Process for improving crystallinity

Abstract

This invention provides a process for increasing the crystallinity of at least one solid material which is less than 100% crystalline, comprising contacting said solid material with solvent in which the solid material is insoluble or poorly soluble (a non-solvent); and applying ultrasound to the solid material when in contact with said non-solvent.

Claims

1. A process for increasing the crystallinity of a crystallizable solid, comprising contacting said crystallizable solid with a non-solvent; and applying ultrasound to the crystallizable solid when in contact with said non-solvent; wherein the crystallizable solid is selected from the group consisting of an active pharmaceutical ingredient, an active agrochemical ingredient, a pharmaceutical excipient, an agrochemical excipient and appropriate mixtures of two or more thereof; and wherein the process further comprises: (i) forming a solution of the crystallizable solid in a solvent; (ii) subjecting the solution, after step (i), to a process selected from the group consisting of rapid precipitation, freeze drying, lyophilisation, rapid expansion of supercritical solutions, spray drying or mixtures thereof, wherein the said dissolved crystallizable solid is converted into a substantially dry solid material; (iii) isolating, after step (ii), the substantially dry crystallizable solid from the liquid and/or gaseous components of the process of step (ii); (iv) treating, after step (iii), said substantially dry crystallizable solid from step (iii) with a non-solvent therefor; (v) applying ultrasound, after step (iv), to the crystallizable solid from step (iv) when it is in contact with said non-solvent; and (vi) optionally, after step (v), separating and/or drying the resultant crystallizable solid from step (v).

2. A process according to claim 1, wherein the crystallizable solid is a particulate solid material having a mass median aerodynamic diameter of up to about 10 m.

3. A process according to claim 1, wherein prior to the application of the above process, the crystallizable solid is less than 50% crystalline.

4. A process according to claim 1, wherein the crystallizable solid is a pharmaceutically active ingredient selected from the group consisting of anti-allergics, bronchodilators, anti-inflammatory steroids and mixtures thereof.

5. A process according to claim 1, wherein the crystallizable solid is obtained from spray-drying.

6. A process according to claim 1, wherein the crystallizable solid produced by the process is at least 90% crystalline.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The process of the invention may be carried out using conventional equipment as shown in the accompanying figures in which:

(2) FIG. 1 shows conventionally spray dried budesonide (with ultrasound treatment).

(3) FIG. 2 shows spray dried budesonide with ultrasound treatment according to the present invention.

(4) FIG. 3 shows a DSC of conventionally spray dried budesonide.

(5) FIG. 4 shows a DSC of spray dried budesonide with dry powder ultrasound treatment of the present invention.

(6) FIG. 5 shows a diagrammatic representation of a conventional spray drying equipment whereby the dry solid collection chamber is replaced by an ultrasonic cell having an ultrasonic probe inserted into the cell.

(7) FIG. 6 shows a bonded transducer apparatus of a similar configuration to that of FIG. 5.

(8) FIG. 7 shows a multiple transducer apparatus of a similar configuration to that of FIGS. 5 and 6. The multiple transducers in this case are circumferentially mounted around a cylindrical duct.

(9) FIG. 8 shows a multiple transducer apparatus similar to that in FIG. 7 where the multiple transducer apparatus is configured in a recirculation loop.

(10) FIG. 9 shows a sample of fluticasone propionate prepared by aerosolisation (according to the invention).

(11) FIG. 10 shows a sample of fluticasone propionate prepared by aerosolisation (according to the invention).

(12) FIG. 11 shows a sample of fluticasone propionate prepared by dispersion and precipitation with ultrasound.

(13) FIG. 12 shows a sample of micronised/milled fluticasone propionate.

(14) FIG. 13 shows aerosol efficiency of various samples.

(15) FIG. 14 shows bar chart representation of Fine Particle Fraction.

(16) FIG. 15 shows aerosol efficiency compared with GSK's Flixotide.

(17) FIG. 16 Particles of Fenoterol hydrobromide post ultrasonic treatment.

(18) FIG. 17 Micronised Fenoterol hydrobromide.

(19) FIG. 18 Particle size distribution data GR005/180/A4 and GR005/179/C are particles prepared by this invention.

(20) FIG. 19 Comparative Fine Particle Fraction (FPF) data using inhalation device.

