Development of Solid Lipid Nanoparticles Containing Ionically Complexed Chemotherapeutic Drugs and Chemosensitizers HO LUN WONG,1 REINA BENDAYAN,1 ANDREW MIKE RAUTH,2 XIAO YU WU1 1
Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, 19 Russell Street, University of Toronto, Ontario, Canada M5S 2S2 2
Ontario Cancer Institute, 610 University Avenue, Toronto, Ontario, Canada M5G 2M9
Received 4 September 2003; revised 15 March 2004; accepted 15 March 2004 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20100
ABSTRACT: The purpose of this study was to develop and characterize a solid lipid nanoparticle (SLN) system containing an anionic polymer for the delivery of cationic antineoplastic agents and chemosensitizers. Ionic complexation was utilized to enhance the loading of these highly water-soluble drugs. The influence of anionic compounds and polymers on drug partition and loading into SLNs was investigated, and dextran sulfate (DS) was found to be the most suitable among those studied. SLNs loaded with doxorubicin and various model chemosensitizers (e.g., verapamil) were thus prepared by incorporating DS using a microemulsion method. The particle size was measured with photon correlation spectroscopy. The mean diameter of the SLNs ranged from 180 to 300 nm, depending on the type and content of the drug and the polymer. The particles possessed weakly negative surface charges as determined by zeta potential measurements. Most polymer-loaded SLNs released half of the drug in the first a few hours and the remaining drug in 15 h or more. The presence of counterions in the medium, especially divalent ions, promoted drug release. Dual drug (doxorubicin/verapamil or quinidine/ verapamil)-loaded DS-SLNs were also formulated, which released both drugs without noticeable interference to each other. These studies have laid the foundation for a ‘‘onebullet’’ dosage form that may provide convenient and effective delivery of multiple drug treatment of tumors. ß 2004 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 93:1993–2008, 2004
Keywords: anticancer drugs; solid lipid nanoparticles; polymer–drug complex; controlled release/delivery; doxorubicin; chemosensitizers
INTRODUCTION To overcome the shortcomings of conventional cancer chemotherapy such as a low therapeutic index and the emergence of the multidrug resistance (MDR) phenotype,1–3 a variety of approaches have been developed. These include the use of cytotoxic drugs in combination, coadministration of chemosensitizers such as the Correspondence to: Xiao Yu Wu (Telephone: 416-978-5272; Fax: 416-978-8511; E-mail:
[email protected]) Journal of Pharmaceutical Sciences, Vol. 93, 1993–2008 (2004) ß 2004 Wiley-Liss, Inc. and the American Pharmacists Association
inhibitors of cellular drug-efflux activities of Pglycoprotein (P-gp), or multidrug resistance-associated protein (MRP) for the reversal of MDR,4–7 and combinational use of chemosensitizers for synergistic therapeutic activities.8–10 Each of these approaches shows some promise as well as limitations. The use of nanoparticles for the delivery of multiple cytotoxic agents and chemosensitizers could eliminate some limitations because they have the ability to deliver drugs at high effective concentrations to the targeted sites with a lower side effect compared to the conventional dosage forms.11 In addition, nanoparticles
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 8, AUGUST 2004
1993
1994
WONG ET AL.
