots and summarized descriptively by means of median survival times with their respective 95% confidence intervals. Statistical analysis used SAS/STAT procedure Lifetest.33 Evaluation was based on pooled data from 3 experiments (dose regimen, 3 × 1.5 mg/kg) and pooled data from 2 experiments (dose regimen, 3 × 2.5 mg/kg).
For the histologic evaluation of tumor growth, the 101/8 glioblastoma was implanted in 14 rats as described above. These and control animals that did not receive any treatment were randomized into 5 groups (n = 2 or 3) that were sacrificed by carbon dioxide asphyxiation on days 2, 3, 5, 8 or 12 after tumor implantation.
For a preliminary histologic evaluation of the results of chemotherapy, the 101/8 glioblastoma was implanted into 12 rats. These animals were randomized into 4 groups (n = 3) that were treated with DOX formulations in the dose of 3 × 1.5 mg/kg as described above. On day 12, all animals were sacrificed. In addition, 2 long-term surviving animals in the group treated with 3 × 1.5 mg/kg DOX-NP+PS without clinical signs of tumor growth were sacrificed on day 180 and studied by histology. Additionally, for pathologic evaluation of side effects, 6 healthy animals were injected with DOX and DOX-NP (3 × 2.5 mg/kg; 3 animals in each group).
Gross pathology and tissue preparation
Randomly selected animals (at least one of each group) underwent necropsy of the whole body. Autopsy of the whole body, including the brain, was also performed in all healthy animals treated with DOX formulations. In all animals, gross pathology of the brain was evaluated. Consecutively, the whole brains were fixed in phosphate-buffered formalin (4%, pH 7.4) for 4–48 hr, then cut into 2 mm thick frontal slices, dehydrated and embedded in paraffin. Lungs, kidneys, spleens, livers and hearts were processed similarly.
Histology and immunocytochemical analysis
Serial sections (5 μm thick) from paraffin-embedded tissues were cut and processed for staining with hematoxylin and eosin (H&E), periodic acid Schiff's reagent (PAS) and Giemsa (Merck) and for immunocytochemistry. Comparative histology and immunocytochemistry was performed using a paraffin tissue array system of 2 mm cylindrical samples containing relevant areas cut from previously evaluated standard-size paraffin blocks of whole brains and other organs. Immunostaining was performed using the indirect avidin-biotin-peroxidase complex method. Primary antibodies were applied at a concentration of 5–25 μg/ml for 1–24 hr. These were polyclonal glial fibrillary acidic protein (GFAP; Dako, Glostrup, Denmark), an astrocyte marker, monoclonal actin-binding cytoskeletal protein ezrin (Sigma)34 and the monoclonal proliferation marker MIB-1 (Dako). MIB-1 required pretreatment by microwave at 600 W for 2 × 5 min in 10 mM sodium citrate (pH 6.2). Biotinylated preadsorbed secondary antibodies, antimouse from sheep (NEN-Amersham, Arlington Height, IL) and antirabbit from swine (Dako) were used in a concentration of 10 μg/ml with an indirect peroxidase system (ABC, Vector, Vector Laboratories, Burlingame, CA). Diaminobenzidine (DAB; Dako) served as a chromogene providing a brown reaction product. Counterstaining was performed with hematoxylin (Harris) 7 g/l (Merck). For blocking of nonspecific activity, nonfat dry milk (1–10%; Nestlé, Frankfurt, Germany) was used. To block the endogenous peroxidase, the sections were immersed for 10–20 min in PBS containing 1% hydrogen peroxide. Cell death (apoptosis) was established using a modified terminal dUTP nick end labeling (TUNEL) staining method with an antidigoxigenin-peroxidase system (Oncor, Boehringer Mannheim, Mannheim, Germany) following the recommendations of the manufacturer.
For evaluation of organ toxicity (Table II), tumor size, histology and immunocytochemistry, a semiquantitative scoring system was used, and the tumor size was estimated on frontal sections of the whole brains as follows: 0, no detectable tumor; 1, up to 1 light microscopy visual field at a magnification of × 20; 2, up to 2 visual fields at the same magnification; 3, up to 3 visual fields; 4, up to 4 visual fields; 5, above 4 visual fields at 20 × magnification. For sizes up to 50% of a visual field, values were given as × 5. For histology and immunocytochemistry, the following score was used: 0, no cells with no positive staining or positive immunoreaction; 1, ≤ 1% cells with positive staining; 2, 1–5% positive cells; 3, 15–10% positive cells; 4, ≥ 10% positive cells. Organ toxicity was estimated macroscopically and microscopically as follows: 0, no signs of organ toxicity; 1, slight edema or minimal other changes; 2, moderate organ alterations; 3, severe signs of toxicity.