(21) FIG. 20 Comparative Fine Particle Fraction (FPF) data using inhalation proprietary test rig.

(22) FIG. 21 Surface energy measurements with IGC at finite and infinite dilution.

(23) FIG. 22 AFM Topology profiles for budesonide micronized and particles of the present invention.

(24) FIG. 23 shows aerosol efficiency of various samples of FP following 1 and 3 months storage.

(25) FIG. 24 shows a bar chart comparing the aerosol efficiency of various samples of FP following 1 and 3 months storage.

(26) FIG. 25 shows a particle fraction distribution graph for various stages of a Next Generation Impactor.

(27) FIG. 26 shows the fine particle fraction of various samples of fluticasone propionate following 1 month storage.

(28) FIG. 27 shows a bar chart representation of fine particle fraction.

(29) FIG. 28 shows fluticasone propionate particles (90% by weight) made by the current invention in combination with salmeterol xinafoate particles (10% by weight).

(30) FIG. 29 shows the cohesive-adhesive balance of fluticasone propionate with lactose.

(31) FIG. 30 shows the excellent homogeneity of FP particles of the present invention when blended with micronized SX.

(32) FIG. 31 shows the FPFLD for FP engineered according to the current invention when blended with micronized SX compared with Advair.

(33) FIG. 32 shows AFM contour plots for surface roughness of sample 3 FP.

(34) FIG. 33 shows AFM contour plots for surface roughness of sample 4 FP.

DETAILED DESCRIPTION OF THE INVENTION

(35) Referring to FIGS. 3 and 4, comparing the DSC traces for the two batches, there is clear indication that application of ultrasound to spray dried particles modifies the physical characteristics of particles. The exotherm (positive peak) at 120 C. is indicative of amorphous to crystalline transformation in the DSC apparatus. In general there is definite improvement in crystalline characteristic of processed material.

(36) Turning to FIG. 5, spray drying with ultrasound apparatus comprises a liquid feed chamber 1, spray drying atomiser and heated gas inlet 2, evaporation chamber 3, cyclonic separator 4, continuous ultrasonic treatment chamber 5, (surrounded by a thermal jacket 6). The conventionally treated spray dried powder is deposited directly into an ultrasonic flow cell chamber 5. Concurrently, a continuous feed of non-solvent 7, is pumped via a pump 8, at a suitable flow rate balanced by the rate of flow of particle slurry 9, to subsequent processing by filtration or drying. Ultrasonic probe 10, irradiates the mixture with ultrasonic energy and the mixture flows through an outlet 11. The solvent vapour, ultrafine particles and gases 12, are expelled via filter 13. The ultrasonic radiation is continued as long as necessary until the desired particle size and crystallinity is achieved. Naturally the feed stream to the spray dryer is balanced with the rate at which particle slurry is removed. The flow rates are controlled such that the residence time in the ultrasonic flow cell chamber 5, is for example, 10 s to 1 hr. Localised cavitation occurring on a microscopic scale promotes changes in fluid temperature and pressure that induces the aforementioned solid state effects. By adjusting the power of the ultrasound, and the residence time in chamber 5, the particle size and morphology can be controlled. The ultrasound has the additional benefit that any crystal deposits within the chamber 5, tend to be removed from the surfaces.

(37) Referring to FIG. 6, spray drying with ultrasound apparatus is of a similar configuration to that of FIG. 5 except that chamber 21 has single bonded ultrasonic transducer 22 located on the external surface of it. The transducer 22 insonates the entire volume of the chamber 21 with sufficient intensity to cause dispersion, deagglomeration and amorphous to crystalline or metastable to stable-crystalline conversion, and by adjusting the power of the ultrasound, and the residence time in the chamber 21, the particle size and morphology can therefore be controlled. The ultrasound has the additional benefit that any crystal deposits within the chamber 21 tend to be removed from the surfaces.

(38) Referring to FIG. 7, spray drying with ultrasound apparatus is of a similar configuration to that of FIGS. 5 and 6 except that chamber 31 has a wrap-around ultrasonic transducers 32 located on the external surface of it. The wrap-around transducers 32 insonates the entire volume of the chamber 31 with sufficient intensity to cause dispersion, deagglomeration and amorphous to crystalline or metastable to stable-crystalline conversion, and by adjusting the power of the ultrasound, and the residence time in the chamber 31, the particle size and morphology can therefore be controlled. The ultrasound has the additional benefit that any crystal deposits within the chamber 31 tend to be removed from the surfaces.

