WO2006051067A1

WO2006051067A1 - A process for preparing formulations of lipophilic active substances by spray freeze drying

Info

Publication number: WO2006051067A1
Application number: PCT/EP2005/055805
Authority: WO
Inventors: Dirk J. Van Drooge; Wouter L. J. Hinrichs; Henderik W. Frijlink; Gerrit Sietse Zijlstra; Bastiaan H. J. Dickhoff
Original assignee: Solvay Pharmaceuticals B.V.; Rijksuniversiteit Groningen
Priority date: 2004-11-10
Filing date: 2005-11-08
Publication date: 2006-05-18
Also published as: CA2585314A1; JP5137579B2; AR052787A1; TW200621310A; EP1811965A1; JP2008519806A; CN101056622A

Abstract

The present invention is related to a method for the preparation of a pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugars alcohols, wherein the lipo philic compound is incorporated in the sugar glass, characterized in that a. said lipophilic compound is dissolved in an organic solvent that is miscible with water and said sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols is dissolved in water; b. the dissolved lipophilic compound and the dissolved sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols are mixed in such a way that a homogeneous mixture is obtained; c. said mixture is spray freeze dried. The invention is further related to pharmaceutical formulations formed containing lipophilic compounds such as Δ9-tetrahydrocannabinol, diazepam and cyclosporin A.

Description

A PROCESS FOR PREPARING FORMULATIONS OF LIPOPHILIC ACTIVE SUBSTANCES BY SPRAY FREEZE DRYING

[0001] The present invention is related to a process for preparing pharmaceutical 5 formulations having improved stability and flow properties by spray freeze drying lypophilic active subtances such as natural cannabinoid compounds, especially Δ⁹- tetrahydrocannabinol (THC). The invention further relates to the formulations obtained having specific particle properties.

10 [0002] WO03/082246 describes the use of a stable sugar based solid dispersion of a lypophilic substance, that preferably is obtained by freeze drying, from a mixture obtained by mixing a solution of the sugar in water with a solution of the lypophilic substance in an organic solvent miscible with wate r. In this patent application the oligo-fructose inulin is shown to be an excellent stabilising carrier due to its high

15 glass transition temperature. It is also shown that adequate incorporation in inulin glasses prevents oxidation of the active compounds which are vulnerable for oxidation. Furthermore, the lyophilized material can be dry granulated to obtain a free flowing powder for direct compression . The above described formulation can be used to prepare oral dosage forms and to prepare a product suitab le for pulmonary

20 administration. In this patent application also spray drying is used, but when using this drying technique no true and complete solid dispersion is obtained, leading to higher decomposition rates during storage than with freeze drying.

[0003] Although the method described in WO03/082246 leads to a product having

25 excellent properties when prepared by freeze drying , it has the serious drawback that it cannot be scaled up easily. The mixture of the hydrophilic sugar in water and the lipopilic compound in an organic solvent miscible with water used for freeze drying has the tendency to give phase separation after a certain period of time, as it is thermodynamically metastable. Only when the solution is rapidly frozen below the

30 Tg', phase separation is prevented. Once the solution is frozen , molecular mobility has decreased to such extend that phase separation is no longer possible. Also when the product is dried and in the glassy state phase separation is not possible.

However, prevention of phase separation during the production process is only obtained when the cooling process is performed rapidly. When cooling of the solution

35 is slow, phase separation of the lipophilic or hydrophilic solutes may occur because of the increased tendency for phase separation at lower temperatures. It is therefore of paramount importance that the solution is cooled to a temperature below the Tg' before phase separation of one of the solutes has occurred. If phase separation occurs, a true solid dispersion is not obtained and patches that consist of clusters of non protected lipophilic compound will occur in the solid dispersion. This will lead to loss of the advantages that are obtained by the formation of true solid dispersions , such as an improved stability against oxidation and decomposition during storage .

[0004] Therefore the freezing of the mixture should take place on a very short term after mixing of the two solutions. This can easily be done on a small scale, but it is difficult to perform this process with large volumes as the freezing time will be too large. Therefore scaling up of a normal freeze drying process can only be performed when using a multitude of small scale batches. Alternatively it has to be accepted that some phase separation takes place, leading to less stable products.

[0005] It is the objective of the present invention to prepare a formulation of a lipophilic substance having good stability , fast aqueous dissolution and improved bioavailability, that can be produced on an industrial scale. It is a further objective of the present invention to prepare a formulation with very fine spherical particles that have advantageous dispersion and aerodynamic properties which make them for example highly suitable for pulmonary administration.

[0006] In the framework of the present invention the expression lipophilic active compound means an active compound having a solubility in water lower than 1 mg/ml. The invention is especially useful for active compounds having such a low solubility in water but is even more useful for compounds having a solubility in water lower than 0.5 mg/ml or even lower than 0.1 mg/ml. Examples of lipophilic active compounds are Δ⁹-tetrahydro-cannabinol, diazepam and cyclosporin A.