have demonstrated good potential in reversing the MDR phenotype.12–15 Nonetheless, the development of an integrated approach to deliver combinations of cytotoxic agents and chemosensitizers with a nanoparticulate system is still in its infancy. Solid lipid nanoparticles (SLNs), a lipid formulation with high adaptability, superior handling properties, and low toxicity,16–18 have the potential to be a generic carrier system for anticancer drugs. For lipophilic drugs, SLNs are particular useful because of high partition of the drugs in the lipid phase. For this reason, free bases of drugs should be used. However, free bases of many antineoplastic agents and chemosensitizers are practically water insoluble, so their salt forms are usually preferred clinically. The cationic charges of these salts may pose an obstacle for efficient drug incorporation into lipid particles. The use of counterions such as organic phosphate to facilitate loading of doxorubicin hydrochloride has shown promise in solving this problem.19 However, small drug molecules, in particular, tend to diffuse rapidly to and concentrate near the particle surfaces during the lipid solidification process, and readily leak out when placed in the medium (burst effect).20 This is particularly undesirable in terms of cancer therapy, considering the high toxicity of most antineoplastic agents, and calls for additional drug release control mechanisms to achieve optimal cancer therapy. Our group has previously reported high-level loading of water-soluble, cationic antineoplastic drugs in microspheres (MS) of crosslinked sulfonylated dextran.21–24 The nonspecific nature of drug–polymer ionic interactions allows convenient loading of structurally unrelated ionic drugs and combinations of these agents into the MS (Internal communication to R.Cheung, University of Toronto, Ontario, Canada).21 Attempts to physically downsize these microspheres (40– 120 mm in diameter) in our preliminary study have resulted in unacceptably fast drug release. Because it has been proven that drug release can be retarded by coating microspheres with a hydrophobic polymer, we anticipate that by complexing cationic drugs with a chemically similar but noncrosslinked anionic polymer, such as dextran sulfate (DS), and incorporating the drug–polymer complex into a SLN system, the lipid matrix will slow down the drug release. The large molecular weight of the drug–polymer complex and the need of counterions for exchange and release of the bound drugs are also anticipated JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 8, AUGUST 2004
to reduce the burst effect. In addition, DS may provide additional therapeutic benefits with its intrinsic antimetastatic and anticancer effects, as recently demonstrated.25,26 The present studies were thus intended (1) to investigate the feasibility of preparing such a SLN system that may combine the advantages of a polymeric carrier and a lipid formulation, (2) to establish a functional polymeric SLN system based on these data, and (3) to assess the capability of the SLN system for the loading and release of combinations of model ionic antineoplastic agents including doxorubicin and various chemosensitizers.
MATERIALS AND METHODS Materials Stearic acid (Fisher Inc, Canada) was purified by recrystallization in 95% ethanol. DS (MW: 5000), sodium alginate (2% viscosity grade), ionic drugs doxorubicin HCl (DOX), verapamil HCl (VER), propranolol HCl (PRP), quinidine sulfate (QS), and other chemicals including diclofenac sodium and zidovudine (AZT) were all purchased from Sigma, Inc. (Canada), unless otherwise specified. DOX is a broad-spectrum cytotoxic drug widely used for the treatment of various cancer types, and VER, PRP, and QS all have established P-gp inhibitory effects.5,7 VER was the primary model drug used throughout the studies because of its chemical stability. Tween-801, a nonionic surfactant approved by FDA for parenteral use,27,28 with established intrinsic chemosensitizing effect was a gift from Uniqema, Canada. Distilled and deionized water (DDIW) was prepared with MilliQ water purifier (Milli-Pore, Canada). Determination of Drug Partition Coefficients Drug partition studies were conducted to select the polymer/drug combinations suitable for the SLN system and to provide basic understanding of the process of drug–polymer complexation necessary for the optimization of the system. A two-phase solvent system was established by mixing one-octanol and DDIW, 10 mL of each. A known amount of a drug was dissolved in the twophase system, followed by addition of a polymer or organic salt if applicable. In the study of VER partition, the total amount of VER was kept at 0.04 mol in each experiment while the amount of DS was varied to give various polymer-to-drug