(39) Referring to FIG. 8, this shows a spray drying apparatus with ultrasound apparatus of a similar configuration to that of FIG. 7 except that chamber 31 is attached to a primary particle collection vessel 41 fitted with thermoregulation jacket 43 and optional stirrer impellor 44, via pump 42, thus creating a continuous closed loop processing system. The ultrasound is applied with sufficient intensity to cause dispersion, deagglomeration and amorphous to crystalline or metastable to stable-crystalline conversion, and by adjusting the power of the ultrasound, and the residence time in the recirculation processing loop 31, 41, 42, the particle size and morphology can therefore be controlled.

(40) The skilled addressee will appreciate that the thermal jacket is designed to help maintain the temperature of the non-solvent at a desired temperature, depending on design.

(41) The term comprising means including as well as consisting e.g. a composition comprising X may consist exclusively of X or may include something additional e.g. X+Y.

(42) Unless defined otherwise, the word substantially does not exclude completely e.g. a composition which is substantially free from Y may be completely free from Y. Where necessary, the word substantially may be omitted from the definition of the invention.

(43) Optional or optionally means that the subsequently described event of circumstances may or may not occur, and that the description includes instances where said event or circumstance occurs and instances in which it does not.

EXAMPLES

Example 1

(44) Budesonide (5 g) was dissolved in 100 mL of dichloromethane. The samples of budesonide powder collected in the ultrasonic chamber were produced using a Bchi-290 laboratory-scale spray dryer (insulins), Bchi, Switzerland). The solution was atomized using nitrogen at 7 bar flowing at approximately at 10 Lpm (Liter/minute). The aspirator was set at 100% and flow rate of solution was set to 10 Lpm. The gas temperature was set to 120 C. Budesonide particles were collected the ultrasonic chamber connected to the end of high performance cyclone separator. In order to apply ultrasound to the spray dried particles, the collection ultrasonic chamber was filled with heptane thermoregulated at 25 C. and was fitted with an ultrasonic probe resonating at 20 kHz. Ultrasound at 20 W power was applied between 30 minutes and 1 hour. The resulting particle slurry was spray dried and particles characterized by optical microscopy and DSC (Differential Scanning calorimetry). The size of the particles were typically in the range of 1-7 m.

(45) The D(10), D(50), D(90) for two representative samples were 1.21, 3.03, 4.63 m and 1.05, 2.99, 3.76 m respectively as determined by Sympatec HELOS laser diffraction.

(46) Differential Scanning Calorimetry

(47) DSC experiments were performed with a DSC Q2000 V24.2 build 107 (TA Instruments, UK). Approximately 3 mg of material was weighed into the sample pan of the DSC and subjected to heating ramp of 100 C./min add heated to 275 C. The DSC measurement was carried out using the following steps. Run 9 (spray dried material not treated with ultrasound according to the present invention) Instrument DSC Q2000 V24.2 Build 107 Module DSC Standard Cell RC Sample px02-262-spray dried Size 2.140 mg Method Fast Heating expt 100 C.-min Weighed sample is heated at rate of 100 C./Min to 275 C. Run 10 (material treated with ultrasound according to the process of the present invention) Instrument DSC Q2000 V24.2 Build 107 Module DSC Standard Cell RC Sample px02-262-post ultrasound Size 3.590 mg Method Fast Heating expt 100 C.-Min Weighed sample is heated at rate of 100 C./Min to 275 C.

Example 2

(48) Example 2 shows the advantages of the particles produced according to the present invention.

