An in vitro--in silico--in vivo approach in biopharmaceutical drug characterization: metformin hydrochloride IR tablets.

The integrated in vitro--in silico--in vivo approach has emerged into a biopharmaceutical toolkit that could accelerate drug development and improve drug product clinical performance in patients. In the present study, the influence of physiologically based media and dynamic dissolution testing on drug release from two metformin hydrochloride immediate release products with proven bioequivalence was tested. Metformin-specific physiologically based pharmacokinetic (PBPK) model was developed based on a range of literature or in silico predicted data using gastrointestinal simulation technology implemented in the Simcyp software package. Various approaches were employed in order to estimate the human effective permeability which was used as input for metformin plasma profile simulation. Influence of the rate and extent of metformin dissolution on drug absorption was evaluated. Both convolution and deconvolution approaches were used in order to establish a correlation between the in vitro and in vivo data. The results obtained indicate that physiologically based dissolution media and glass bead dissolution device exhibit certain advantages over the compendial dissolution apparatus and simple buffers which tended to be over-discriminative. Gastrointestinal simulation technology implemented in the Simcyp Simulator was successfully used in developing drug-specific PBPK model for metformin. Simulations indicate that in vitro dissolution kinetics has no significant effect on metformin absorption, if more than 65% of drug is released in 1 hour. Level A in vitro-in vivo correlation was obtained using both convolution and deconvolution approaches.

Mucoadhesion on urinary bladder mucosa: the influence of sodium, calcium, and magnesium ions.

The aim of the present work was to establish if different cations present in the lumen of the urinary bladder at the time of application affect the mucoadhesion strength of cationic chitosan, anionic sodium carboxymethyl cellulose (NaCMC), and nonionic hydroxypropyl cellulose (HPC). The mucoadhesion strength of polymeric films was determined on pig urinary bladder mucosa. Sodium, calcium, and magnesium ions decreased the mucoadhesion strength of all three polymers except NaCMC, whose detachment forces were not influenced by the presence of sodium. Lower mucoadhesion strength in the presence of cations should be considered when drug delivery systems, for example microspheres, containing the tested mucoadhesive polymers are applied intravesically. In the majority of the experiments, cations decreased the mucoadhesion strength of the polymers already in concentrations normally present in urine. For stronger mucoadhesion, application of microspheres into the empty urinary bladder would be recommended. Additionally, the mucoadhesion properties of the tested polymers could be controlled by the selection of a proper medium for the suspension of microspheres. Namely, for all three polymers bivalent calcium and magnesium had stronger influence on mucoadhesion compared to univalent sodium, and with increasing concentrations of cations mucoadhesion strength of the polymers decreased.

Enhanced permeability of the urinary bladder wall: the role of polymer charge.

The urothelium is usually impermeable to substances present in the urine. In the current work the possibility of using different absorption enhancers in the development of intravesical drug delivery systems was explored. To establish the role of the polymer charge on its ability to improve bladder wall permeability, cationic poly-L-arginine, anionic NaCMC and alginate as well as nonionic HPC and HPMC were tested. The permeability experiments were performed on isolated pig urinary bladders. We established that the charge of the polymer affects its ability to enhance the permeability of the urinary bladder wall, but to a limited extent. Positively charged polymers were the most promising absorption enhancers for the urinary bladder wall. In order to significantly enhance the permeability of the bladder wall, higher concentrations of poly-L-arginine were needed compared to chitosan. Moreover, chitosan reached the plateau of its absorption enhancement effect after 60 min, while poly-L-arginine increased the permeability continuously over 90 min. In contrast to polycarbophil, two other anionic polymers, NaCMC and alginate, did not significantly enhance the permeation of pipemidic acid into the tissue. Interactions between the polymers and the drug might prevail over the potential effect of NaCMC and alginate on tissue permeability. Furthermore, for the nonionic polymers HPMC and HPC an insignificant influence on bladder wall permeability was determined. Therefore, the selection of absorption enhancers for intravesical drug delivery systems is limited and cannot be done only on the basis of polymer charge.

Kinetic model of drug distribution in the urinary bladder wall following intravesical instillation.

