|Home | About | Journals | Submit | Contact Us | Français|
Convection-enhanced drug delivery (CED) is a novel approach to delivering drugs into brain tissue. Drugs are delivered continuously via a catheter, enabling large volume distributions of high drug concentrations with minimum systemic toxicity. Previously we demonstrated that CED formation/extent of small molecules may be significantly improved by increasing infusate viscosities. In this study we show that the same methodology can be applied to monodispersed maghemite nanoparticles (MNPs). For this purpose we used a normal rat brain model and performed CED of MNPs over short infusion times. By adding 3% sucrose or 3%–6% polyethylene glycol (PEG; molecular weight 400) to saline containing pristine MNPs, we increased infusate viscosity and obtained increased CED efficacy. Further, we show that CED of dextran-coated MNPs (dextran-MNPs) resulted in increased efficacy over pristine MNPs (p < 0.007). To establish the use of MRI for reliable depiction of MNP distribution, CED of fluorescent dextran-MNPs was performed, demonstrating a significant correlation between the distributions as depicted by MRI and spectroscopic images (r2 = 0.74, p < 0.0002). MRI follow-up showed that approximately 80%–90% of the dextran-MNPs were cleared from the rat brain within 40 days of CED; the rest remained in the brain for more than 4 months. MNPs have been tested for applications such as targeted drug delivery and controlled drug release and are clinically used as a contrast agent for MRI. Thus, combining the CED method with the advantages of MNPs may provide a powerful tool to treat and monitor brain tumors.
One of the most promising methods for efficient drug delivery into brain tumors is convection-enhanced drug delivery (CED). CED involves positioning the tip of an infusion catheter within the brain tissue and supplying the drug through the catheter while maintaining a positive pressure gradient from the tip of the catheter during infusion. The catheter is connected to a pump that delivers the drug and maintains the desired pressure gradient throughout delivery. Drug delivery rates are typically about 0.5–4.0 μl/min with infusion distances on the order of centimeters. CED involves formation of pressure gradients in the tissue. Convective pressure occurs under normal conditions of bulk flow of brain interstitial fluid due to changes in hydrostatic and osmotic pressures,1 under the pathologic conditions of vasogenic edema,2 which involves an increase in extracellular fluid volume, and after direct infusion of solutions into the brain parenchyma.3 CED supplements diffusion and greatly enhances the distribution of molecules in the brain,4,5 capable of obtaining in situ drug concentrations several orders of magnitude greater than those obtained by systemic administration. The concentration profile is relatively flat up to the flow front.6
We have recently published the results of a phase I/II clinical trial where recurrent glioblastoma multiforme patients received CED of Taxol with a significant antitumor response rate.7 CED of other drugs such as Tf-CRM107,8 TP-38,9 and PE38KDEL10 was also found to have clinical activity of varying degrees. Furthermore, recent animal studies show that liposomes can be delivered via CED into brain and brain tumors, as a vehicle for therapeutic agent delivery.11–14
Accumulated clinical experience shows that there is significant variability in the extent of convection among drugs and among patients. For that reason, a reliable real-time, noninvasive monitoring technique is required, as well as means to increase CED efficacy. We have recently shown that low-viscosity infusates tend to backflow along the catheter track while high-viscosity infusates tend to form efficient convection.15 We also showed that CED formation and extent may be improved by increasing infusate viscosities, thus increasing treatment effects. In this study we used our novel high-viscosity methods to obtain efficient CED of maghemite nanoparticles (MNPs) of approximately 80 nm average diameter, which can be directly depicted by MRI, in a normal rat brain model.