(49) The aerosolisation efficiency of three batches of engineered fluticasone propionate (FP) produced by various processing was assessed in binary dry powder inhaler (DPI) formulations. Batches studied include:

(50) Sample 2 Prepared by aerosolistion method exemplied in this invention as shown in SEM FIG. 9. Fluticasone propionate (4 g) was dissolved in 100 mL of acetone. The samples of Fluticasone propionate powder collected in the ultrasonic chamber were produced using a Bchi-290 laboratory-scale spray dryer (Bchi, Switzerland). The solution was atomized using nitrogen at 7 bar flowing at approximately at 10 Lpm (Liter/minute). The aspirator was set at 100% and flow rate of solution was set to 10 Lpm. The gas temperature was set to 120 C. Fluticasone propionate particles were collected the ultrasonic chamber connected to the end of high performance cyclone separator. In order to apply ultrasound to the spray dried particles, the collection ultrasonic chamber was filled with heptane thermoregulated at 25 C. and was fitted with multiple bonded transducers (akin to FIG. 7) resonating at 20 kHz. Ultrasound at 20 W power was applied between 30 minutes and 1 hour. The resulting particle slurry was spray dried and particles characterized by optical microscopy and DSC (Differential Scanning calorimetry). The size of the particles were typically in the range of 1-6 m. The D(10), D(50), D(90) were 1.35, 3.25, 5.63 m as determined by Sympatec HELOS laser diffraction.

(51) Sample 3 Prepared by aerosolistion method exemplied in this invention as shown in SEM FIG. 10. Sample 3 was prepared by the same method as sample 2, except that 3 g of FP was used in sample 3. The D(10), D(50), D(90) were 0.99, 2.55, 4.97 m as determined by Sympatec HELOS laser diffraction.

(52) Sample 4 Prepared by alternative precipitation approach as shown in SEM FIG. 11. Sample 4 was prepared as described in WO 2008/114052 A1. This method does not use initial solution atomization. Instead this prior art involves the dispersive antisolvent crystallization brought about by adding a solution of fluticasone propionate in acetone to heptane antisolvent in the presence of an ultrasonic field. This leads to particles significantly smoother than particles formed using the method of the current invention. The D(10), D(50), D(90) were 1.14, 2.67, 5.11 m as determined by Sympatec HELOS laser diffraction.

(53) The samples were compared to an additional binary DPI formulation containing micronized FP and formulations extracted from a Flixotide Discus inhaler.

(54) The aerosolization efficiency of samples 2, 3 and 4 of engineered FP were evaluated using binary formulations containing 0.4% w/w FP.

(55) Each binary formulation contained 0.016 g FP and 3.984 g lactose (ML001, DMV-Fonterra, Vehgel, Netherlands) and was prepared by geometric mixing. Following this, the blend was subsequently prepared using a Turbula T2F (Willy A Bachofen A G, Basel, Switzerland) at 46 rpm for 45 minutes.

(56) Following content uniformity testing, 12.51 mg of each blend was loaded into size 3 hydroxypropylmethyl cellulose capsules (HPMC, Shionogi Qualicaps S A, Basingstoke, UK). The capsules were stored at 44% RH for 24 h prior to in vitro performance testing.

(57) Testing was performed using a Next Generation Impactor (NGI) with pre-separator, which was connected to a vacuum pump (GE Motors). Prior to testing, the pre-separator was filled with 15 ml of mobile phase and the cups of the NGI cups were coated with 1% v/v silicone oil in hexane to eliminate particle bounce.

(58) For each experiment, four individual capsules of the same formulation were discharged into the NGI at 60 Lpm for 4 s via a Rotahaler (GSK, Ware, UK) DPI device. Additionally, blisters from a Flixotide Diskus (GSK, Ware, UK) were emptied and loaded into size 3 HPMS capsules and discharged into a NGI at 60 Lpm1 for 4 s via a Rotahaler. Following aerosolization, the NGI apparatus was dismantled and the inhaler, capsules and each part of the NGI was washed down into known volumes of HPLC mobile phase.

(59) The mass of drug deposited on each part of the NGI was determined by HPLC. This protocol was repeated three times for each blend, following which, the mass median aerodynamic diameter (MMAD), geometric standard deviation (GSD), fine particle dose (FPD) and fine particle fraction of the emitted dose (FPFED) were determined. The FPD represented the mass of drug that was collected on stages 3-8 of the NGI.

(60) The aerosolization efficiency is shown in FIG. 13. The percentage fine particle fraction is shown in FIG. 14.