Intravesical administration of cytotoxic agents is commonly used in urological practice for treatment of superficial bladder cancer. The leading motive is optimisation of drug delivery near the site of action and reduction of systemic toxicity. Bladder pharmacokinetics is complicated by several mechanisms. The objectives of this work were to develop a kinetic model of drug distribution in the bladder wall following intravesical instillation and to study the effect of various parameters on tissue and systemic drug exposure and explore the potential benefits of permeability enhancing effects of chitosan (CH) and polycarbophil (PC) through simulation. Key elements of the model are variable urinary drug concentration due to urine formation and voiding, biphasic diffusion in the bladder tissue and systemic absorption. Model parameters were estimated from bladder-tissue concentration profiles obtained in previous in vitro experiments with pipemidic acid (PPA) as a model drug. The results support further investigations on application of CH and PC in intravesical drug delivery. Both polymers increase permeability of the bladder wall by diffusion enhancement in the urothelium and presumably by improving the contact with the bladder surface. The developed mathematical model could serve for optimisation of intravesical drug delivery and future development of intravesical drug delivery systems.

Determination of main low molecular weight antioxidants in urinary bladder wall using HPLC with electrochemical detector.

The aim of the present work was to develop validated HPLC method using electrochemical detector for simultaneous detection of low molecular weight antioxidants (LMWA) in urinary bladder. Furthermore, the method was applied to study the distribution of LMWA in urinary bladder wall. The ascorbic acid (AA), glutathione in reduced (GSH) and oxidized (GSSG) form and uric acid (UA) were resolved by isocratic elution from C18 reversed-phase column. The bladder tissue sample preparation involved extraction with meta-phosphoric acid solution for LMWA stabilization. The AA, GSH and UA tissue peak was identified by different approaches. The obtained method validation parameters were in acceptable range: intra-day precision (<4.4%), intra-day accuracy (<8.4%), inter-day precision (<9.4%) and inter-day accuracy (<15.6%). Additionally, the method provided good linearity (r2>0.99) and recoveries (98.9-112.6%). The distribution of LMWA in urinary bladder was determined by measuring their concentration in bladder wall layers: urothelium, lamina propria, muscularis and serosa. The validated method was able to quantify the reduced form of all three LMWA in all four bladder wall layers. The LMWA concentrations were decreasing from urothelium to serosa except of UA. The developed HPLC method with electrochemical detection of LMWA is simple, fast and can be used for simultaneous quantification of LMWA in tissues, which contain low concentrations of antioxidants.

Biomechanical study of different plate configurations for distal humerus osteosynthesis.

Fractures of the distal humerus are most commonly fixed by open reduction and internal fixation, using plates and screws, either in a locking or in a non-locking construct. Three different plating systems are commonly used in practice. The most important differences between them are in plate orientation, which affects both the rigidity of the osteosynthesis and invasiveness of the surgical procedure. Unfortunately, there is no common agreement between surgeons about which plate configuration brings the best clinical outcome. In this study, we investigate the theoretical rigidity of plate osteosyntheses considering two types of AO/ASIF configurations (90° angle between plates), Mayo clinic (Acumed) configuration (180° between plates) and dorsal fixation of both plates. We also compared the results for cases with and without contact between the bone fragments. In the case of no bone contact, the Mayo clinic plate configuration is found to be the most rigid, followed by both AO/ASIF plate configurations, and the least rigid system is the Korosec plate configuration. On the other hand, no significant differences between all types of fixation configurations are found in cases with contact in-between the bone fragments. Our findings show that this contact is very important and can compensate for the lack of load carrying capacity of the implants. This could therefore incite other implant fixation solutions, leading to less invasive surgical procedures and consequently improved clinical outcome.

Influence of chitosan and polycarbophil on permeation of a model hydrophilic drug into the urinary bladder wall.

Influence of dispersions of mucoadhesive polymers chitosan and polycarbophil on permeability properties of urinary bladder was investigated in vitro on isolated porcine urinary bladder. Pipemidic acid as a model hydrophilic drug was used. Its distribution in the bladder wall was determined from actual tissue concentrations by a method based on sectioning of frozen tissue and extraction of tissue slices. Pipemidic acid tissue concentration versus tissue depth profiles were evaluated by a diffusion model assuming constant diffusion coefficient. Increase in bladder wall permeability was observed in the presence of both polymers. Apparent permeability (mean+/-S.D.) of urinary bladder wall was increased 2.7+/-2.9 and 2.8+/-2.0 times for chitosan, and 2.3+/-2.0 and 4.3+/-4.2 times for polycarbophil at 0.5 and 1.0%, w/v polymer concentration, respectively. This increase is a consequence of the increased permeability of urothelium. These findings support investigations on application of chitosan and polycarbophil in development of mucoadhesive intravesical drug delivery systems. Experimental model may be applied to evaluate the results of experiments with drugs used in intravesical therapy.