Magnetic nanoparticles are considered to be spherical particles with magnetic properties, ranging from a few nanometers to approximately 100 nm in size. These particles, because of their spherical shape, very large surface area, and magnetic properties, may have a wide range of potential applications, such as drug delivery, MRI, specific cell labeling and separation, cell tracking, diagnostics, hyperthermia, and biocatalysis.16–22 In recent years, extensive efforts have been made to design nanoparticles for specific drug targeting applications.23–27 Of particular interest for the present study is the use of magnetic nanoparticles for specific targeting via MRI. Recently, we have prepared and characterized new uniform maghemite (γ-Fe2O3) nanoparticles. These MNPs are formed by controlled nucleation and then growth of thin maghemite films onto appropriate nuclei.17,28 The resulting nanoparticles are superparamagnetic, biodegradable, and nontoxic, and have a narrow size distribution. Due to the superparamagnetic properties, it is possible to perform the imaging and assessment of the MNP distribution using MRI. Superparamagnetic MNPs produce a dramatic long-range disturbance in magnetic field homogeneity, causing marked shortening of T2* relaxation time. (T2* is the characteristic time constant that describes the decay of transverse magnetization, taking into account the inhomogeneity in static magnetic fields and the spin-spin relaxation time.) This results in a rapid loss of phase coherence and loss of MRI signal in the area of the particles. The goals of this study were (1) to establish the use of MRI for a reliable depiction of the MNP distribution in a normal rat brain model; (2) to use MRI to optimize infusate viscosity parameters to obtain maximum MNP distribution in minimum infusion time; and (3) to study the clearance time of MNPs from brain tissue for future use in sustained drug release.
Gelatin (type A from porcine skin, 300 bloom), dextran (molecular weight [MW] 35,000–45,000), divinyl sulfone (DVS), glycine, sucrose, and polyethylene glycol (PEG) (MW 400) were all purchased from Sigma (Rehovot, Israel). Salts for buffers were purchased from Bio-Lab Ltd. (Jerusalem, Israel). Formaldehyde was purchased from Gadot Biochemical Industries, Ltd. (Haifa Bay, Israel). Water was purified by passing deionized water through an Elgastat Spectrum reverse osmosis system (Elga Ltd., High Wycombe, UK).
Three types of MNPs were prepared: (1) pristine MNPs; (2) MNPs coated with dextran (dextran-MNPs); and (3) fluorescent dextran-MNPs. These MNPs were mixed in various solutions containing saline or 3% or 6% PEG (MW 400) in saline or 3% sucrose in saline. Relative viscosity was measured in an Ostwald-type viscometer relative to saline solution at 20°C. The MNP solutions were infused into the striatum of normal Sprague-Dawley (SD) rats (males, 250–300 g). MRI and spectroscopic imaging methods were used to assess the MNP distribution in the brain. For spectroscopic imaging, brains were harvested and submerged in formaldehyde for 24 h and then sliced at the site of the injection. The study was performed in accordance with the guidelines of the Animal Care and Use Committee of Sheba Medical Center, which is recognized by the Israeli authorities for animal experimentation.
MNPs of narrow size distribution were prepared by nucleation followed by controlled growth of magnetic iron oxide layers onto gelatin nuclei, as described previously.17,28 Briefly, FeCl2 solution (10 mmol/5 ml 0.01 normal [N] HCl) was added to 80-ml aqueous solution containing 200 mg porcine gelatin, followed by NaNO2 solution (7 mmol/5 ml H2O). After a reaction time of 10 min, the pH was raised to 9.5 by adding NaOH aqueous solution (1 N). This procedure was repeated four more times. The MNPs were then washed from excess reagents using magnetic columns. Dextran coating was performed by shaking the aqueous suspension of the MNPs containing 2% dextran (MW 35,000–45,000) at 85°C for 1 h. The dextran-coated MNPs were then washed by means of magnetic columns. In order to prevent leakage of dextran from the nanoparticles’ surface to the aqueous continuous phase, crosslinking of the adsorbed dextran with DVS was performed by shaking the dextran-MNPs (2.5 mg/ml) with DVS ([MNP]/[DVS] = 1 mg/4.3 μl) at pH 10.5 and 60°C. The dextran-MNPs were then washed with bicarbonate buffer (0.1 M, pH 8.3) using magnetic columns. The remaining residual activated double bonds of the DVS were then blocked with 1% glycine aqueous solution. Excess glycine was washed from the obtained dextran-MNPs using magnetic columns and saline. Fluorescent dextran-MNPs were prepared similarly, by adding rhodamine molecules to the gelatin solution, thereby entrapping the rhodamine within the MNP.
The diameter and size distribution of the MNPs were measured by a submicron particle analyzer (N4MD, Coulter Electronics Ltd., Hialeah, FL, USA) and a transmission electron microscope (TEM) (JEM-1200EX, JEOL, Japan). Magnetic measurements were performed on a sample of dextran-MNPs that was introduced into a plastic capsule. Measurements at room temperature were performed using an Oxford Instrument vibrating sample magnetometer (VSM). Magnetization was measured as a function of the external field being swept up and down (−16,000 oersteds [Oe] < Happlied < 16,000 Oe, in steps of 200 Oe).