(61) The aerosolisation efficiency as determined by percentage fine particle fraction (% FPF) of Samples 2 and 3 was significantly greater than that of micronized FP. This data is shown in FIG. 15 and clearly displays aerosol efficiency compared with the formation from a Flixotide. The increase in the performance on inclusion of these materials was dramatic for the inhaler used in this study. Sample 4 had a significantly lower % FPF than micronized FP, which was related to the surface morphology of the particles. These data suggested little difference between the aerosolisation efficiency of samples 2 and 3.

(62) Comparative in vitro data shows that the performance of particle produced according to the present invention are overwhelmingly and surprisingly superior to performance of both conventionally milled particles and specially prepared (ultrasonically precipitated) particles that have vastly different surface, and geometric properties. In all cases the particles have the same particle size range. In a specific example for Fluticasone propionate, the FPF for optimal particles was 54% greater than micronized/milled and over 200% greater than precipitated material. The significant differences and improved performance of these optimal particles can therefore be attributed not to size, although this is an important design criterion, but a range of other properties that describe these vastly improved particles.

(63) The SEM images shown in FIGS. 9 and 10 (according to the invention) clearly show clear contrast in terms of shape and surface roughness compared to precipitated and milled material shown in FIGS. 11 and 12 respectively (not according to the invention). The roughness and rounded/spheroid 3D shape has a profound impact on their performance.

(64) All formulations were stored for one and three months at 25 C./75% RH. The FPF of the formulation containing micronized fluticasone propionate decreased by almost 50% after storage for one month and almost 70% after storage for three months compared to the FPF at t=0. However, the FPF of formulations containing particles prepared by the current invention were not as affected following storage in stress conditions. The FPF of sample 3 decreased 6% after storage for one month and decreased 3% after storage for three months compared to the FPF at t=0. The FPF of sample 2 decreased 7% after one months storage and 28% after three months storage compared to the FPF at t=0. There was a small yet statistically significant decrease, 29%, in the fine particle fraction of the formulation containing sample 4 which was prepared by alternative precipitation approach after one month with a decrease of 24% after 3 months storage compared to the FPF at t=0. This may be related to the planar morphology of the particles, and therefore, the particles are likely to develop greater adhesion to the lactose as result of capillary forces. The data are shown in FIGS. 23 and 24. It is clear that the FPF of samples 2 and 3 maintained a higher FPF than the micronized sample or sample 4 when stored for one or three months in the conditions described. The FPF of samples 2 and 3 decreased by a smaller percentage than the micronized sample when stored for one or three months in the conditions described. This shows that samples 2 and 3 were more stable than the micronized sample after storage for one or three months. Sample 4 has a much lower FPF than samples 2 and 3 after storage for one or three months.

Example 3

(65) Example 3 shows the advantages of the particles produced according to the present invention for formulation in a pMDI. FIG. 25 is a particle fraction distribution graph for various stages on the NGI and clearly shows the superior performance in terms of FPF for sample 1 and sample 5. Both samples were produced in a similar manner to sample 2 of Example 2. Sample 1 was prepared from atomization of a 3% solution of fluticasone propionate in 20% dichloromethane in methanol (3 g in 100 ml) and Sample 5 was prepared from atomization of 3% solution in acetone (3 g in 100 ml). As shown in FIG. 25, the FPF on stage 5 (cut-off 1.36 m) of the NGI for both samples showed over 100% increase compared with micronized material, alongside similar FPF for stage 4 (cut-off 2.30 m). The FPF values were 11.80% and 13.80% on stage 5 for samples 5 and 1 respectively compared with 6.20% for micronized material. It is clear therefore, that samples 1 and 5 have a similar FPF to the micronized product at stage 4, and a greater FPF than the micronized product at stage 5.

Example 4

(66) The aerosolisation efficiency of samples 2, 3 and 4 of fluticasone propionate (FP) produced as described in Example 2 was assessed in combination dry powder inhaler (DPI) formulations containing micronized salmeterol xinafoate (SX) using a Rotahaler unit dose DPI device (GSK, Ware, UK) and Cyclohaler unit dose DPI device (TEVA Pharmaceuticals, Netherlands).