Influence of physicochemical and biological parameters on drug release from microspheres adhered on vesical and intestinal mucosa.

The object of our work is to develop mucoadhesive microspheres to be applied into the urinary bladder. In the present study the microspheres were prepared and the release of a model drug after their adhesion to mucosa was evaluated. The microspheres were prepared by solvent evaporation method using Eudragit RL or hydroxypropylcellulose as matrix polymers and one out of five different polymers as mucoadhesives or non-mucoadhesive references. A method for the evaluation of the drug release from microspheres adhered on guinea pig urinary bladder and small intestine mucosa was developed and the influence of the following parameters on this process was followed: mucoadhesion strength of polymeric films, swelling of polymers and the drug release from microspheres. The results showed that the detachment forces were decreasing in the following order: CMCNa > Carbopol 934P > HPC > EE.HCl = PVP/VA. Carbopol swelled to the largest volume among all polymers and the drug release from microspheres was more retarded when Eudragit RL was used as matrix polymer. When comparing the results of pipemidic acid release from microspheres adhered on intestinal mucosa with detachment forces, similar ratios among the mucoadhesive polymers can be seen. On the other hand, differences between two mucosae were observed. These differences are due to the amount of mucus on mucosa and might also be influenced by the charge of mucus. The goal of our work at this point of investigation was achieved by microspheres containing carboxymethylcellulose as mucoadhesive and Eudragit RL as matrix polymer because they provide the longest release time from microspheres adhered on vesical mucosa and sufficient high strength of mucoadhesion.

The enhancement of pipemidic acid permeation into the pig urinary bladder wall.

The influence of interactions between polycarbophil and calcium on a model drug permeation into the pig urinary bladder wall was investigated. Pipemidic acid was used as a model drug. One percent w/v polycarbophil dispersion significantly increases the permeation of pipemidic acid into the urinary bladder wall. The enhanced absorption of pipemidic acid caused by polycarbophil is significantly less pronounced in polycarbophil dispersions containing calcium. The enhancement of pipemidic acid permeation into the urinary bladder wall could be due to the opening of tight junctions, which causes higher paracellular permeability. In the case of polycarbophil dispersion with calcium some carboxylic groups of polymer are already occupied with calcium, present in the dispersions. As a consequence extracellular calcium binds to polycarbophil in lower extent if compared with polycarbophil dispersion without calcium and transport is increased to a lesser degree. We concluded that the mechanism of drug absorption enhancement caused by polycarbophil could be similar for urinary bladder as described in the literature for intestinal mucosa.

Mucoadhesion on pig vesical mucosa: influence of polycarbophil/calcium interactions.

The influence of polycarbophil/calcium interactions on the mucoadhesive properties of polycarbophil has been examined. Polycarbophil dispersions and films with different concentrations of calcium or sodium ions were prepared and the following parameters were measured: detachment force on pig vesical mucosa, zeta potential, pH and viscosity. Polycarbophil detachment force decreased significantly in the presence of calcium but not sodium. Both ions decrease the pH of polycarbophil dispersions. On the other hand, altering the pH of hydrated polycarbophil films in the absence of added ions had an insignificant effect on detachment force. Both ions reduce the absolute values of polycarbophil zeta potential, calcium more efficiently than sodium. We could conclude that decreased mucoadhesion strength of polycarbophil in the presence of calcium is due to the chelation of polycarbophil carboxylic groups by calcium and crosslinking of polymer. The crosslinked polymer chains would be expected to be less flexible, and therefore, interpenetrate to a lesser extent with the glycosaminoglycans of mucus. Additionally, the interactions between functional groups of polycarbophil and mucus glycosaminoglycans are lowered due to the calcium, blocking the carboxylic groups. The mechanism of calcium influence on viscosity of polycarbophil dispersions appears to be different: repulsion between ionised carboxylic groups of polycarbophil prevails over the crosslinking of polycarbophil by calcium.

The study of drug release from microspheres adhered on pig vesical mucosa.