[0007] It has surprisingly been fou nd that spray freeze drying of a lipophilic compound from a mixture of a sugar solution in water and a solution of the lipophilic compound is an organic solvent miscible with water, yielding a sugar glass, can be performed on a large scale leading to a product having the desired properties and even superior properties compared with the product obtained according to the freeze drying method disclosed in WO03/082246 , such as better stability, an optimal aerodynamic particle size and aerodynamic particle size distribution for pulmonary administration without any particle size reduction, easier de-agglomeration when used for pulmonary administration . [0008] In the spray freezing technique small droplets of the , shortly before, mixed solution are sprayed into cold (temperatures far below Tg') gas (e.g. air) or a cold fluid (e.g. liquid nitrogen). The droplet size of the aerosol depends on different factors such as the intended use of the particles and the amount of solid material in the solution. In general the size of the aerosol droplets will be between 1 and 5000 μm, preferably between 1 and 500 μm and most preferred between 5 and 500 μm . With this freeze drying technique large volumes of the solution can be frozen and further freeze dried to form a powder. Moreover, high concentrations of the solutes can be applied, provided spraying is performed fast enough to prevent phase separation after mixing of both solutions.

[0009] The spray freezing technology is known from the prior art. Y -F Maa and SJ. Prestrelski (Current Pharmaceutical Biotechnology 2000, 1, 283-302) and Y-F Maa et al., Pharm. Res. 1999, ^5, 249-54 describe the use of different powder production techniques for biopharmaceutical powders such as protein/peptide based drug formulations, which have to be regarded as powders containing hydrophilic active substances. In the first publication it is concluded that for convenience and simplicity lyophilization and spray drying are the methods of choice for the formation of biopharmaceutical powders. As some proteins are thermally labile and cannot resist heat denaturation by hot air, for these proteins spray freeze drying may be the method of choice. The second publication compares the two processes of spray drying and spray freeze drying in the preparation of inhalation powders. In this publication it is stressed that the spray freeze drying process might involve more stressful events which might affect protein's stability than the spray drying process. In both papers the stability is related to effect of the drying process on the quality of the product formed during the drying process . Both papers are, however, silent about the stability of the powder product during storage.

[0010] The present invention relates to a method for the preparation of a pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugars alcohols, wherein the lipophilic compound is incorporated in the sugar glass, characterized in that a) said lipophilic compound is dissolved in an organic solvent that is miscible with water and said sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols is dissolved in water; b) the dissolved lipophilic compound and the dissolved sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols are mixed in such a way that a homogeneous mixture is obtained; c) said mixture is spray freeze dried.

[0011] Mixing of the two solutions is preferably done continuously or semi- continuously just before spraying of the droplets of the mixture of the active compound and the sugar(s) or sugar alcohol(s). Semi -continuously in the framework of the present invention means that the two solutions are made batch wise, but that mixing and spray drying preferably takes place continuously until the solutions are fully used. The time between mixing and spray freeze drying is preferably shorter than 15 minutes , more preferably shorter than 10 minutes and most preferred less than 5 minutes.

[0012] The dry substance content of the mixture just before spray freeze drying is preferably higher than 5%, more preferably higher than 8% and even more preferred higher than 10%. The content of the active substance in the mixture just before spray freeze drying is preferably higher than 0.5%, more preferably higher than 1.0%, even more preferably higher than 2.0% and most preferred 4.0% or higher.

[0013] For the spray freeze drying process several apparatus can be used, such as the apparatus described in US 5,922,253.

[0014] In the framework of the present invention the expression sugar includes polysugars and the expression sugar alcohols includes poly sugar alcohols. The sugar glass formed preferably has a glass transition temperature of above 50⁰C at normal environmental conditions. Preferred sugars in the present invention are non - reducing sugars. A non -reducing sugar is a sugar, which does not have or can not form reactive aldehyde or ketone groups. Examples of non -reducing sugars are trehalose and fructanes such as inulines.

[0015] Preferred non -reducing sugars to use in the present invention are fructans or mixtures of fructans. A fructan is understood to mean any oligo- or polysaccharide which contains a plurality of anhydrofructan units. The fructans can have a polydisperse chain length distribution, and can have a straight or branched chain. Preferably the fructans contain mainly β-1 ,2 bonds, as in inulin, but they can also contain β-2,6 bonds, as in levan. Suitable fructans can originate directly from a natural source, but may also have undergone modification.

[0016] Examples of modifications are reactions known per se that lead to a lengthening or shortening of the chain length. In addition to naturally occurring polysaccharides, also industrially prepared polysaccharides, such as hydrolysis products which have shortened chains and fractionated products having a modified chain length are suitable in the present invention. A hydrolysis reaction to obtain a fructan having a reduced chain length can be carried out enzymatically (for instance with endoinulase), chemically (for instance with aqueous acid, physically (for instance thermally) or by the use of heterogeneous catalysis (for instance with an acid ion exchanger).

[0017] Fractionation of fructans, such as inulin, can be achi eved inter alia through crystallization at low temperature, separation with column chromatography, membrane filtration and selective precipitation with an alcohol. Other fructans, such as long-chain fructans, can be obtained, for instance through crystalli zation, from fructans from which mono-and disaccharides have been removed. Fructans whose chain length has been enzymatically extended can also serve as fructan in the present invention. Further, reduced fructans can be used, which are fructans whose reducing end groups, normally fructose groups, have been reduced, for instance with sodium borohydride, or hydrogen in the presence of a transition metal catalysts.

[0018] Fructans which have been chemically modified, such as crosslinked fructans and hydroxyalkylated fructans, can also be used. The average chain length in all these fructans is expressed as the number-average degree of polymerization (DP). The abbreviation DP is defined as the average number of sugar units in the oligo - or polymer.