(67) The aerosolization efficiency was evaluated in combination DPI formulations also containing micronized SX. Each combination formulation contained 0.16000 g FP (sample 2, 3 or 4), 0.02320 g SX and 3.8168 g lactose (ML001, DMV-Fonterra, Vehgel, Netherlands) and was prepared by geometric mixing. Following this, each blend was subsequently prepared using a Turbula T2F (Willy A Bachofen A G, Basel, Switzerland) at 46 rpm for 45 minutes. The blend strength of each combination formulation equated to 500 g FP and 50 g Salmeterol base. This matches the dose strength of Advair 500/50. A sample of Advair 500/50 was used as the micronized example. The aerolisation efficiency of four different formulations containing FP samples 2, 3, 4 or micronized FP was measured.

(68) As shown in FIG. 26, at T=0, the FPF of samples 2, 3 and 4 using a Cyclohaler unit dose DPI device are greater than the micronized sample. When the Rolahaler unit dose DPI device is used, the FPF of samples 2 and 3 is greater than sample 4 and the micronized sample. This shows that the fluticasone propionate particles of the present invention, samples 2 and 3, overall have a higher FPF in the two devices. While sample 4 performs well in the Cyclohaler unit, it has a poor performance in the Rolahaler unit. Samples 2 and 3 have a high FPF in both units.

(69) The four samples were stored for 1 month at 25 C./75% RH. The % FPF of micronized FP and the samples 2, 3 and 4 was not significantly affected under stressed storage conditions. These data show that particles prepared by the current invention should afford stability to DPI formulations.

(70) Following content uniformity testing, 12.50.5 mg of each blend was loaded into size 3 hydroxypropylmethyl cellulose capsules (HPMC, Shionogi Qualicaps S A, Basingstoke, UK). The capsules were stored at 44% RH for 24 h prior to in vitro performance testing.

(71) Testing was performed using a Next Generation Impactor (NGI) with pre-separator, which was connected to a vacuum pump (GE Motors). Prior to testing, the pre-separator was filled with 15 ml of mobile phase and the cups of the NGI cups were coated with 1% v/v silicone oil in hexane to eliminate particle bounce.

(72) For each experiment, two individual capsules of the same formulation were discharged into the NGI at 60 Lpm for 4 s via a Rotahaler (GSK, Ware, UK) and 90 Lpm for 2.8 s via a Cyclohaler (TEVA Pharmaceuticals, Netherlands) to ensure both devices were operated such that 4 kPa pressure drop was generated.

(73) Additionally, blisters from a commercially available Advair 500/50 Diskus (GSK, USA) were emptied and 12.5 mg of formulation was transferred into size 3 HPMC capsules and discharged into a NGI at 60 Lpm for 4 s via a Rotahaler and at 90 Lpm for 2.8 s via a Cyclohaler.

(74) Following aerosolization, the NGI apparatus was dismantled and the inhaler, capsules and each part of the NGI was washed down into known volumes of HPLC mobile phase. The mass of drug deposited on each part of the NGI was determined by HPLC. This protocol was repeated three times for each blend, following which, the fine particle dose (FPD) and fine particle fraction of the loaded dose (FPFLD) were determined. The FPD represented the mass of drug that was collected on stages 3-8 of the NGI.

(75) FIG. 26 shows the FPF specifically for FP for an Advair 500/50 equivalent blend of FP, prepared according to this example, and mechanically micronized SX particles. FIG. 27 graphically shows the FPF performance of particles whereby the FP was prepared by this invention. Samples 2 and 3 (cyclo_2 and cyclo_3, and rota_2 and rota_3) clearly show superior performance in terms of FP, and thus clearly indicates that the cohesion between FP and FP particles and adhesion of lactose to FP can be controlled. Above all the micronized components used in this study are from commercially available devices whereby the micronized material has undergone several weeks if not months of conditioning. Conversely particles made by the current invention are highly stable even when freshly prepared, and to reiterate, this again states that particles prepared by the current invention should afford stability to DPI formulations.

(76) FIG. 28 shows particles of FP sample 3 made by the current invention. They were then blended into a combination consisting of salmeterol xinafoate (10% by weight) and fluticasone propionate (90% by weight), wherein the salmeterol xinafoate particles were micronized and the fluticasone propionate particles are prepared according to the present invention.