Improved treatment success in rats treated with doxorubicin bound to nanoparticles
Treatment regimens were chosen based on the results of the above-mentioned toxicologic study.35 These results indicated that the maximum tolerated dose for DOX formulations was approximately 7.5 mg/kg. Therefore, chemotherapy was carried out using total doses of 7.5 mg/kg (3 × 2.5 mg/kg) and 4.5 mg/kg (3 × 1.5 mg/kg). The results of chemotherapy are presented in Table I; Figures 1 and 2 show the Kaplan-Meier survival curves of the rats exposed to treatment with various DOX formulations. The data in Table I are calculated by using the median of the survival times and, as a result, mainly focus on the short-term survival results, whereas the Kaplan-Meier survival curves emphasize the long-term survivors.
Most prominently, the animals treated with 3 × 1.5 mg/kg of DOX-NP+PS (Fig. 1, Tables I) showed a statistically significant increase in survival times compared to the controls (ISTC 85%) and the doxorubicin solution (DOX, ISTD 24%). This result was confirmed in a series of 3 experiments. In addition, cases of long-term remission (after 180 days) were consistently observed. The overall percentage long-term remissions in these groups exceeded 20% (5/23; Fig. 1, Table I). The survival time was also increased in the groups treated with 3 × 1.5 mg/kg of DOX-NP (ISTC 38%) and DOX+PS (ISTC 65%); there were also 2 long-term survivors (2/23, 9%) in the latter group. In contrast, the median of survival in the group treated with DOX alone administered in the same regimen was similar to the latter 2 groups (ISTC 54%); none of the animals in the DOX group survived for more than 65 days. Treatment using 3 × 2.5 mg/kg of DOX-NP+PS led to 169% ISTC or 43% ISTD. There were 2 long-surviving animals (117 and 163 days) in this group. Administration of 3 × 2.5 mg/kg of DOX, DOX+PS and DOX-NP provided a medium antitumor effect manifested as 88% ISTC, 108% ISTC and 62% ISTC, respectively.
Pathology of glioblastoma implantation sites
Cerebral tumor implantation sites were confirmed at gross inspection in all brains, including long-term surviving animals with no histologic tumor detection. The injection canal was macroscopically identifiable on frontal brain slices in 95% of the animals. The BBB integrity for Evans Blue solution after tumor implantation was macroscopically and microscopically confirmed. Tumors of animals sacrificed at day 12 exhibited only small focal necroses and only occasional central hematomas but considerable surrounding brain edema, which appeared to be dependent on the tumor size. There were clear differences in tumor size between control animals and treated animals of all groups (Table II).
Table II. Histologic Evaluation of Intracranially Implanted 101/8 Glioblastoma After Treatment with Formulations of Doxorubicin
Number of animals
Control, day 12
Histopathologic and immunocytochemical evaluation of treatment with different formulations of doxorubicin
The results of the preliminary histologic evaluation of brains from animals with implanted 101/8 glioma are shown in Figure 3 and Table II. All animals showed histologic signs of previous tumor implantation. Nontreated control animals developed a maximal tumor size with comparatively low numbers of necrotic areas around day 12 (Fig. 3a). Therefore, this time point was chosen for comparative histologic evaluation. Some clearly recognizable patterns were observable (Table II). Tumor size was considerably larger in nontreated controls versus animals with doxorubicin treatment, with an even smaller tumor size in animals with DOX-NP+PS treatment (Fig. 3a and e). Similarly, after this treatment, animals had a tendency toward more solid tumor growth and accompanying slight inflammatory infiltrates in areas surrounding the tumor (Fig. 3h). This inflammatory response was most pronounced in the group receiving doxorubicin bound to polysorbate-coated nanoparticles, but also evident in the group receiving DOX+PS (Table II). In contrast, nontreated control animals had a more diffusely invasive tumor growth without notable inflammation and higher numbers of necrotic tumor areas (Fig. 3a and b). They also had higher rates of apoptosis and proliferation (Fig. 3d, Table II), delineating a higher cellular turnover. No tumor was detected in 2 long-term surviving animals (Fig. 3i–l). These rats had large gliotic cerebral scars at the site of original tumor implantation (Fig. 3i and j) with focal reactive inflammatory changes, confirming previous correct positioning of the tumor implant. However, there was no detectable cellular proliferation (Fig. 3k). Also, there was no evidence for metastasis.