SOLID LIPID NANOPARTICLES CONTAINING IONICALLY COMPLEXED CHEMOTHERAPEUTIC DRUGS
ratios. The molar ratio of DS to VER was calculated as the molar ratio of their charges on the basis that each dextran unit contains an average of 2.3 free ionizable sulfate groups as described in the product monograph, and one verapamil molecule was assumed to carry one full positive charge. The mixture was stirred and heated to 708C. Samples were drawn from the organic phase and diluted with a warm mixture of ethanol/0.3 M calcium chloride solution (80:20 v/v ratio). The calcium ions served as an ion exchange species to release the drug that complexed with the polymer or the organic salt. The drug concentrations in the samples were determined by UV-visible spectrophotometer (Agilent 8453 UV spectrophotometer, Germany). A set of experiments using preequilibrated one-octanol and water mixture were also conducted to check the possibility of the effect of solvent equilibration. The partition results were essentially identical to those obtained without preequilibration. Hence, only the data without preequilibration were reported. Preparation of Ionic–Polymer Solid Lipid Nanoparticles VER-, PRP-, and QS-SLNs The particles were prepared based on the microemulsion technique as described by Aquilano, et al.,29 with some adjustments made for the purposes of this study. In short, a specified amount of the drug(s) was dissolved in 1 g of water and 0.46 to 0.52 g of Tween-801, followed by addition of an anionic polymer (e.g., DS) or an organic salt in several small portions under constant stirring. Following the addition of 0.18 to 0.25 g of stearic acid, the mixture was heated to 72–748C, and then slowly cooled down to 65–678C until the formation of a translucent lipid emulsion. The emulsion was quickly dispersed in water at 2–48C at a volume ratio of 1 part emulsion to 15 parts water. Except for the study of drug loading and size measurement, where freshly prepared samples were used, the samples were dialyzed using dialysis tubing (Spectropor, LMCO ¼ 300 kDa) against DDIW at room temperature for 1 h to reduce the concentrations of unloaded drugs to below 40% of their initial levels. DOX-Containing SLNs DOX was relatively prone to aggregation with polymer upon prolonged heating. A different
1995
treatment was therefore applied to the preparation of SLNs containing DOX. Predetermined quantities of DOX or DOX and other drug(s) were sequentially dissolved in Tween-801 and water as described above. DS was added in three portions with an interval of 1 min, and the mixture was stirred at room temperature overnight in the dark until there was no visible precipitate. The drug– polymer–surfactant mixture was vacuum dried, then heated with melted stearic acid at 728C and stirred until homogenous. Water preheated to the same temperature was added once to the mixture to produce a microemulsion. The SLNs were prepared and the unloaded drug was removed in the same manner as previously described. In the above experiments, the amount of drug added in the SLNs was kept constant (for every 100 mg stearic acid, 10 mg of VER or PRP, or 2.5 mg of DOX was added, respectively) with the amount of DS being varied. The molar ratio of DS to drug was computed as the molar ratio of ionic charge of (polymer to drug). Determination of Particle Size, Morphology, and Surface Charge Samples of SLNs loaded with the drug/polymer of interest were diluted with ultrasonic-degassed DDIW, and the particle size distribution and the zeta potential of the particles were measured with an instrument combining the functions of photon correlation spectroscopy (PCS) and zeta potential measurement (Nicomp Zetasizer 380, USA). Multiple minor peaks were occasionally observed. To allow comparison of particle size, for each particle type, the weighted means of all peaks of individual samples were first calculated, and the overall mean of these weighted means was reported. The effect of exposure to ionic medium on the size distribution of drug-loaded DS-SLNs was studied in a calcium fortified buffer, prepared by dissolving 0.05 mol of calcium chloride per liter of phosphate buffer (pH 7.4, h ¼ 0.1 M). The particles were incubated in the buffer at 378C. The VER-DSSLNs used in this study were loaded with 5.5% w/w of VER on the basis of total particle weight. Particles were diluted with DDIW prior to the measurements and two 5-min cycles were run for each sample. The data were collected at predetermined time points from 15 min to 2 h using PCS. The morphology of the VER-DS-SLNs (freshly prepared or incubated in a calcium fortified buffer at 378C for 24 h) was examined by transmission electron microscopy (TEM) (Hitachi JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 8, AUGUST 2004
1996
WONG ET AL.