The object of our work is the preparation of a mucoadhesive drug delivery system intended for intravesical application. In the present work, microspheres with Eudragit RS matrix polymer and different mucoadhesive polymers, i.e. chitosan hydrochloride (Ch), sodium salt of carboxymethyl cellulose (CMC) and polycarbophil (PC) were prepared to evaluate their influence on the mucoadhesive properties of microspheres. Different parameters were determined and their influence on pipemidic acid release from microspheres adhered on intact and damaged pig vesical mucosa was evaluated: swelling of polymers, mucoadhesion strength of polymeric films and drug dissolution according to USP XXIV method. The dissolution rate from microspheres containing different mucoadhesive polymers decreases as follows: PC>Ch>CMC. PC swelled to the largest volume among all polymers and as a result the fastest release of the drug from PC microspheres was obtained. The release rate of pipemidic acid from microspheres adhered on intact mucosa followed the order PC>CMC>Ch. These results show that both drug dissolution and mucoadhesion strength strongly influence drug release from adhered microspheres. The slowest release from Ch microspheres could be interpreted by the largest mucoadhesion strength of Ch polymeric films. The release rate of pipemidic acid from microspheres adhered on damaged mucosa followed the order PC=Ch>CMC. The results obtained on pathologically changed mucosa model support the indication of the role of glycosaminoglycans and polymer charge in the mucoadhesion process on vesical mucosa. Analysis of release data shows that the drug dissolution profiles follow the Higuchi kinetics better than the release profiles from adhered microspheres and different kinetics might be a consequence of different release mechanisms.

Formulation of controlled release microspheres containing nicardipine: the role of pharmacokinetic modeling and computer simulation.

Nicardipine is an antihypertensive drug of the dihydropyridine series. It has high solubility in an acidic and low solubility in an alkaline medium. It is rapidly absorbed, extensively presystemically metabolized and excreted in the urine and faeces, mainly as inactive metabolites. Since the duration of its action can be extended by prolonging the absorption interval, the design of controlled release formulation is reasonable. The aim of the present study was to prepare microspheres which would release nicardipine at a decreased rate in gastric and increased rate in intestinal juice during a 12 h interval. Pharmacokinetic modeling based on compartment analysis and supported by analog computer and digital simulation technique showed that the target steady state peak plasma concentrations of 32 microg/l and trough plasma concentration of 7 microg/l would be maintained if nicardipine were incorporated in a formulation releasing the drug as follows: 25% after 1 h, 40% after 2 h, 65% after 4 h, 80% after 6 h, 90% after 8 h and 100% by 12 h. Microspheres have been prepared from hydroxypropylmethylcellulose phthalate polymer using the solvent evaporation method. Drug content, scanning electron micrographs, particle size distribution and dissolution profile were determined. In vitro nicardipine release was described by a biphasic square root of time kinetics and was in accordance with the above values relating to the dissolution. Furthermore, a composed first-pass pharmacokinetic model with derived release function as an input was developed to predict nicardipine plasma concentrations after single- and 12 h multiple-dosage-regimen scheme administration of controlled release microspheres.

Biopharmaceutic evaluation of some 4-quinolone derivatives.

The relationships between physicochemical properties (ionization, solubility, partition, protein binding) of 4-quinolone derivatives and their pharmacokinetics was studied by means of in vitro methods. This information is needed to clarify the erratic absorption behaviour of quinolone antibacterials following oral administration. The ionization constants of ofloxacin (OF), pefloxacin (PF), norfloxacin (NF) and ciprofloxacin (CF) were determined by solubility, potentiometric, spectrophotometric, and octanol-water partition methods. The quinolones are ampholytes showing small values of partition coefficients (order of magnitude 1) resulting from their hydrophilic nature. Within the series, both partition coefficients and the ionization constants differ only slightly (5.5 less than pK1 less than 6.5, and 7.5 less than pK2 less than 8.5). The intrinsic solubility of nonionized forms of drugs was obtained to be of 1g/l order of magnitude. Linear relationship was observed between the log solubility versus reciprocal temperature. The dissolution of NF is an endothermic reaction with dissolution enthalpy H = 20.4 kJ/mol. In vitro diffusion of OF, PF, NF, and CF was studied by the means of Sartorius in vitro absorption model. The rate of diffusion and the projected in vitro rate of absorption are not functions of structural differences within the series (diffusion rate constant range was 0.5 less than kd less than 5.0 .10-3 cm/min). Binding of NF, CF, PF and OF to bovine serum albumin (BSA) was studied using equilibrium dialysis in the absence and presence of urea. It was found that the quinolones were moderately bound (40-60%) to BSA and that urea in pathologic concentrations diminished their binding for 10-25%.