The tumor cells stained negative for GFAP (Fig. 3g), but weakly positive for ezrin, an actin-binding astroglial cytoskeleton protein (Fig. 3f). Areas surrounding the tumor showed slight proliferative activity (Fig. 3d) derived either from diffusely invading tumor cells or from reactive astroglia, which appears more likely, since these cells are strongly GFAP-positive (not shown). Apoptosis was rarely observed in some microglial cells outside the tumor perimeter. There was no neuronal apoptosis (Fig. 3l).
Lower doxorubicin toxicity in rats treated with doxorubicin bound to nanoparticles
Signs of organ toxicity occurred at day 12 only in animals treated with doxorubicin alone at the maximal dosage of 3 × 2.5 mg/kg, whereas at this dose level with DOX-NP+PS no clinical symptoms and only minimal gross alterations of the lung without overt histopathology were observed in 1 out of 3 animals. Most importantly, the nanoparticle formulation of doxorubicin produced less peripheral toxicity after systemic administration.
In particular, autopsy of the whole body in healthy animals treated with 2 different doxorubicin formulations revealed an empty gastrointestinal tract in all animals treated with doxorubicin only (n = 3), while no overt pathology was visible. These changes were not observed in animals treated with doxorubicin bound to nanoparticles (n = 3). The healthy animals treated with doxorubicin only showed slight signs of lung edema, which was confirmed by histology and absent in all other groups. In one case (doxorubicin only), we found a small spot of intramural cardiac bleeding, which could be confirmed by histology as a small fresh infarction site. Some of the tumor-bearing animals treated with doxorubicin only showed slight lung edema without overt histopathology. All other organs did not show any signs of toxicity in any of the groups. Necropsy of long-term surviving rats in the group treated with 3 × 1.5 mg/kg DOX-NP+PS did not reveal any pathology.
Cerebral neuronal apoptosis as a sign of neuronal toxicity was absent in any of the studied groups. In one animal of the group of healthy rats treated with doxorubicin only, apoptosis of few glial cells was observed. Nonspecific signs of CNS toxicity such as increased expression of GFAP (glial fibrillary acidic protein) or ezrin on distant astrocytes or degenerative morphologic changes of neurons were entirely absent in treated animals on day 12 and in long-term survivors. In addition, there was no indication of chronic glial activation in areas distant from the tumor site in long-term surviving rats. Original GFAP expression of glioblastomas is usually lost after several cycles of cell culture.36 Any GFAP expression found in the rat model is due to reactive astrocytes surrounding the tumor site. The expression of the actin-binding protein ezrin, which is also typically present in astrocytes or astrocytic tumors, remains much more stable34, 37 and is therefore more suitable to demonstrate glial derivation of tumor cells in experimental glioblastoma. However, ezrin may also be present in surrounding reactive astrocytes, and for this reason less likely to differentiate between glioblastoma cells and reactive astrocytes than GFAP.
In the present study, the therapeutic potential of doxorubicin bound to nanoparticles for the therapy of rats with intracranially implanted glioblastomas was investigated. A significant increase in survival time and more than 20% long-term remission animals were found in the group of rats treated with doxorubicin bound to polysorbate-coated nanoparticles (DOX-NP+PS) as compared to control animals or animals treated with other formulations of doxorubicin (Tables I and II). By histology, doxorubicin-treated animals had an overall slower tumor growth with lower tumor sizes, lower cellular turnover and an increased concomitant inflammatory reaction. The attachment of doxorubicin to nanoparticles also had a favorable effect on doxorubicin organ toxicity. There was no indication of short-term neurotoxicity. It is possible that in addition to the doxorubicin nanoparticles overcoated with polysorbate 80 (DOX-NP+PS), the doxorubicin solution in combination with polysorbate 80 (DOX+PS) could also have a therapeutic effect.