Under full anesthesia, a midline scalp incision was made to identify the bregma. A 1-mm burr hole was made in the right region of the skull, 2 mm lateral and 3 mm anterior to the bregma. A 33-gauge needle attached to a 1,000-μl syringe (Gastight, Hamilton Co., Reno, NV, USA) was placed stereotactically 6 mm deep in the rat striatum. In order to increase the sensitivity of the test, we introduced challenging CED conditions by inserting the injector needle only 4.5 mm into the rat striatum. The infusion was performed using a BASI syringe pump (Bioanalytical Systems, Inc., West Lafayette, IN, USA) at a rate of 4 μl/min for 15 min (unless stated otherwise).
T2*-weighted MR images were acquired immediately after the convection treatment using a 0.5-T interventional GE MRI system and a dedicated, specially designed animal coil. T2*-weighted images were acquired with 3-mm slices, no gap, 14 × 10.5-cm2field of view, a gradient echo (GRE) sequence with a 256 × 256 matrix, repetition time (TR) = 350 ms, echo time (TE) = 14.5 ms, flip angle = 60°, and two signal averages. Spectral images were obtained with the SD300 system based on Fourier transform (FT) spectroscopy (Applied Spectral Imaging [ASI], Migdal Ha’Emek, Israel). The spectral images were analyzed off-line with dedicated ASI-developed software. The principle of spectral imaging, combining conventional imaging with FT spectroscopy, has been described elsewhere.29 An excitation filter of 550 ± 20 nm and an emission filter of >600 nm were used for rhodamine fluorescence detection.
The area (in mm2) of infusate distribution was calculated from the T2*-weighted MR images acquired immediately after CED treatment. Regions of interest (ROIs) were defined over the entire signal void region in each slice (excluding the ventricles). The number of pixels in the ROIs were counted and multiplied by the area of a single pixel. Data are presented as average ± SD. The differences in ROI sizes were analyzed using the nonparametric Wilcoxon two-sample test. Significant differences between the groups were considered when the calculated p value was 0.05.
In the case of fluorescent dextran-MNP infusate, the ROIs were also calculated from the spectroscopic images, and a correlation between the ROI sizes derived from the two imaging methodologies was calculated using Pearson’s correlation test.
Prussian blue staining (Diagnostic BioSystems, Pleasanton, CA, USA) was performed to demonstrate dextran-MNP microdistribution in the rat striatum (dextran-MNPs stained blue). Briefly, 23 days after dextran-MNP convection, the brains were extracted and fixed in 4% glutaraldehyde for 48 h. The brains were paraffin embedded, and 5-μm slices were chosen from the site of convection. The slices were deparaffinized and placed in a Prussian blue solution for 20 min, and then in a nuclear fast red solution, which stained the nuclei red and other tissue pink. Photographs were acquired with a charge-coupled device (CCD) digital camera connected to a light microscope (Olympus BX60, Olympus, Tokyo, Japan).
The mean diameter of the dextran-MNPs as measured by TEM was 14 ± 4 nm. An example of a TEM photomicrograph is shown in Fig. 1A. The light scattering measurements (Fig. 1B) of the dextran-MNPs dispersed in saline resulted in an average diameter of about 81 ± 9 nm with no observable aggregates. A hysteresis loop at room temperature of the dextran-MNPs is shown in Fig. 1C. Similar results were obtained for pristine MNPs (data not shown).
Dextran-MNPs (0.2 mg/ml) were infused at a rate of 4 μl/min for 15 min into the striatum of 31 rats. Rats were scanned by MRI immediately after treatment (Fig. 2). Efficient CED was depicted as homogeneous signal loss throughout the striatum. Inefficient CED, caused by backflow of the infusate along the catheter path, resulted in release into the ventricular system, and was depicted as accumulation of signal loss in the ventricles. The average area of the signal void induced by the dextran-MNP distribution was 25.4 ± 8.1 mm2 (median = 24.5 mm2, range = 6.0–42.4 mm2).
In order to study the clearance rate, we infused dextran-MNPs (0.2 mg/ml) at a rate of 4 μl/min for 15 min into the striatum of 11 rats. The clearance rate was regularly monitored by MRI until most of the dextran-MNPs were cleared from the brain (Fig. 3). Data show a close to linear rate down to approximately 80%–90% clearance after 40 days, at which a plateau is reached.