Prediction of the in vivo performance of enteric coated pellets in the fasted state under selected biorelevant dissolution conditions.

The purpose of this research was to predict the in vivo dissolution of lansoprazole from enteric coated pellets in the fasted state using a biorelevant flow-through dissolution method with low flow rates and volumes close to those in vivo. Additionally, a novel rotating stirring element, composed from magnet inserted in a silicone tube, was used to produce the movement of the pellets and expose them to slightly increased physical stress. Obtained dissolution results were compared to the dissolution results of our previous work using the USP IV with higher flow rate (11 ml/min). As drug release from enteric coated pellets usually starts in the small intestine, the influence of pellets' residence time in the gastric medium and additionally the effect of different media on drug release was studied. Prolongation of residence time in an acidic medium had only minor effect on the release rate after initial lag time, but significantly reduced the total amount of the drug released from both tested formulations, which was attributed to the drug's degradation in an acidic medium. The increased physical load on the pellets induced by the rotating stirring element compensated for the decrease of flow rate from 11 ml/min using the USP IV to 3 ml/min using the non-compendial system. Considering also gastric emptying kinetics good prediction of the in vivo release was achieved compared to in vivo absorption data obtained from a pharmacokinetic study under fasting conditions. Thus, using more physiologically relevant dissolution conditions, expressed through low volume and lower flow rate, and in combination with increased mechanical stress we obtained equally good in vitro/in vivo correlation as using USP IV and higher flow rates. Comparison of the dissolution results obtained with two different systems provided additional insight into product behaviour and improved prediction of in vivo performance.

The influence of gastric emptying kinetics on the drug release from enteric coated pellets in fasted state: an in vitro/in vivo correlation.

The absorption and bioavailability of drugs can be substantially affected by the transit of dosage forms through the gastrointestinal (GI) tract. Gastric emptying is one of the most critical parameters contributing to this inter- and intra-individual GI transit variability. It is especially important for the delayed release dosage forms whose release depends on the local environment and begins when the dosage form passes pylorus and comes into contact with higher pH medium in small intestine. The purpose of our research work was to predict the in vivo dissolution from enteric coated pellets for population and establish a good in vitro/in vivo correlation (IVIVC) with mean in vivo absorption profiles, obtained in a pharmacokinetic study under fasting conditions. The dissolution tests were carried out on a USP 4 - flow-through cell with enteric coated pellets containing an acid-labile drug and formulated as orodispersible tablets. Using several residence times in an acidic medium, we simulated the gastric emptying of the pellets and the exposure of different fractions of the pellets to the gastric medium for different periods of time. The amount of drug released decreased with the increasing time of exposure to the acidic medium due to the drug's degradation. The mean in vivo dissolution profiles, which were predicted on the basis of experimentally determined dissolution profiles and mathematical model of pellets' gastric emptying, gave a very good IVIVC with the mean in vivo absorption profiles.

Characterization of calcium alginate beads containing structurally similar drugs.

The aim of this study was to investigate the characteristics of alginate beads prepared by ionotropic gelation in which structurally similar drugs were incorporated. For this purpose theophylline and theobromine were selected as model drugs. The influence of incorporated drugs on bead characteristics such as size, shape, and morphology, as well as encapsulation efficiency, was examined. It was found that theobromine as well as theophylline content in beads significantly decreased with increasing hardening time due to drug diffusion into the hardening media. In theobromine beads the drug content was extremely improved by dropping the alginate and drug solution into an acidic calcium chloride solution, while theophylline content was to some extent improved by the hardening of beads in a calcium chloride solution saturated with the drug. The most evident difference between theophylline and theobromine beads was in their shape and morphology. Theobromine beads were round, while theophylline ones had an irregular shape with an extremely wrinkled surface. The distinction in shape was highly dependent on drug content. Additionally, it was demonstrated that beads' shape was dependent on preparation conditions as well. On the basis of x-ray powder diffraction (XRPD) and differential scanning calorimetry (DSC) analyses and scanning electron microscope (SEM) photographs it was found that the most of the drug in bead was present in an amorphous state. Therefore, it is suggested that some drug-alginate interactions could be present in beads and might be responsible for the different shape of theophylline and theobromine beads. Thus it can be concluded that the preparation of beads by ionotropic gelation cannot be generalized even though structurally similar drugs are incorporated.