[0019] Even more preferred reducing sugars in the present invention are inulins or mixtures of inulins. Inulins are oligo- and polysaccharides, consisting of β-1 ,2 bound fructose units with an α-D-glucopyranose unit at the reducing end of the molecule and are available with different degrees of polymerization (DP). The preferred inulins are inulins with a DP of greater than 6 or a mixtures of inulins wherein each inulin has a DP of greater than 6. Even more preferred are inulins or mixtures of inulins with a DP of between 10 and 30. Most preferred are inulins or mixtures of inulins with a DP of between 15 and 25. lnulin occurs inter alia in the roots and tubers of plants of the Liliaceae and Compositae families. The most important sources for the production of inulin are the Jerusalem artichoke, the dahlia and the chicory root. Industrial production starts mainly from the chicory root. The main difference between inulins originating from the different natural sources resides in the degree of polymerization (DP), which can vary from about 6 in Jerusalem artichokes to 10-14 in chicory roots and higher than 20 in the dahlia. Inulin is an oligo - or polysaccharide which in amorphous condition has favorable physicochemical properties for the application as auxiliary substance in pharmaceutical formulations.

[0020] These physicochemical properties are: (adjustable) high glass transition temperature, no reducing aldehyde groups and normally a low rate of crystallization. Further inulin is non toxic and inexpensive.

[0021] The weight ratio of lipophilic compound to sugar or sugar alcohol is typically in the range of between 1 : 1 to 1 :200, more preferably in the range of between 1 :10 and 1 :50 and most preferred in the range between 1 :12 and 1 : 25.

[0022] Organic solvents which are suitable to form a mixture that is stable for a sufficient amount of time with the sugar, water and the lipophilic compound are solvents which are mixable with water such as dimethylsulfoxide (DMSO), N₁N - dimethylformamide (DMF), acetonitrile, ethylacetate, 1 ,4-dioxane and lower alcohols. As the solvents have to be removed by spray drying or freeze drying the solvents should preferably also have a reasonable vapor pressure at the drying temperature. Therefore lower 1 ,4 dioxane and alcohols, defined as Ci-C₆ alcohols, wherein the alkyl chain may be branched or unbranched are preferred. The more preferred alcohols are C₂-C₄ alcohols such as ethanol, n -propyl alcohol and f-butyl alcohol. The most preferred solvents are 1 ,4-dioxane and f-butyl alcohol.

[0023] Preferred compounds to be formulated are natural cannabinoid compounds. In the framework of the present invention the expression "natural cannabinoid compound" includes non -natural derivatives of cannabinoids which can be obtained by derivatization of natural cannabinoids and which are unstable like natural cannabinoids. The preferred natural cannabinoid compound is Δ⁹-tetrahydro- cannabinol. [0024] In a further aspect the present invention also relates to a pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugar alcohols, obtained by spray freeze drying, wherein the lipohilic compound is incorporated in the sugar glass , characterized in that said composition consists of spherical particles having a mean geometric particle size of between 6 and 5000 μm, preferably between 6 and 500 μm, even more preferably between 8 and 25 μm and with a span lower than 4. With spherical in the framework of the present invention is meant that the outer perimeter of the particle has no sharp edges and the aspect ratio of the two dimensional projection is over 0.6. (see A.M. Bouwman et al, Powder Technology 2004, 146, 66 - 72 and P. Schneiderhόhn, Eine vergleichende Studie ϋber Methoden zur qua ntita- tiven Bestimmung von Abrundung und Form an Sand kόrnen, Heidelberger Beitrage zur Mineralogie und Petrographie 1954, 4, 82-85.). In the compositions according to the present invention preferably no guest-host complex is formed between the lipophilic compound and said sugar, said sugar alcohol, the mixture of sugars or the mixture of sugar alcohols.

[0025] The particles obtained have a porosity of 70% or higher, preferably 80% or higher, more preferred 85% or higher and most preferred 90% or higher.

[0026] The particles obtained have a specific surface of above 40 m²/g, preferably above 80 m²/g and most preferred above 100 m²/g.

[0027] In an even further aspect the present invention also relates to a pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugar alcohols, obtained by spray freeze drying, wherein the lipohilic compound is incorporated in the sugar glass, characterized in that said composition consists of sp herical particles having a mean aerodynamic particle size of between 1 and 5 μm and a span lower than 5.

[0028] The above particle size is directly obtained by the spray freeze drying process, without any particle size reduction step, such as milling. The product according to the present invention contains an amount of degradation products lower than 10 % and a percentage of phase separation lower than 15 %. In dissolution tests using aqueous dissolution media, that guarantee sink conditions, the material dissolves within 45 minutes. The physical properties of the particles (e.g. aerodynamic particle size distribution, shape and the fragility of the particles) make the production especially suitable for dispersion into an aerosol that can be used for pulmo nary administration. When the particle size is reduced due to breakage during dispersion, this increases the chance for peripheral lung deposition.

[0029] The following examples are only intended to further illustrate the invention, in more detail, and therefore these examples are not deemed to restrict the scope of the invention in any way.

Example 1. Materials and Methods Example 1a. Materials

The following materials were of analytical grade and used as supplied: methanol, ethanol and tertiary butanol (TBA). Inulin, type TEX803!, having a number average degree of polymerization (DP) of 23, was provided by Sensus, Roosendaal, The Netherlands. Δ⁹-tetrahydrocannabinol was a gift of Unimed Pharmaceuticals Inc., Marietta, USA. Demineralized water was used in all cases. Diazepam was obtained from Sigma-Aldrich Chemie GmbH, Steinheim, Germany. Cyclosporin A was obtained from Bufa B.V., Uitgeest, The Netherlands.

Example 1b. Methods

Dissolution experiments

Dissolution experiments were carried out in triplo in a 0.25% SDS (w/v) solution using an USP dissolution apparatus I (basket).

