Langmuir 2006, 22, 911-916
911
Amorphous Drug Nanosuspensions. 2. Experimental Determination of Bulk Monomer Concentrations Lennart Lindfors,*,† Sara Forsse´n,† Pia Skantze,† Urban Skantze,† Anna Zackrisson,‡ and Ulf Olsson§ Pharmaceutical and Analytical R&D, AstraZeneca R&D Mo¨lndal, Sweden, and Physical Chemistry 1, UniVersity of Lund, Sweden ReceiVed August 30, 2005. In Final Form: NoVember 1, 2005 A simple turbidimetric method was developed to measure the bulk concentration of drug in nanosuspensions. The bulk concentrations measured were in the range from 1 µM to 1 mM. The accuracy of the method was checked by determination of the bulk concentration of crystalline nanosuspensions, i.e., the crystalline solubility, which compared favorably to solubilities measured by a conventional method. Results obtained for amorphous nanosuspensions agreed with predictions using a theory describing the relative solubility between a supercooled liquid and a crystal. Further, it was found that the bulk concentration in Ostwald ripening inhibited amorphous nanosuspensions and could be lowered by incorporation of higher amounts of the inhibitor, in agreement with predictions using the Bragg-Williams theory of nonideal solutions.
1. Introduction Amorphous nanosuspensions of drugs are attractive formulations in many cases. On a lab-scale, they are prepared by rapid mixing of a drug dissolved in an organic water-miscible solvent and an aqueous stabilizer solution. Furthermore, they contain low amounts of additives and are, thus, expected to induce minimal side effects in various in vivo studies. One problem with these suspensions is the more or less pronounced tendency to crystallize and also the Ostwald ripening process that will take place unless proper inhibitors are added.1 Both these phenomena are likely to be related to the bulk concentration of the drug in the suspensions. Therefore, it is important to determine the bulk concentration in such systems. In conventional solubility experiments, a saturated solution is separated from excess particles by, e.g., filtration or centrifugation, and the concentration in the filtrate or supernatant is then determined by, e.g., spectroscopic methods.2 However, filtration, or more specifically ultrafiltration, has proven to be difficult with nanosuspensions, probably because of particles blocking the pores of the membrane. Moreover, both separation methods lead to high local particle concentrations that may induce crystallization in the case of amorphous nanosuspensions. In this paper, we have investigated an alternative method of determining the solubility, based on the scattering of light from colloidal drug nanosuspensions. The basic idea is simple, and the method has been used for example to determine the critical micellization concentration (cmc) of surfactants.3,4 The scattered light intensity depends strongly on the particle size. Although the scattering from a molecular dispersed solution in general is very small, a colloidal dispersion of the same concentration may * Corresponding author. Telephone: +46 31 7761000. Fax: +46 31 7763834. E-mail:
[email protected]. † AstraZeneca R&D Mo ¨ lndal. ‡ Present address: Physical Chemistry, University of Go ¨ teborg, Sweden. § University of Lund. (1) Lindfors, L.; Hedberg, P.; Skantze, U.; Rasmusson, M.; Zackrisson, A.; Olsson, U. Langmuir 2006, 22, 906. (2) Yalkowsky, S. H.; Banerjee, S. Aqueous Solubility: Methods of Estimation for Organic Compounds; Marcel Dekker Inc.: New York, 1992. (3) Lindman, B.; Wennerstro¨m, H. Current Topics in Chemistry; SpringerVerlag: Berlin, 1980; p 83. (4) Kato, T.; Kanada, M.; Seimiya, T. Langmuir 1995, 11, 1867.