7000H, Japan). The nanoparticles were stained with ruthenium tetroxide (2% w/v aqueous solution) prior to examination. Size measurements of randomly selected nanoparticles were conducted by the computer software package provided by the manufacturer of the microscope (Hitachi, Japan) and were included in the pictures. These data provided supporting information about the particle size. Determination of Percent Drug Loading and Encapsulation Efficiency In each measurement 2 mL of the freshly prepared drug-loaded SLN suspension was filtered with a 0.1-mm filter unit (Millipore, USA). The drug concentration(s) in the filtrate was measured with vis-UV spectrophotometry. To take into consideration the amount of drug adsorbed onto the filters, the calibration curves were prepared using standard drug solutions that were filtered in the same manner. The total amount of loaded drug was calculated using the formula: (total weight of drug addedcalibrated drug concentration in filtrate total volume of SLN suspension). In the case of multiple drugloaded particles, absorptions at multiple wavelengths corresponding to the absorption peaks of the component drugs were simultaneously measured, and the amounts of individual drugs in the solutions were resolved by software using the Multicomponent Analysis Program (Agilent ChemStation, USA). Percent drug loading was calculated as (weight of drug loaded 100%)/ (total weight of particles) and drug encapsulation efficiency using the formula: (weight of drug loaded 100%)/(total weight of drug added in the preparation). Drug Release Studies Owing to the submicron size of the particles and fast drug release compared to hydrophobic drugloaded SLNs, the traditional method for measuring drug release, for example, assaying remaining drug in the particles after separation of the nanoparticles from the release medium by centrifugation, cannot be applied to the present SLN system. Hence, a nondestructive, automated method, the so-called dialysis-tubing method, which is well-accepted in studies of release kinetics of nanoparticles,30–32 was used in most of the measurements of drug release. In a typical experiment, 5 mL of SLN suspension in DDIW at JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 8, AUGUST 2004
378C was enclosed in a dialysis tubing with high permeability (Spectropor: LMCO ¼ 300 kDa). The two ends of the tubing were then sealed using tubing clamps. Negligible leakage of the particles was confirmed by preliminary tests using blank SLNs, which showed no noticeable change in the light transmittance of the release medium. Drug release was initiated by immersing the filled and sealed tubing in a large volume of releasing medium (typically 40 times of the volume of the suspension). The composition of the release medium was varied to study the influence of type of ions (i.e., calcium or sodium) and ionic concentration. In some experiments, calcium ions (in the form of 2 M calcium chloride) were added to DDIW 1 h later to investigate the impact of the divalent ions on the drug release rate. The amount of drug released from the SLNs was calculated as follows: Mt ¼ Co Vo þ Ci Vi
ð1Þ
where Mt is the amount of drug released from the particles at a specified time; Co is the drug concentration outside the dialysis tubing, which is due to drug release from the SLNs; Ci is the drug concentration inside the dialysis tubing, which is due to drug release from the SLNs; Vo is the volume of medium outside the dialysis tubing; Vi is the volume of medium inside the dialysis tubing. Because unremoved free drug is present initially in the dialysis tubing, the actual drug concentrations measured inside and outside the dialysis tubing are contributed by both unremoved free drug and drug released from the SLNs. To deduct the contribution by the unremoved free drug, eq. 2 was derived (see the Appendix for detailed derivation). Mt ¼ Co;total Vo þ Ci;total Vi Ci;un Vi
ð2Þ