Determination of porosity after spray freeze drying

The porosity (ε) after spray freeze drying was measured according to the following procedure. Inulin was dissolved in water/TBA mixtures of 6/4 v/v. The inulin concentration (c) was varied from 13.3 mg/ml up to 100 mg/ml. These solutions were slowly pumped through a tube to generate equally sized droplets. The volume of the generated droplets (V_dlOp) was determined by counting the number of drops necessary to fill a volume of 5.00 ml. The droplets were frozen by dropping them into a bucket filled with liquid nitrogen. The frozen solution spheres were photographed by a digital camera together with a ruler for calibration. Sigma Scan Pro 5.0 (Jandel Scientific, Erkrath, Germany) was used to determine the cross sectional area of the frozen droplets. Subsequently, the diameter was calculated. The diameter of the spray freeze dried particles (d_p) was determined according to the same procedu re. The porosity was calculated with the following equation:

, _H3 _ ^Vdrop^C r < - Pinulin

For the density of inulin, p_/nu&,, 1.534 g/cm³ was taken (HJ. C. Eriksson et al., Int. J. Pharm. 2002, 249, 59-70).

Scanning electron microscopy (SEM)

Double sided adhesive tape was placed on an aluminium specimen holder upon which a small amount of powder was deposited. The particles were coated with approximately 10-20 nm gold/palladium, using a sputter coater (Balzer AG, type 120B, Balzers, Liechtenstein). Scans were performed using a JEOL scanning electron microscope (JEOL, type JSM-6301 F, Japan) at an acceleration voltage of 1.5 kV. All micrographs were taken at a magnification of 2000.

Laser diffraction

The geometric particle size distribution was measured with a Sympatec HELOS compact KA laser diffraction apparatus (Sympatec GmbH, Clausthal -Zellerfeld, Germany). The powder was dispersed using a RODOS dry powder dispenser at 0.5 bar or using an inhaler adapter (INHALER, Sympatec GmbH, Clausthal -Zellerfeld, Germany) in combination with a test inhaler based on air classifier technology at 60 L/min for 3 seconds (A. H. de Boer et al. Int. J. Pharm. 2002, 249, 233-245; A. H. de Boer et al., Int. J . Pharm . 2003, 260, 187-200). A 100 mm lens was used and calculations were based on the Fraunhofer theory. All data given are the mean of at least four measurements.

Differential Scanning Calorimetry

Thermal behaviour of the spray freeze dried powders was determined by modulated differential scanning calorimetry (MDSC) on a differential scanning calorimeter (DSC2920, TA Instruments, Gent, Belgium). A modulation amplitude of 0.318⁰C, a modulation period of 60 seconds and a heating rate of 2°C/min was used. Calibration was performed with indium. Standard aluminium sample pans were use d. During measurement, the sample cell was purged with nitrogen at a flow rate of 35 mL/min. Before scanning, the sample pan was heated at 2°C/min to 50⁰C to remove all residual moisture. Subsequently, the sample was cooled to -20⁰C and then scanned up to 180⁰C. The glass transition temperature (Tg) was defined as the inflection point of the change in specific heat in the reversing signal.

BET analysis

A 5-point nitrogen adsorption isotherm at 77 K was measured with a Tristar surface analyser Micromeritics Instrument Corporation, Norcross (GA), USA. The BET theory (S. Brunauer et al., J. Am. Chem. Soc. 1938, 60, 309-319) was used to calculate the surface area. Duplicate analyses were performed with all spray freeze dried powders taken from a vacuum desiccator. For every drug load two different batches were analysed.

Stability study To investigate the degradation of pure THC, 20 ml_ glass vials were charged with 70 μl_ of a solution of THC in methanol containing 2.52 mg THC. They were left overnight under a flow of dry nitrogen to allow for methanol evaporation. The resulting thin layers of THC spread over the bottom of the vials (4.5 cm ²). Spray freeze dried and freeze dried material containing THC were weighed in vials. All samples were stored in climate chambers of 20°C/45%RH and 60°C/8%RH. Samples (n=3) were taken at different time intervals and analysed by means of HPLC using the method described by van Drooge et al. (Eur. J. Pharm. Sci. 2004, 21 , 511 -518). Briefly, samples were extracted with methanol. A Waters 717+ autosampler was used to inject 50μl_ of supernatant on a precolumn (HPLC precolum inserts, μBondapak C18 Guardpak) followed by a Chrompack Nucleosil 100 C18 column (4.6x250 mm). Absorbance at 214 nm was measured with a UV detector (Shimadzu SPD-M6A). Chromatograms and peak areas were analysed with an integrator (waters 741 Data Module) and Kromasystem 2000 software. The flow rate of the eluens (methanol/water 92/8 (v/v) plus 5 drops concentrated sulphuric acid per litre eluens) was set at 1.0 mL/min. In a chromatogram of untreated THC, a large peak was observed at a retention time of 7.5 min. In every series of HPLC -runs some calibration samples were included. Cascade lmpactor Analysis