Prussian blue staining of brains extracted 23 days after treatment showed that most of the residual MNPs were located either intracellularly or adjacent to the cell surfaces. Examples are shown in Fig. 4.
In order to visualize the MNP distribution in the brain directly, we infused a high concentration (2 mg/ml) of dextran-MNPs, resulting in a clear depiction of the treated region (Fig. 5).
To establish the use of MRI for reliable depiction of the dextran-MNP distribution in the rat brain, we used fluorescent dextran-MNPs (0.2 mg/ml) dispersed in saline. These fluorescent dextran-MNPs were infused at a rate of 4 μl/min for 15 min into the striatum of 12 rats (Fig. 6). The fluorescent dextran-MNP distribution areas in the striatum, as depicted by the MR images and the spectroscopic images, were calculated. The correlation (Pearson’s correlation; r2 = 0.74, p < 0.0002) between these two imaging methodologies (Fig. 6) was found to be significant, thus establishing the validity of MRI for monitoring CED of MNPs at the given concentration.
In order to increase the sensitivity of the test, we introduced challenging CED conditions (see Materials and Methods section). The rats were divided into two groups of seven each. One group underwent CED with pristine MNPs (0.2 mg/ml) while the other group underwent CED with dextran-MNPs (0.2 mg/ml). The average area of distribution (calculated from the MR images) of pristine MNPs and dextran-MNPs was 13.4 ± 6.1 mm2 (median = 12.2 mm2, range = 4.9–22.2 mm2) and 23.9 ± 4.6 mm2 (median = 25.9 mm2, range = 16.6–29.7 mm2), respectively. The dextran-MNP distribution area was significantly larger than that of the pristine MNPs (p < 0.007).
Using the challenging CED conditions, we examined the change in CED area distribution caused by increased viscosity infusates. For that purpose, two suspensions of pristine MNPs were prepared. One consisted of MNPs suspended in saline (0.2 mg/ml, relative viscosity = 1.00 ± 0.03), and the other consisted of MNPs suspended in 3% sucrose in saline (0.2 mg/ml, relative viscosity = 1.09 ± 0.03). The rats were divided into two groups. Each group received only one of the MNP suspensions. Infusion into the rat striatum was performed at a rate of 4 μl/min for a duration of 15 min. The rats were scanned by MRI immediately after treatment. The calculated distribution areas of MNPs in saline and MNPs in 3% sucrose solution were 13.4 ± 6.1 mm2 (median = 12.2 mm2, range = 4.9–22.2 mm2) and 23 ± 3.4 mm2 (median = 23.2, range = 17.2–28.2 mm2), respectively, with a significantly larger area for MNPs in 3% sucrose (p < 0.007). In order to further study the influence of increased viscosity on the CED distribution area of pristine MNPs, we infused a solution of pristine MNPs (0.2 mg/ml) in 3% or 6% PEG (relative viscosity, 1.112 ± 0.028 and 1.252 ± 0.031, respectively). The resulting distribution areas, as calculated from the MRI, were compared with those of pristine MNPs in saline (Fig. 7). The distribution areas of pristine MNPs in saline and 3% and 6% PEG solutions were 17.9 ± 6.8 mm2 (median = 18.0 mm2, range = 4.2–24.8 mm2), 24.0 ± 4.7 mm2 (median = 24.9 mm2, range = 14.6–28.9 mm2), and 24.7 ± 3.0 mm2 (median = 25.5, range = 19.6–29.0 mm2), respectively. CED extent of pristine MNPs in 3% PEG and 6% PEG was significantly larger than that of pristine MNPs in saline (p < 0.053 and p < 0.011, respectively).
This study demonstrates the feasibility of obtaining efficient CED of MNPs in a normal rat brain model. The dextran-MNP particle size as measured by TEM was 14 ± 4 nm. In contrast, light scattering measurements showed a size of 81 ± 9 nm. These differences are attributed to the fact that TEM measures the dry diameter, while light scattering determines the hydrodynamic diameter, which takes the hydrated layers on the particle’s surface into account. Furthermore, we show that the M(H) curve of dextran-MNP does not saturate at 10,000 Oe, and that the obtained magnetic moment at 10,000 Oe is about 41 electromagnetic units [emu] g−1. In addition, the M(H) curve does not exhibit any coercivity. Both features are typical of superparamagnetic behavior.