have a significant turbidity.5 Hence, if colloidal particles are gradually added to a solvent, they will at concentrations below the solubility dissolve and the solution will essentially not scatter light. As the concentration is increased above the solubility, however, the solution is saturated with “monomers” and the added particles no longer dissolve. The scattered intensity then increases significantly and, initially, linearly with the particle concentration, and the solubility is determined from the onset of intensity increase. The method is of course also applicable to liquid-liquid dispersions, and a similar approach was used to determine the solubility of a hydrocarbon in a micellar solution.6 Here we have investigated six different drug compounds, the molecular structure of which are given in Figure 1. Both crystalline and amorphous nanoparticles were prepared, where the amorphous particles in general also contained a second, water insoluble, compound to inhibit Ostwald ripening. In Figure 2, we show Cryo-TEM micrographs of crystalline and amorphous nanosuspensions, respectively, of the compound felodipine. The Ostwald ripening rate depends on the solubility, and for the compound C4, the solubility, and hence the ripening rate, was so low that inhibition was not necessary. After a description of the experimental conditions in section 2, we present and discuss our experimental results in section 3, where we also will discuss how the amorphous and crystalline solubilities are related. To our knowledge, the data presented here represents the first direct measurement of amorphous drug solubilities. 2. Experimental Section 2.1. Chemical Components. Nifedipine was purchased from Sigma, whereas felodipine, bicalutamide, C1, C2, and C4 were obtained from AstraZeneca. All drug structures are shown in Figure 1. N,N-Dimethylacetamide, DMA (Aldrich), sodium dodecyl sulfate, SDS (Millchem UK Ltd), the disodium salt of Aerosol OT, AOT (Cytec Industries Inc.), poly(vinylpyrrolidone) K30, PVP (BASF), Miglyol 812N (Hu¨ls, an approximately 60/40 (w/w) mixture of C8 and C10 triglycerides), and 1-decanol (Aldrich) were used as received. 2.2. Preparation of Amorphous Nanosuspensions. Amorphous nanosuspensions of the drug compounds were prepared by rapidly (5) See, for example: Hunter, R. J. Introduction to Modern Colloid Science; Oxford University Press Inc.: New York, 1993. (6) Kabalnov, A. S. Langmuir 1994, 10, 680.
10.1021/la052367t CCC: $33.50 © 2006 American Chemical Society Published on Web 12/21/2005
912 Langmuir, Vol. 22, No. 3, 2006
Figure 1. Molecular structures of the drug compounds chosen for this study.
Figure 2. Cryo-TEM images of (a) a 10 mM crystalline nanosuspension of felodipine and (b) a 5 mM amorphous nanosuspension of felodipine. injecting a drug solution (typically 100 mM drug dissolved in DMA) into an aqueous stabilizer solution in a vial placed on an ultrasonic bath (Elma Transsonic Bath T460/H). The stabilizer solutions
Lindfors et al. contained 0.2% (w/w) PVP and 0.25 mM SDS in all cases. Ostwald ripening inhibitors used were Miglyol (nifedipine, felodipine and bicalutamide) and a Miglyol/decanol 1:1 (w/w) mixture (C1 and C2). The drug/inhibitor ratio was in all cases 4:1 (w/w). For the compound C4, no inhibitor was used. The average particle size of the amorphous suspensions was measured by fiber optic quasi elastic light scattering (FOQELS, Brookhaven Instruments Corporation), and the particle sizes reported are averages computed by the second cumulant method. 2.3. Preparation of Crystalline Nanosuspensions. A crude suspension of 10% (w/w) drug in 1.33% (w/w) PVP K30, 0.067% (w/w) AOT was made by sonication in an ultrasonication bath. This suspension was milled for 4 × 30 min, with 15 min pauses, at 700 rpm using the Fritsch Planetary Micromill P7 equipped with 3.6 mL milling bowls and 0.6-0.8 mm milling beads of zirconium oxide. The volume averaged particle size (diameter) of the crystalline suspensions was measured using a Malvern Mastersizer 2000, to confirm that the amount of particles larger than 1 µm was negligible. 2.4. Light Scattering Method. Small volumes of drug suspension were successively added directly to a fluorescence cuvette containing pure liquid (water, 1% (v/v) DMA or 10% (v/v) DMA, respectively) and mixed to give the desired concentrations. The light scattering intensity at 700 nm was recorded at a scattering angle of 90 ° as a function of total drug concentration. As a light scattering setup, we here used a Perkin-Elmer LS 55 Luminescence Spectrometer, setting both the emission and excitation wavelengths to 700 nm.7 For ionizable compounds, special precautions were taken to measure the solubility of the neutral forms (e.g., by measurements of solution pH). In some experiments the crystalline solubility was studied as a function of temperature. In those cases crystalline nanosuspensions with different concentrations were prepared by dilution. The water used was cooled to a temperature below 10 °C prior to the dilution, and the vials directly transferred to a temperature controlled water bath to ensure that the nanoparticles did not dissolve. The temperature of the suspensions was then increased from 10 to 80 °C in steps of 5 °C. At each temperature the suspension was equilibrated for at least 15 min after which the scattering intensity was recorded. With increasing temperature the solubility increases and hence the scattered intensity decreases. For a given concentration, the temperature above which all the particles were dissolved, was determined from a plot of the scattered intensity versus temperature. Above this temperature the scattered intensity leveled off at a constant low value. 