In vitro deposition of the powder formulations was tested wi th a multi-stage liquid impinger (MSLI) of the Astra type (Erweka, Heusenstamm, Germany). A flow rate of 60 L/min was used for 3 seconds according to the procedure described by the European Pharmacopeia 4^th Ed. 2002. A mixture of water and ethanol (90%v/v water) was used as solvent since the use of pure water resulted in inhomogeneous solutions and improper rinsing due to the low aqueous solubility of THC. Each impactor stage was filled with 20 ml_ of solvent. In the final stage a dry glass filter (Gelman Sciences, type A/E, Michigan, USA) was used for the retention of particles that passed the fourth stage. A previously described test inhaler based on air classifier technology (A. H. de Boer et al. Int. J. Pharm. 2003, 260, 187-200) was used under controlled ambient conditions (20°C/50%RH) and in each experiment 10 inhalations were performed. All powders used were pre -equilibrated in a climate chamber at 20⁰C and 45%RH. Two independently produced spray freeze dried batches were analysed. The deposition was defined as the weight fraction powder relative to weight of the powder used in the cascade impactor analysis and was calculated from the drug load and the inulin concentration. The inulin concentration on each of the different stages was analysed using the Anthrone assay (T. A. J. Scott et al. Analytical Chemistry 1953, 25, 1656-1661). Samples of 1.00 mL were mixed with 2.00 mL Anthrone reagent 0.1%w/v in concentrated sulphuric acid. Due to the enthalpy of mixing, the sample was heated to its boiling point. T he boiling mixture was then cooled to room temperature. After 45 minutes the sample was vortexed and 200 μL of sample was analysed in a plate reader (Benchmark Platereader, Bio-Rad, Hercules, USA) at 630 nm. In every assay two 11 point calibration curves o f the appropriate spray freeze dried powder in the appropriate medium was established. In each of the experiments the recovery was above 90%.

Example 2. Preparation of spray freeze dried powder of THC to form an inulin glass. To produce a spray freeze dried powder, an aqueous inulin solution of various concentrations and a 10-mg/mL THC in TBA solution were prepared (Table I). Table I: Composition of the different mixtures used to produce the solid dispersions.

Subsequently these solutions were mixed at a volume ratio water/TBA of 6/4. The solution containing both THC and inulin was sprayed with the 0.5mm nozzle of the Bϋchi 190 mini spray dryer (Bϋ chi, Flawil, Switserland). The liquid feed rate was 10.5 mL/min and the atomising air flow was set at 400 L_n/h. The outlet of the nozzle was positioned about 10 cm above liquid nitrogen. Hot water was pumped through the jacket of the nozzle in order to avoid freezing of the solution inside the nozzle. The resulting suspension (frozen droplets of the solution in liquid nitrogen) was transferred into the freeze dryer (Christ, model Alpha 2-4 lyophilizer, SaIm and Kipp, Breukelen, The Netherlands). Vacuum was applied as soon as all nitrogen was evaporated. During the first 24 hours the pressure was set at 0.220 mbar and the shelf temperature at -35°C (condensor temperature -53°C). During the second 24 hours, the shelf temperature was gradually raised to 20⁰C wh ile the pressure was decreased to 0.05 mbar. After removing the samples from the freeze drier, they were stored over silicagel in a vacuum desiccator at room temperature for at least 1 day. As can be seen in table I, the drug load was varied by spray free ze drying solutions of various inulin concentrations while keeping the THC concentrations constant. When solid dispersions were prepared by freeze drying (for comparison) a previously described freeze drying procedure was followed (D. J. Van Drooge et al., Eur. J. Pharm. Sci. 2004, 21 , 511 -518). This procedure uses the same instrument settings as applied during drying of spray freeze dried material . Example 3. Characteristics of spray freeze dried THC containing powder

The spray freeze dried solid dispersions appeared as a white powder with a low bulk density ranging from about 20 to 85 mg/cm³ and a very high bulk porosity ranging from 94% to 99% depending on the total solid concentration in the solution.

Furthermore, the powder easily swirled up, which is a first indication of its applicability for inhalation.

The SEM pictures of the different powders are shown in Figure 1 (representative

SEM pictures for drug loads of 4,8,12,16,20 and 30 wt-% designated A, B, C, D, E and F, respectively) .

They showed a high porosity and a rough surface in all cases. The surface texture does not change when the drug load increased but somewhat more broken particles were observed at the highest drug load indicating high fragility.

Due to handling problems, the porosity of the spray freeze dried particles could not be measured directly. However, an estimation could be performed with larger spheres. The effect of solute concentration on droplet formation, freezing and particle size after drying was investigated. The results are depicted in Table II. The droplet

Table Il Size (relative to droplet size) and porosity of particles during spray freezing process inulin cone. (mg/mL) 100 50.0 25.0 13.3 droplet size (%) 100 ± 0.2 100 ± 0.1 100 ± 0.5 100 ± 0.7 frozen droplet size (%) 102 ± 5.1 104 ± 2.4 103 ± 2.1 104 ± 2.6 particle size (%) 84.1 ± 2.4 79.7 ± 2.8 78.2 ± 2.7 66.9 ± 3.0 porosity of particle (%) 89.0 ± 0.24 93.6 ± 0.13 96.6 ± 0.09 97.1 ± 0.08 density of particle 169 ± 3.63 99.4 ± 2.06 52.6 ± 1.37 44.6 ± 1.26 (mg/cm³)

sizes were 3.45 mm and independent of inulin concentration. After freezing a small increase in diameter was observed, indicating that a water/TBA solution containing inulin expands slightly upon freezing. The expansion was irrespective of inulin concentration. Furthermore, as can be expected, spray freeze drying of lower concentrated solutions yielded particles of higher porosities. However, after lyophilization of the frozen solution spheres, all particles were significantly smaller. During drying, particle diameters decreased to 84.1% of the droplet size for the most concentrated solution and even more (79.7-66.9%) for particles with lower inulin concentrations. This implies that particles prepared from low concentrated solutions shrink more during drying which is caused by their higher porosity and their consequently lower strength.