Determination of microsphere solidification time in the solvent evaporation process.

The aim of this work was to define the time of microsphere solidification during the solvent evaporation process. Microspheres were prepared by the solvent evaporation method, using acetone/liquid paraffin solvent system, ketoprofen as a model drug and Eudragit RS as a matrix polymer. Two sets of experiments were performed--in the first one the initial temperature of the emulsion system was 5 degrees C and in the second one 25 degrees C. In each set, two batches of microspheres were compared at constant emulsion stirring rate 250 and 1000 rpm and intermediate batches where the emulsion stirring rate was lowered from 1000 to 250 rpm at pre-defined times after the beginning of the inner phase solvent evaporation. By comparison of the properties of these microspheres, an insight was obtained into the mechanism of microspheres formation. The criterion for determination of microsphere solidification time was the resemblance between the microsphere properties of the batches prepared by stirring rate change and the batch prepared by constant stirring at 1000 rpm. A stirring rate change after the solidification has no influence on microsphere properties, that means that they are the same as of the batch prepared by constant stirring at 1000 rpm. The results of the sieve analysis and particle size distribution of microspheres show that the time of microspheres solidification is in the interval between 15-20 min if the initial temperature is 5 degrees C and between 10-20 min if the initial temperature is 25 degrees C. From the release profiles of ketoprofen, one can infer that the times of solidification for both initial temperatures are a bit lower. The microscopic pictures, which enable one to follow the processes in the system, confirmed the result obtained by the sieve analysis. In spite of its inability to distinguish between single particles and agglomerates, the sieve analysis enabled one to determine the actual time of solidification, while the drug release determination was not sensitive enough to trace small differences in surface area due to particles aggregation.

Eudragit E microspheres containing bacampicillin: preparation by solvent removal methods.

Eudragit E microspheres containing bacampicillin hydrochloride were prepared by solvent evaporation and solvent extraction methods. Three different systems of solvents were used: methyl acetate, acetone and methanol/liquid paraffin. The success of the procedures depended mostly on the lipophilicity of the solvent. The particle size of microspheres was determined by sieve analysis. The results showed that the average size of microspheres is influenced greatly by the type of solvent. Scanning electron microscopy was used for observation of the shape of microspheres. Microspheres prepared by the solvent evaporation method in systems with acetone and methyl acetate were all of a regular spherical shape. The surface of all other microspheres were folded. The influence of magnesium stearate content in microspheres was also studied in terms of different methods, solvents and processing conditions.

Formulation and evaluation of vinylpyrrolidone/vinylacetate copolymer microspheres with griseofulvin.

Regular spherical microspheres of 220-260 microns average size have been prepared from vinylpyrrolidone/vinylacetate copolymer using a solvent evaporation method. Griseofulvin has been incorporated into these microspheres and its physical characterization has been carried out by differential scanning calorimetry (DSC), X-ray diffractometry and X-ray photoelectron spectroscopy. An increase of solubility was observed only with the 1:3 drug/polymer microspheres and the comparison of the dissolution profiles of microspheres with pure griseofulvin resulted in an enhancing effect. Furthermore the release rate of griseofulvin, incorporated into the microspheres, was shown to be biphasic and dependent upon the penetration of water into the microspheres, hydration and dissolution of the polymer and finally dissolution of the drug.

The influence of stirring rate on biopharmaceutical properties of Eudragit RS microspheres.

Eudragit RS microspheres containing pipemidic acid, as a model drug, were prepared by the solvent evaporation method using an acetone/liquid paraffin solvent system. The aim of the work was to evaluate the influence of stirring rate on the average particle size, particle morphology, drug content and release kinetics, as well as the influence of particle size on microsphere morphology, drug content and release kinetics. Stirring rate has been found to significantly influence the average diameter of microspheres. The average diameter decreases as the stirring rate increases. This can be explained by production of a finer dispersion of droplets when higher stirring rates are applied and, consequently, by the formation of smaller microspheres. With increasing stirring rate and increasing fraction particle size the drug content also increases. It is assumed that this dependence is a consequence of an uneven diffusion of the drug from the inner to the outer emulsion phase, and an uneven encapsulation of drug particles during the preparation. Drug release follows the Higuchi model. As seen from SEM photographs, larger microspheres are more porous and the microspheres produced at higher stirring rates are more porous than those produced at lower stirring rates. This explains the unexpected finding that the release rate increases as the fraction particle size and the stirring rate increase.