2.5. Conventional Solubility Measurements. The crystalline solubility of drugs in aqueous solutions was determined by adding an excess of the drug to typically 1 mL of water. The mixture was allowed to equilibrate using magnetic stirring for at least 48 h at room temperature (∼23 °C) and was then passed through a 0.2 µm hydrophilic PTFE filter. The concentration was determined by UVHPLC. 2.6. Differential Scanning Calorimetry Measurements. The melting temperature, Tm, and enthalpy of melting, ∆Hm, were determined by differential scanning calorimetry (DSC) analysis on samples of crystalline drug material using a Mettler-Toledo DSC 820 in a open vial configuration and a scanning speed of 10 K/min. The entropy of melting, ∆Sm, was calculated as ∆Sm ) ∆Hm/Tm. 2.7. Cryo-TEM Studies. Cryo-TEM (cryo transmission electron microscopy) images were taken on nanosuspensions at 25 °C in CEVS (controlled environment vitrification system). The samples were applied as a thin film on a metal plate coated with a porous polymer film, vitrified in liquid ethane at -170 °C, and studied at the boiling temperature of nitrogen in a Zeiss EM 902 (accelerator voltage 80 kV).
3. Results and Discussion Solubility of the Crystalline State. In the solubility experiments, a small amount of a concentrated particle suspension was (7) Mouga´n, M. A.; Coello, A.; Jover, A.; Meijide, F.; Tato, J. V. J. Chem. Ed. 1995, 72, 284.
Bulk Concentrations in Nanosuspensions
Figure 3. Measured scattering intensity vs total felodipine concentration for (a) a crystalline nanosuspensions in water (9) and (b) in 10% (v/v) DMA (0). The solid lines are linear fits to the high concentration data. The crystalline solubility is estimated from extrapolating the line to zero intensity.
successively added directly to a fluorescence cuvette containing the solvent. The solvents investigated here were pure water and 1 and 10% (v/v) solutions of N,N-dimethylacetamide (DMA) in water. The outcome of the light scattering experiments actually depends on the protocol for how the concentrations are varied. By gradually increasing the concentration, the bulk is gradually saturated with drug as the concentration is increased up to the solubility, and a further increase of the concentration does not dissolve the particles. If on the other hand a concentrated suspension would have been diluted, the particles would gradually decrease in size as the drug would dissolve to saturate the growing bulk. DMA was used in the preparation of amorphous nanosuspensions, and it is therefore of interest to know how solubilities depend on the DMA concentration. In Figure 3, panels a and b, we show the results from solubility experiments on crystalline felodipine in pure water and 10% (v/v) DMA, respectively. As can be seen, the light scattering intensity remains low up to a certain concentration, above which it increases approximately linearly with the concentration. The concentration onset of the increase in intensity can be identified
Langmuir, Vol. 22, No. 3, 2006 913
with the solubility of the drug in the solvent. For ctot < Sc, where ctot denotes the total drug concentration and Sc is the crystalline solubility, the particles are completely dissolved in the solvent and essentially no light scattering is obtained. For ctot > Sc, on the other hand, the bulk is saturated with drug and additionally added particles remain in solution and scatter light. For low particle concentrations, interparticle interactions can be neglected, and we expect the light scattering intensity to increase linearly with the particle concentration. It should be noted however, that the particles are electrostatically stabilized by adsorbed ionic surfactant, and the linear regime may be small. The solubilities are evaluated from the experiments in Figure 3, panels a and b, by fitting the high concentration data to a straight line. With this procedure, we obtain Sc ) 2.6 µM for the solubility in water and Sc ) 23 µM for the solubility in 10% (v/v) DMA. To check the reproducibility of the method, the experiment with crystalline nanoparticles of felodipine in water was repeated four times at different occasions (different millings) and by different experimentalists, giving Sc ) 2.1 ((0.5) µM with the standard deviation given within brackets. The relative standard deviation obtained was thus, ∼25%, which may serve as an estimate of the uncertainty of the method for all of the