The geometric volume median diameter (x₅₀) of all THC containing powders was analysed with laser diffraction using two different dispersion methods. Firstly, the materials were dispersed with a RODOS disperser at a relatively low pressure of 0.5 bar in order to minimize the dispersion forces during the measurement. Secondly, the powders were dispersed by means of the test inhaler at 60 L/min for 3 seconds in order to measure the geometric particle size that actually leaves the inhaler. These test conditions correspond with the conditions during cascade impactor analysis. With RODOS measurements it was found that the geometric volume median diameter of all powders except for the 30wt-% drug load more or less corresponded with estimations from SEM pictures (see Figure 2: Median volume diameters of spray freeze dried powder, determined by laser diffraction with: RODOS dispersion at 0.5 bar (shaded columns) and with test inhaler dispersion at 60 L/min for 3 seconds (open columns) (error bars represent standard deviations, n > 4)).

At a drug load of 30wt-%, the particle size seems smaller. Apparently, due to their higher porosity, the particles are so fragile that the relatively Io w dispersion forces generated with the RODOS are already large enough to break up and de - agglomerate these powders. Much larger dispersion forces that are generated when the powders are dispersed with the test inhaler result in smaller particles (see Fig. 2). In this case, also less porous and less fragile particles (lower drug loads) are broken and de-agglomerated. Apparently, they are fragile enough to allow for disruption by the applied dispersion forces. Disruption may be advantageous to obtain high alveolar deposition during inhalation (see further).

The BET specific surface areas of all powders ranged from about 70 to 110 m²/g. These very high specific surface area's are in accordance with previously reported data on spray freeze dried materials . Finally, the powders were characterized by modulated differential scanning calorimetry (MDSC). In Table III, the glass transition temperatures (Tg's) of THC, amorphous inulin and the different solid dispersions are presented. As reported before, THC remains also above the Tg in the amorphous state since it resists crystallization. A Tg of 9.3°C was observed for the pure THC. The inulin type used in this study has a Tg of 155°C. The results show that incorporation of THC in inulin glasses does not affect the Tg of inulin. Table III. Glass transition temperatures found in solid dispersions with various drug loads. All mixtures were prepared by spray freeze drying.

Only at the highest drug load a Tg of THC could be discerned. This indicates that at this drug load either THC molecules are homogeneously dispersed in the inulin but form a percolating system or that THC is not dispersed homogeneously throughout the inulin carrier anymore. In either case THC molecules are neighbouring resulting in a Tg of pure THC.

Example 4. Stability of THC in the spray freeze dried inuline glass powder as function of drug load.

The spray freeze dried solid dispersions containing THC appearing as a white powder, showed no coloration in due time, which is a first indication of effective stabilization of the labile THC by the inulin glass . The results of a more thorough investigation on the stabilization of THC is shown in Figure 3 (THC content as a function of storage time in spray freeze dried powders with 4 and 8wt-% THC and in pure THC samples. A: storage at 20°C/45%RH; shaded squares: pure THC, open squares: 4wt-%, solid squares 8wt-%. B: storage at 60°C/8%RH; shaded squares: pure THC, open squares: 4wt-%; solid squares: 8wt-%). The THC content in spray freeze dried powders containing 4 and 8wt-% THC initially is plotted as a function of time. It was found that pure THC degrades completely within about 50 days when exposed to air of 20°C/45%RH. (see Fig. 3A) However, when it is incorporated in the glassy inulin matrix, about 80% of the THC could be recovered after 300 days. When the more stressful storage condition of 60°C/8%RH is chosen, pure THC degraded completely within 15 days, (see Fig. 3B) Again the glassy inulin matrix decelerated THC degradation. No differences in degradation rate were observed betwee n the 4 and 8% drug load. Apparently, for both drug loads THC was effectively shielded from its environment by a matrix of inulin and thereby strongly stabilised. To investigate the effect of drug load on THC stabilisation in the solid dispersions in more detail, spray freeze dried powders of a wide range in drug loads were evaluated. To investigate the effect of freezing rate on THC stabilisation, solid dispersions produced by freeze drying instead of spray freeze drying were subjected to a stability study. It appeared that at 20°C/45%RH all spray freeze dried powders effectively stabilised the THC even up to a drug load of 30 wt-% (see Figure 4 : Stability at 20°C/45%RH of THC in solid dispersions as a function of drug load. Given are the recoveries of THC (Black squares: spray freeze dried batches after 3.5 months; white diamonds: freeze dried material after 1.5 months, standard deviations all < 15%). All spray freeze dried powders contained over 85% of the original THC content after storage for 3.5 months. When freeze dried cakes were exposed to same environment, significantly more THC was degraded even though the storage was only 1.5 months. Especially at high drug loads spray freeze drying yields substantially better stabilised material. It can be conclu ded that spray freeze drying is the optimal process for the production of solid dispersions, not only because particles are easily obtained but also because THC is strongly stabilized for all drug loads evaluated.

Example 5. Effect of batch size/freezing rate on stability of freeze dried THC in inulin.