The results show that efficient CED into rat striatum could be obtained with short infusion times (15 min) of dextran-MNPs dispersed in saline. By introducing more challenging CED conditions, thus simulating better actual clinical conditions, we increased the sensitivity of the experiment to the infusate/MNP characteristics. By applying these conditions, we showed that coating the MNPs with dextran increased the distribution area obtained by CED. One explanation for this observation may be the effect of surface coating on protein adsorption. The steric brushes of the dextran macromolecules are believed to reduce interactions with proteins.30,31 Thus, it may be that dextran-MNPs pass more easily through the tissue than do pristine MNPs. Indeed, it has been shown that coating of nanoparticles is essential for preventing nonspecific adsorption of plasma proteins. Biocompatible polymers such as dextran and PEG have been used for this purpose.32–34
Previously, we showed that low-viscosity infusates tend to backflow along the catheter track while high-viscosity infusates tend to form efficient convection. In addition, our previous results suggest that CED formation and extent may be significantly improved by increasing infusate viscosities, thus increasing treatment effects.15 Indeed, by applying the more challenging CED conditions, we showed that the increased viscosity methodology could be applied to pristine MNPs and that it significantly improved the efficiency of CED.
The long clearance time of dextran-MNPs from the rat brain tissue may be explained by the Prussian blue staining depicting the residual MNPs either intracellularly or adjacent to the cell surface. This suggests that the clearance observed in the first 30–40 days may be of extracellular MNPs. The long presence of the particles in the cell vicinity may suggest the use of MNPs as a vehicle for sustained drug release. The MNPs used in this study may be advantageous for in vivo use. No toxicity was detected up to 120 days after CED. In addition, the MNPs are biodegradable, have a narrow size distribution, and are stable at physiological pH.
In an earlier study35 on CED of dextran-coated mono-crystalline iron oxide nanocompounds (MION-46), Kroll et al. evaluated the effect of dose, volume of injection, and infusion time on CED volume during intra-cerebral CED. Our study demonstrates that other parameters, such as viscosity and surface coating, should also be taken into account since they significantly affect the distribution area of CED and allow for higher CED flow rates, thus shorter infusion times.
The correlation between the distributions as depicted by MRI and spectroscopic images establishes the use of MRI for reliable MNP imaging. However, calibration is still needed for different MNP concentrations, since the MRI artifacts caused by MNPs are strongly affected by MNP concentration. In this study, we used MNPs as a model to show the feasibility of obtaining efficient CED of large particles with our high-viscosity methodologies. The ability to convect large particles efficiently implies that it may be possible to efficiently convect other similar size particles such as gene-therapy–related products, liposomes, and so on. Indeed, it has been shown elsewhere that the volume of CED of adenovirus and MION-46 did not significantly differ in a rat brain model.36
Initial clinical experience shows that there is significant variability in the efficient formation and extent of convection among patients, as well as among drugs. Careful examination of the radiological clinical data points to backflow along the catheter path or leakage into low-density regions as one of the main obstacles in forming efficient convection clinically.
We have shown that high-viscosity infusates reduce the sensitivity to backflow, thus increasing the probability for forming efficient convection.15 These results were consistent with the exceptionally high CED efficacy observed in clinical studies with CED of Taxol.7,37 As far as the authors are aware, Taxol is the highest viscosity infusate used in clinical CED-based trials, and the observed response rates were exceptionally high.
The results presented in this article are an extension of the previous study, which dealt with small molecules, suggesting that high viscosity may increase CED efficacy of large particles, MNPs, as well.
The application of CED together with MNPs may enable the use of drugs and/or drug carriers that have been considered inappropriate for convection treatment of brain tumors. The use of MNPs as a drug carrier may enable the delivery of large therapeutic agents and targeted drug delivery. In addition, the long clearance time of MNPs may enable slow drug release.
We thank Dr. Tammy Lublin-Tennenbaum for her help in the synthesis and characterization of MNPs. Thanks also to Mr. Gregory Tamar for his help in performing the CED procedure in the animals, to Dr. Genady Kostenich for his help in performing the spectroscopic imaging, and to Mrs. Sharona Salomon for handling the infusates.