Polysorbate-coated nanoparticles represent an effective delivery system for chemotherapy of brain tumors
The pharmacokinetics of doxorubicin bound to polysorbate 80-coated poly(butyl cyanoacrylate) nanoparticles after intravenous administration in healthy rats has been studied previously21 and showed significant doxorubicin concentrations of around 6 μg/g in the brain after intravenous injection of a single dose of 5 mg/kg. In contrast, the brain concentrations were below the detection limit of 0.1 μg/g in all other groups, including doxorubicin in saline, doxorubicin plus polysorbate 80 and doxorubicin bound to nanoparticles without polysorbate coating. In addition, nanoparticles are easily prone to endocytosis and are taken up by a large number of solid tumors due to the enhanced permeability and retention effect.26, 27, 28, 29 Taken together with the known ability of glioma cells for phagocytosis,38 the results of the study of Gulyaev et al.21 suggested that doxorubicin bound to the polysorbate 80-coated NP was likely to be effective against brain tumors. This expectancy was met in the present study. The results of the present experiments demonstrated the high efficacy of this formulation for therapy of rats bearing 101/8 glioblastoma. The nanoparticles provided a significant increase of the survival time of 24–43 ISTD % compared to controls and DOX for all dose ranges studied (Table I). The overall best result was obtained in the group treated with 3 × 1.5 mg/kg of DOX-NP+PS; in this group over 20% of the rats (5/23) showed a long-term remission (Table I). These animals were sacrificed after 6 months and no signs of tumor could be observed by histologic examination. Although the mean survival time was longer in the group treated with 3 × 2.5 mg/kg of DOX-NP+PS, indicating a dose dependence of the treatment success, long-term surviving animals of this group died before day 180. The latter was possibly due to biologic variance in a relatively low number of animals. A considerably more extensive toxicology study showed a clear dose dependence of doxorubicin effects.35 The small additional number of animals studied here for organ toxicity indicated that the use of nanoparticles might reduce the organ toxicity of doxorubicin in the group of animals treated with the maximum dosage. However, the maximum dose applicable may still be limited by the systemic toxicity of doxorubicin.39, 40, 41, 42
Potential mechanisms of drug delivery to brain
The most likely mechanism of the drug transport with nanoparticles appears to be low-density lipoprotein (LDL) receptor-mediated endocytosis due to selective adsorption of apolipoproteins B and/or E to the particle surface after injection into the blood.43 The particles thus seem to mimic lipoprotein particles and are taken up by the brain endothelial cells that express numerous LDL receptors. The drug may then be transported into the brain by diffusion following release from the very rapidly biodegrading nanoparticle polymer. It is also possible that the nanoparticles are transcytosed, although no concrete evidence for this mechanism exists at present.
Recent studies revealed concerns about a disruption of the BBB with the polysorbate-coated nanoparticles.44 However, it was shown by Alyaudtin et al.45 that the BBB was not significantly disrupted by a general opening of the tight junctions but that the volume of distribution measured by the spaces accessible to the intravascular marker inulin was increased by other means. At concentrations of poly(butyl cyanoacrylate) nanoparticles and polysorbate 80 that achieve significant drug delivery to the brain, there is little in vivo or in vitro evidence to suggest that a generalized toxic effect on the BBB is the primary mechanism for drug delivery to the brain.46
However, it is known that the BBB is partially disrupted as a result of tumor growth, especially at the late stages of tumor development. Hence, it is very likely that the administration of DOX, DOX+PS and DOX-NP on days 5 and 8 could also enable drug transport to the tumor site across the leaky endothelium. However, by binding to poly(butyl cyanoacrylate) nanoparticles coated with polysorbate 80 (DOX-NP+PS), doxorubicin (and possibly other antitumor drugs) can also be transported across the intact BBB.21 The delivery across the intact BBB enables a cytostatic effect also at tumor sites with an intact BBB. This is especially important in humans since the brain is so much larger than in the rat, with the consequence that drugs have to cover much longer diffusional distances to reach the target site following localized drug delivery. As a result, local drug injections and drug-loaded implants have only a limited efficacy since they are able to deliver the drug only over very short distances,16, 47 whereas drugs bound to polysorbate–coated nanoparticles can reach the whole brain. Thus, the use of coated nanoparticles may even present better options of the treatment of human gliomas than liposomes, which so far did not provide a successful long-term therapy of glioblastomas with doxorubicin.20 However, a direct comparison of both vehicles in the same model system has not yet been performed.