100 mL of solution containing inulin (type TEXI803, degree of polymerization 23) and THC dissolved in a mixture of water and TBA was frozen using liquid nitrogen. It t ook several minutes to completely freeze 100 mL of such a solution. Furthermore, small vials containing 0.4 mL or 2 mL of the same solution were frozen. After lyophilization solid dispersions were obtained with a theoretical drug load of 4%. Both batches were put in a vacuum desiccator for on e day. The stability of THC in the slowly cooled batch was very limited, because as soon as this material was transferred from the freeze dryer to the vacuum desiccator, it turned purple indicating THC degradation. To test the long term stability of THC both batches were exposed to 20⁰C and 45% Relative Humidity (RH). The results are depicted in Figure 5: Effect of batch size on THC stability. The immediate degradation of the slow freezing batch appeared to be large (about 22%). Furthermore, the long term stability was poor compared to material obtained by vial freezing. It can therefore be concluded that slow cooling results in a fraction of THC that is not stabilized at all and another fraction poorly stabilized, whereas vial freezing yields material with improved stability.

Example 6. In vitro deposition behaviour of the spray freeze dried THC containing powders

The geometric particle sizes, reported as the volume median diameter in the shaded bars in figure 2, indicated that the particles produced with spray freeze drying are rather large for an application in pulmonary drug delivery. Generally particles between 1 and 5 μm having a density of approximately 1 mg/cm ³ are considered suitable for inhalation . After dispersion with the inhaler adapter, the geom etric particle size of the powders measured with laser diffraction was about this size. However, the size limits refer to the aerodynamic diameter d_aer_o, which is determined by the geometric diameter d_geo, the density of the particle p_p (estimations are given in table II), and the reference density p_r (the density of water taken as 1 g/cm³) The shape factor χ equals 1 for spherical particles and is larger than 1 for non -spherical particles. The aerodynamic diameter can be calculated according to the followi ng equation.

Since the particles in this study are extremely porous, i.e. p_p is very small, the aerodynamic diameter will be substantially smaller than the geometric diameter.

When the particles are assumed to be spherical, the aerodyn amic diameter will be approximately 40-20% of the geometrical diameter depending on the porosity and density of the particles. Therefore, it was interesting to subject the powders to cascade impactor analysis, because the results are governed by the aerodynamic diameter. Moreover, the outcome of cascade impactor analysis is considered to be predictive regarding the suitability for inhalation in vivo.

The air classifier type inhaler was used in the cascade impactor analysis because laser diffraction analysis showed small particles leaving the inhaler, caused by the strong dispersion forces typical for this type of inhaler (A. H. de Boer et al., Int. J. Pharm. 2003, 260, 187-200).

As can be seen in Figure 6 (Cascade impactor results obtained with spray freeze dried powders having different drug loads, (cross hatched = 4% THC, dotted = 8% THC; black = 12% THC, grey = 16% THC, white = 20% THC) (duplicate of two independently produced batches, error bars indicate highest and lowest value) ), all powders showed high fine particle fractions. The fine particle fraction (FPF), here defined as the sum of the 3^rd, 4^th and the filter stage relative to the total dose, was very high for all powders. This implies that all powders showed excellent inhalation behaviour since the FPF is assumed to represent deep (peripheral) lung deposition during in vivo inhalation. All powders showed similar inhalation behaviour except for the powder with 4% drug load. For unknown reasons, this powder showed a high retention in the inhaler and an FPF of only 35%. However, all other materials showed less inhaler retention and a fine particle fraction of 40 -50%. These results indicate that spray freeze dried powders in combination with an air classifier based inhaler is very promising for pulmonary delivery. Moreover, these in vitro inhalation simulations were performed with unformulated material: only spray freeze dried powder was used without any additional excipients or formulation techniques that could further improve the aerosolisation behaviour.

Example 7. Effect of batch size/freezing rate on mode of incorporation of diazepam.

To investigate the effect of freezing rate on the mode of incorporation, the lipophilic model drug diazepam was incorporated in inulin (type TEXI803) by means of v ial freeze drying (volumes of 2 ml in a vial were frozen) and spray freeze drying. When phase separation occurs during freezing, each phase in the amorphous solid dispersion should exhibit a glass transition temperature (Tg). Differential Scanning Calorimetry (DSC) was used to measure the number of Tg's. In the thermograms obtained from the DSC measurement, solid dispersions with high drug loads (35 wt- %) were compared with a physical mixture of amorphous diazepam and amorphous inulin. The results are depicted in Figure 7 (Thermograms of solid dispersions and physical mixture containing diazepam and inulin).

In the physical mixture (trace 1) two Tg's could be discerned. In the vial freeze dried solid dispersion two Tg's were also discerned, however the Tg of diazepam was less pronounced, which indicates that phase separation is only partial. However, when a solid dispersion was produced by spray freeze drying, only one Tg could be discerned. It can therefore be concluded that phase separation during freezing between inulin and lipophilic drug can be prevented by fast cooling (small amounts of liquid). Example 8. Preparation of spray freeze dried cyclosporine containing powders with different drugloads.