Example 5

(77) The cohesive-adhesive balance (CAB) of micronized Fluticasone Propionate (FP) and batches of FP engineered by the method of the current invention (samples 2 and 3) and by a different method exemplified in WO 2008/114052 A1 (sample 4), as described in Example 1, were determined with respect to crystalline substrates of FP and lactose monohydrate. The CAB force balance of the different batches with respect to the crystalline substrates was determined as follows:

(78) Probe Preparation:

(79) Particles (n=3) of all batches of FP were attached onto standard V-shaped tipless cantilevers (DNP-020, DI, CA, USA) using an epoxy resin glue (Araldite, Cambridge, UK).

(80) Production of Smooth Lactose & Drug Crystals:

(81) Smooth crystals of lactose were produced on cooling of a heated saturated droplet sandwiched between cover slips. Smooth drug crystals were produced using sitting-drop anti-solvent crystallisation, in which the active was dissolved in acetone and the anti-solvent employed was water.

(82) AFM Force Measurements:

(83) Individual force curves (n=1024) were conducted over a 10 m10 m area at a scan rate of 4 Hz and a compressive load of 40 nN. Environmental conditions were maintained at a constant temperature of 20 C. (1.5 C.) and relative humidity 453%.

(84) The CAB analysis of micronized FP suggested that for equivalent contact geometry, the adhesive FP-Lactose interactions of micronized FP are 1.36 times greater than the cohesive FP-FP interactions.

(85) The CAB analysis of sample 2 FP suggested that for equivalent contact geometry, the adhesive FP-Lactose interactions of sample FP are 1.17 times greater than the cohesive FP-FP interactions.

(86) The CAB analysis of sample 3 FP suggested that for equivalent contact geometry, the adhesive FP-Lactose interactions of sample 3 FP are 1.16 times greater than the cohesive FP-FP interactions.

(87) The CAB analysis of sample 4 FP suggested that for equivalent contact geometry, the adhesive FP-Lactose interactions of sample 4 FP are almost equal to the cohesive FP-FP interactions.

(88) As shown in FIG. 29, the micronized FP was significantly more adhesive to lactose than samples 2, 3 and 4 of FP prepared. Sample 4 was the least adhesive to lactose and SX and therefore, suggests that the surface energy of this material is significantly different from the other batches. However, the adhesion values ranged between 400-800 nN, which reflects the greater contact radius of particles of this material when contacting the different substrates. The greater contact radius of this material will result in limited aerosolisation upon formulation of sample 4 in dry powder inhaler (DPI) formulations.

(89) In contrast, the adhesion values relating to FPlactose interaction of sample 3 ranged from 35-89 nN and the adhesion values of sample 2 ranged from 128-169 nN, which reflects the smaller contact radius of particles of the present invention when contacting the different substrates. The small contact radius of this material will result in greater FP aerosolisation upon formulation of samples 2 and 3 in carrier-based DPI formulations compared to micronized FP, with adhesion values ranging from 169-249 nM.

(90) CAB analysis confirmed that sample 3 may have smaller contact radii than the other materials, whereas sample 4 may have greater contact radii, which is related to the surface geometry of these particles.

(91) These data demonstrate that the particle engineering strategy as exemplified by this invention are able to afford control on both surface energy and particle contact geometry, both of which are critical quality attributes of drug particles in DPI formulations.

(92) FIG. 29 shows the FP-Lactose interactions with respect to different contact geometry of particles taking note of the relatively high forces of both adhesion and cohesion for sample 4.

Example 6

(93) The aerosolization efficiency of sample 3 of fluticasone propionate (FP) produced as described in Example 2 was assessed in combination dry powder inhaler (DPI) formulations containing micronized salmeterol xinafoate (SX) using a Rotahaler unit dose DPI device (GSK, Ware, UK) and Cyclohaler unit dose DPI device (TEVA Pharmaceuticals, Netherlands).

(94) The aerosolization efficiency was evaluated in combination DPI formulations also containing micronized SX. Each combination formulation contained 0.16000 g FP, 0.01160 g SX and 3.8284 g lactose (ML001, DMV-Fonterra, Vehgel, Netherlands) and was prepared by geometric mixing. Following this, the blend was subsequently prepared using a Turbula T2F (Willy A Bachofen A G, Basel, Switzerland) at 46 rpm for 45 minutes. The blend strength of the combination formulation equated to 500 g FP and 25 g Salmeterol base, therefore this was a 500/25 formulation.