different nanosuspensions. With this assumption, the results agree with the corresponding values obtained using the conventional method, Sc ) 2.1 µM in water and Sc ) 28 µM in 10% (v/v) DMA. It should, however, be noted that the solubility data obtained with the light scattering method will, of course, never be more accurate than the quantitative analysis, here performed using HPLC, of the stock nanosuspensions (prior to dilutions). Temperature Dependence of the Crystalline Solubility. One particular advantage with the light scattering method over the conventional one is that it is simple to use also at elevated temperature. With the conventional method, the excess crystals, after equilibration, have to be separated from the bulk solution before it is analyzed, and both the separation step and the subsequent analysis have to be performed under controlled temperature. Using the light scattering method, a concentration series of samples can be prepared, on which the light scattering is recorded as a function of temperature. The solubility in general increases with increasing temperature, and, for a given particle concentration, the intensity decreases with increasing temperature as the particles gradually dissolve. Eventually, the particles are dissolved completely, and the light scattering intensity essentially vanishes, or levels off at a very small value. As an example, we show in Figure 4 the intensity versus temperature for a nanosuspension of crystalline bicalutamide having a concentration of 52 µM in water. From the curve in Figure 4, we obtain that this concentration corresponds to the solubility at 62 °C. In Figure 5, we have plotted the obtained solubilities for two substances, bicalutamide and felodipine, respectively. The two compounds have significantly different solubility curves, reflecting the relative stability of their crystalline states.8 Solubility of the Amorphous State. The solubility of the amorphous state represents an experimental challenge, that is difficult to obtain with the conventional method. It is difficult to obtain sufficiently large amorphous particles that can be filtered and the amorphous state is often labile and crystallization may be induced by mechanical shear or other disturbances. Here, we also performed light scattering experiments on amorphous drug nanosuspensions. In Figure 6, we show intensity versus concentration obtained for amorphous nanosuspensions of felodipine in pure water (Figure 6a) and in 10%(v/v) DMA (Figure 6b), (8) Neau, S. H.; Bhandarkar, S. V.; Hellmuth, E. H. Pharm. Res. 1997, 14, 601.
914 Langmuir, Vol. 22, No. 3, 2006
Lindfors et al.
Figure 4. Measured scattering intensity vs temperature for a sample of crystalline nanoparticles of bicalutamide (52 µM) (9) in pure water. The solid line are linear fits to the high-temperature data. The temperature at which the prepared concentration dissolved is estimated by extrapolating to the line of zero intensity to 62 °C.
Figure 6. Measured scattering intensity vs total felodipine concentration for (a) an amorphous nanosuspension in water (b) and (b) in 10% (v/v) DMA (O). The solid lines are linear fits to the high concentration data. The amorphous solubility is estimated from extrapolating the line to zero intensity. Figure 5. Measured crystalline solubility vs temperature using crystalline nanosuspensions of felodipine (9) and bicalutamide (0).
respectively. Here, the amorphous particles contain, besides felodipine, also Miglyol as an additive to inhibit Ostwald ripening.1 The weight ratio of felodipine-to-Miglyol is 4:1. Comparing the data with the corresponding data for the crystalline particles, we see directly that the amorphous solubility is approximately 10 times higher that the crystalline solubility. As for the crystalline solubility, the result for the amorphous particles is also an approximately 10 times increase in the solubility when water is replaced by a 10% (v/v) DMA solution, as solvent. From an extrapolation of the high concentration data in Figure 6, panels a and b, we estimate the solubilities of felodipine to be 21 µM in water and 0.25 mM in 10% (v/v) DMA, respectively. The presence of the second component (Miglyol) in the particles affect the drug solubility. We denote the pure drug amorphous solubility S0a and the amorphous solubility from the composite particles Sa. For Ostwald ripening inhibition to work, the second component has to be miscible with the (amorphous) drug, which
also implies that Sa e S0a , because mixing in a second component lower the drug chemical potential. If we model the composite particles with the Bragg-Williams model of regular solutions,9 we have
Sa(xd) ) S0a xd exp{χ(1 - xd)2}
(1)
Here, xd is the mole fraction of drug in the particles and χ is the interaction parameter. The particles that we add to the solvent have initially a composition x0d and it is essentially the solubility of these particles, Sa(x0d), that we estimated from the high concentration data of Figure 6. However, when these particles are added to water at low concentrations,