5 batches of different composition were prepared by dissolving Cyclosporin A (CsA) in terf-butanol (TBA) and inulin (DP23) in demineralised water. The concentrations of CsA in TBA and inulin in water were adjusted to achieve a 5%, 10%, 20%, 30% and 50% (w/w CsA/inulin) drugload with a total concentration of 65 mg/ml when the TBA/CsA solution was mixed with the water/inulin solution in a ratio of 40% (v/v) TBA/CsA and 60% (v/v) water/inulin. A batch of pure CsA was also produced in a TBA/water solution, albeit in a lower total concentration of 3 mg/ml. The partial concentration of CsA in TBA used in the pure CsA batch was comparable to the partial concentration of a 5% (w/w) formulation. After mixing the TBA/CsA solution with the water/inulin solution in a 40:60 (v/v) ratio the resulting solution was sprayed over a bowl of liquid nitrogen. The solution was sprayed using a two -fluid nozzle with an orifice of 0.5 mm, a fluid flow rate of 3 ml/min and an atomizing air flow rate of 500 l/h. Upon completion of the spraying procedure the bowl of liquid nitrogen con taining frozen droplets of TBA/water was transferred to a lyophilizer. After most of the liquid nitrogen was evaporated a lyophilizing procedure was started. To sublimate excess solvent the sample was exposed to a shelve temperature -35 ⁰C and a pressure of 0.220 mbar. After 24 hours the temperature and pressure was incrementally increased over a 3 hour period to 20 ⁰C and 0.05 mbar in order to evaporate absorbed solvent in the glassy formulations. Subsequently all produced formulations were stored in a vacuum exsiccator Dissolution experiments were carried out as described in Example 1b. All formulations were weighed in the basket to an amount of 30 mg CsA, ie. in all dissolution experiments 30 mg CsA was dissolved. The results of the dissolution experiments are depicted in Figure 8. For the dissolution of pure CsA 30 mg was weighed and 70 mg pure inulin (also spray freeze dried) was added. This sample is indicated in Figure 8 as "physical mixture (Ph. mix)". From the results it can be concluded that incorporation of CsA into inulin (DP23) increases the rate of dissolution up to a drugload of 50% (w/w) compared to not -incorporated CsA (CsA Ph. Mix). Although the 5 and 10% (w/w) formulations dissolve faster than the 20, 30 and 50% the effect is clearly visible even at the latter drugloads.

Claims

1. A method for the preparation of a pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugars alcohols, wherein the lipophilic compound is incorporated in the sugar glass, characterized in that a. said lipophilic compound is dissolved in an organic solvent that is miscible with water and said sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols is dissolved in water; b. the dissolved lipophilic compound and the dissolved sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols are mixed in such a way that a homogeneous mixture is obtained; c. said mixture is spray freeze dried.

2. A method according to claim 1 , wherei n said step b and c are performed in a continuous or semi-continuous way.

3. A method according to claims 1 -2, characterized in that said sugar or mixture of sugars is a non -reducing sugar or a mixture of non -reducing sugars.

4. A method according to claims 1 -3, characterized in that said sugar or mixture of sugars is a fructane or a mixture of fructanes .

5. A method according to claim 4, characterized in that said fructane or mixture of fructanes is inulin or a mixture of inulins, preferably inulin with a DP of greater than 6 or a mixtures of inulins wherein each inulin has a DP of greater than 6 .

6. A method according to claims 1 -5, characterized in that said organic solvent is 1 ,4-dioxane or a Ci-C₆ alcohol, preferably a C₂-C₄ alcohol.

7. A method according to claims 1 -6, characterized in that said lipophilic compound is a natural cannabinoid compound .

8. A method according to claim 7, characterized in that said natural cannabinoid compound is Δ⁹-tetrahydrocannabinol.

9. A method according to claims 1 -6, characterized in that said lipophilic compound is diazepam.

10. A method according to claims 1 -6, characterized in that said lipophilic compound is cyclosporin A.

11. A pharmaceutical composition obtainable by the method according to claims 1 - 10.

12. A pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugar alcohols, obtained by spray freeze drying, wherein the lipohilic compound is incorporated in the sugar glass, characterized in that said composition consists of spherical particles having a mean geometric particle size of between 6 and 5000 μm.

13. A pharmaceutical composition comprising a lipophilic compound and a glass of a sugar, a sugar alcohol, a mixture of sugars or a mixture of sugar alco hols, obtained by spray freeze drying, wherein the lipohilic compound is incorporated in the sugar glass, characterized in that said composition consists of spherical particles having a mean aerodynamic particle size of between 1 and 5 μm.

14. A pharmaceutical composition according to claim 12-13, characterized in that the porosity of said composition is 80% or higher.

15. A pharmaceutical composition according to claims 12-14, characterized in that said lipophilic compound does not form of a guest -host complex with said sugar, sugar alcohol, mixture of sugars or mixture of sugar alcohols .

16. A pharmaceutical composition according to claim 1 2-15, characterized in that said sugar or mixture of sugars is a non -reducing sugar or a mixture of non- reducing sugars.

17. A pharmaceutical composition according to claims 1 2-16, characterized in that said sugar glass has a glass transition temperature of above 50⁰C at normal environmental conditions.

18. A pharmaceutical composition according to claims 1 2-17, characterized in that said sugar or mixture of sugars is a fructane or a mixture of fructanes .

19. A pharmaceutical composition according to claim 18, characterized in that said fructane or mixture of fructanes is inulin or a mixture of inulins, preferably inulin with a DP of greater than 6 or a mixtures of inulins wherein each inulin has a DP of greater than 6.

20. A pharmaceutical composition according to claim 19, characterized in that said inulin or each inulin in the mixture has a DP of between 10 and 30, preferably between 15 and 25.

21. A pharmaceutical composition according to claims 12-20, characterized in that said lipophilic compound is natural cannabinoid compound.

22. A pharmaceutical composition according to claim 21 , characterized in that said natural cannabinoid compound is Δ⁹-tetrahydrocannabinol.

23. A pharmaceutical composition according to claims 12-20, characterized in that said lipophilic compound is diazepam.

24. A pharmaceutical composition according to claims 12-20, characterized in that said lipophilic compound is cyclosporin A.