where Co;total is the total drug concentration outside the dialysis tubing contributed by both drug released from the SLNs and unremoved free drug; Ci,total is the total drug concentration inside the dialysis bag that is due to both drug released from the SLNs and presence of unremoved free drug; and Ci;un is the initial concentration of unremoved free drug inside the dialysis tubing. The drug concentration in the medium outside the dialysis tubing, Co,total, was continuously assayed by vis-UV spectrophotometry. The values of Ci;un were evaluated using the encapsulation
SOLID LIPID NANOPARTICLES CONTAINING IONICALLY COMPLEXED CHEMOTHERAPEUTIC DRUGS
efficiency determined above and the percent of unloaded drug removed by dialysis in DDIW in 1 h. For example, for an encapsulation efficiency of 70% in a loading solution with an initial concentration of 1.0 mg/mL, and 60% of unloaded drug removed by dialysis as estimated from free drug solutions, Ci;un was calculated to be 0.12 mg/mL. Because Ci,total at t > 0 could not be directly assayed due to the presence of SLNs, it was indirectly determined from the measured Co,total and its change rate based on the following relationship derived from Fick’s Law: DM DCo;total Vo ¼ ¼ PAðCo;total Ci;total Þ Dt Dt
ð3Þ
where DM is the amount of drug diffusing out of the dialysis tubing in time Dt; Dt is the time intervals; P is the membrane permeability; and A is the area of dialysis membrane. For practical reasons, the rate of Co,total increase in a 2-min time frame was assumed to be the instantaneous change rate, DCo,total/Dt. The value of the product P A was empirically determined with free drug solutions of various concentrations under the same experimental condition, and was verified to be a constant within the investigated concentration range of the drugs. The value of Ci,total was determined from eq. 3 and then inserted into eq. 2 calculate the value of Mt at a specified time. To avoid excessive fluctuation or noise, the average values of Mt in 10-min windows were used for drug profile plotting.
1997
In addition to the aforementioned method, a manual sampling-filtration method, later referred as the ‘‘filtration method,’’ was also employed in some study for comparison. Aliquots of VER-DSSLN suspension in calcium chloride solution were manually drawn at predetermined time points, and filtered with 0.1-mm filter units. The amounts of verapamil released were determined by spectrophotometry from the drug concentration in the filtrate with calibration for the adsorbed drug as previously described. Statistical Analysis One-sided Student’s t-test was performed to determine if enhanced drug partitioning in the octanol-water two-phase system and improved drug loading in SLN preparation were achieved with anionic polymer or organic salt complexation. Unless indicated otherwise, a p-value of 0.01 was considered significant.
RESULTS Effect of Ionic Polymers and Organic Salts on Drug Partitioning The effect of various anionic polymers and organic anions on enhancing the distribution of ionic drugs into the organic phase in a standard oneoctanol/water two-phase system is summarized in Table 1. In general, the anionic polymers and organic anions tested demonstrated a significant
Table 1. Effect of Anionic Polymers and Organic Anions in Enhancing Cationic Drug Partition in the Lipophilic Phasea Apparent Drug Partition Coefficient (One-Octanol/Water) Koct/aq Polymer or Organic Salt Addedb Drugc VER PRP QS AZT Diclofenac
Nil (Control)
Dextran Sulfate
Sodium Alginate
Alginic Acid
Sodium Stearate
Sodium Acetate
0.66 0.01 0.24 0.01 1.19 0.32 1.30 0.08 0.28 0.03
18.7 8.4 2.70 0.08 2.22 0.59d 1.21 0.14 0.37 0.05
1.47 0.06 0.61 0.02 1.52 0.27 N.D.e N.D.
0.59 0.01 0.29 0.00 1.22 0.21 N.D. N.D.
15.9 0.4 2.47 0.05 2.13 0.15 N.D. N.D.
2.60 0.18 1.71 0.11 1.67 0.10 N.D. N.D.
a Each value represents the mean SD of at least three separate experiments, rounded to three significant figures, unless otherwise stated. Values significantly larger than the corresponding controls (p < 0.01, one-tailed) are printed in bold. b Polymer or salt with the number of anionic functional groups equimolar to cationic charge of the drug added. c Drugs used were in their commercially available salt forms (verapamil hydrochloride, propranolol hydrochloride, quinidine sulfate, diclofenac sodium; AZT has no salt form). d p value >0.01 but