Potential immune mechanisms of drug actions in brain
While the animal numbers were entirely sufficient to evaluate the efficacy of the doxorubicin-loaded nanoparticles in tumor-bearing rats, histology using a relatively low number of animals was added to confirm the technical aspects of tumor transplantation and the toxicology results of previous studies. This number was not sufficient for a comprehensive statistical evaluation. However, a histologic pattern was emerging, with some results being unexpected and pointing toward potential mechanisms of drug interaction in the brain, which, although still speculative, merits further research. As expected, smaller tumor sizes, less diffuse growth and lower proliferative activity were found in animals treated with doxorubicin. Rather surprisingly, apoptosis rates that we would have expected to increase48, 49 were lower as well. It is likely that the results obtained here reflect the biologic variance in a small sample. It is also possible that the used observation window based on clinical appearance was not optimal for this purpose. Similarly, the number of animals employed for histology was probably too low to detect clear-cut differences in tumor sizes between the different treatment groups, with the exception of the observed inflammatory reaction in the 2 groups given polysorbate 80-containing formulations that was absent in other groups. Histologic evaluation of brains from rats of the group treated with DOX-NP+PS points toward an additional mechanism of drug action in the CNS, which may be specifically related to the polysorbate coating of the nanoparticles. These animals show a tendency toward increased inflammatory response to the tumor, which may indicate an increased cytotoxic immune response in a system that is immunosuppressive under normal conditions.50, 51, 52 One could speculate that this immune response may result in several related consequences by either abrogating a local tolerance induction to the tumor, or by inducing chronic activation of surrounding glia leading to their increased phagocytic activity.50, 51, 53, 54 These events could occur in parallel and could be related as well to a decreased barrier function of the vasculature in these animals, facilitating further enrichment of the chemotherapeutic compounds. Possible effects of surfactant-induced alterations of the BBB cannot be excluded. The observed slight inflammatory reactions are likely to be beneficial in terms of a more effective tumor removal. One potential disadvantage could consist in a chronic extensive glial activation, which, however, was not observed in long-term surviving animals, although glial scarring was prominent enough at tumor sites to confirm previous correct tumor implantation. Thus, further studies will be necessary to elucidate the immunologic effects of polysorbate coating of nanoparticles.
No indication of drug-related neurotoxicity
A potential neurotoxicity of doxorubicin has aroused controversy.10, 11, 14, 35, 55 For example, findings by Neuwelt et al.55 indicated that doxorubicin was neurotoxic for the brain following osmotic opening of the BBB and intracarotid artery injection. However, Boiardi et al.,10 Lippens11 and Koukourakis et al.14 demonstrated in clinical trials that liposomal doxorubicin or a similar anthracycline, daunorubicin, could be effectively delivered to the brain after i.v. injection but showed no significant adverse effects in the patients. Moreover, the cytotoxicity of doxorubicin in organotypic multicellular spheroids of human gliomas appeared to be higher than that of BCNU (carmustin).56 We have recently demonstrated that the toxicity of doxorubicin bound to nanoparticles was rather lower than that of a solution of doxorubicin in saline.35 In the present study, we found only limited systemic toxicity that was dose-dependent and obviously not enhanced but rather alleviated by polysorbate coating of nanoparticles. Indications of short-term neurotoxicity, such as increased apoptosis in areas distant from the tumor, increased expression of GFAP or ezrin on distant astrocytes and degenerative morphologic changes of neurons, were entirely absent in treated animals on day 12 as well as in long-term survivors. In addition, there was no indication of chronic glial activation in areas distant from the tumor site in long-term surviving rats. Moreover, long-term survivors did not exhibit any obvious neurologic symptoms. However, long-term neurotoxicity in the form of subtle alterations of synapse function or glutamate turnover cannot be entirely excluded.
In conclusion, this study showed that nanoparticles provide an therapeutically effective distribution of doxorubicin into the brain in vivo. It is also evident that this approach offers new opportunities for noninvasive chemotherapy of brain tumors. At the present stage, doxorubicin bound to polysorbate-coated nanoparticles appears to be the doxorubicin formulation that is most effective in the CNS and that may be the most likely to elicit an accompanying immune reaction to the tumor. There are no indications of short-term neurotoxicity.
The authors thank Peter Stoldt (Edinger-Institute, Frankfurt/Main, Germany), Martin Michaelis (Institute of Pharmaceutical Technology, University of Frankfurt, Frankfurt/Main, Germany), David Begley (King's College, London, U.K.) and Hagen v. Briesen (Institute of Biomedical Research, Georg-Speyer-Haus, Frankfurt/Main, Germany) for helpful discussions and support during this study. Supported by the Deutsche Forschungsgemeinschaft (Graduiertenkolleg Arzneimittel: Entwicklung und Analytik as well as travel grants to A.S.K. and S.E.G.), INTAS, the Boehringer-Ingelheim-Fonds and the Hermann-Schlosser-Foundation (to S.C.J.S.). This study was also enabled by a generous gift of doxorubicin by the Sicor Company (Rho, Italy).