(95) Assessment of the content uniformity of the formulation containing micronized FP and micronized SX with a 500/25 formulation suggested poor homogeneity and therefore, this formulation was not characterised by in vitro impaction studies. In contrast, the formulation containing FP sample 3 and micronized SX exhibited very good homogeneity as shown in FIG. 20. This shows the % Relative Standard Deviation for Sample 3 for FP was 3.43 and for SX was 4.55. This compared to the micronized FP-SX formulation, where the % Relative Standard Deviation of FP was 8.76 and of SX was 15.95.

(96) Following content uniformity testing, 12.50.5 g of the blend containing the FP sample 3 blend or Advair was loaded into size 3 hydroxypropylmethyl cellulose capsules (HPMC, Shionogi Qualicaps S A, Basingstoke, UK). The capsules were stored at 44% RH for 24 h prior to in vitro performance testing.

(97) Testing was performed using a Next Generation Impactor (NGI) with pre-separator, which was connected to a vacuum pump (GE Motors). Prior to testing, the pre-separator was filled with 15 ml of mobile phase and the cups of the NGI cups were coated with 1% v/v silicone oil in hexane to eliminate particle bounce.

(98) For each experiment, two individual capsules of the same formulation were discharged into the NGI at 90 Lpm for 2.8 s via a Cyclohaler (TEVA Pharmaceuticals, Netherlands) to ensure both devices were operated such that 4 kPa pressure drop was generated.

(99) Following aerosolization, the NGI apparatus was dismantled and the inhaler, capsules and each part of the NGI was washed down into known volumes of HPLC mobile phase. The mass of drug deposited on each part of the NGI was determined by HPLC. This protocol was repeated three times for each blend, following which, the fine particle dose (FPD) and fine particle fraction of the loaded dose (FPFLD) were determined. The FPD represented the mass of drug that was collected on stages 3-8 of the NGI.

(100) Performance data suggested a FPD for sample 3 in a 500/25 formulation of 79 g and 11 g for FP and SX, respectively as shown in FIG. 30. As shown in FIG. 21 this translated to a fine particle fraction of 44% for SX compared with 15.8% for the engineered FP.

(101) FIG. 31 shows the FPFLD measured as described above, for formulations containing 500 g of Sample 3 FP and 50 g of SX (Sample_3 500/50), 500 g of Sample 3 FP and 25 g of SX (Sample_3 500/25) and Advair (Advair.sub. 500/50) which contained 500 g of FP and 50 g of SX. This Figure shows that in Sample_3 500/50 and Sample_3 500/25, SX had a much higher FPFLD than Advair_500/50. The FPFLD of SX of the Sample_3 500/25 was higher than the Sample_3 500/50.

(102) This implies that the engineered FP has a dramatic effect on both the content uniformity of the blend and facilitates an increase in SX with respect to FPF, also implying that for a given formulation using engineered FP significantly less SX can be used in the blend to achieve comparable FPF for both FP and SX. These data suggest that using engineered FP, prepared by the current invention, it is possible to formulate combination DPI products containing half of the nominal strength currently deployed in the Advair product.

Example 7

(103) The surface roughness and surface area of micronized Fluticasone Propionate (FP), sample 3 of FP prepared by the current invention and sample 4 of FP, as described in Example 2, were determined using atomic force microscopy (AFM) and BET surface area analysis, respectively. The roughness of imaged areas was quantified using the mean (R.sub.a) and root mean square (R.sub.q) of the variations in the height of the imaged surface. Furthermore, the surface area of samples was determined by a five-point BET nitrogen adsorption analysis.

(104) The surface topography of the FP samples was investigated with TappingMode atomic force microscopy (AFM) using a Multimode AFM, J-type scanner, Nanoscope IIIa controller (all from DI, Cambridge, UK) and a silicon tip (model number OMCL-AC240TS, Olympus, Japan) to image three randomly selected 1 m1 m square areas on the surface of particles of each material with a resolution of 512512 pixels and a scan rate of 1 Hz. The roughness of imaged areas was quantified using the mean (R.sub.a) and root mean square (R.sub.q) of the variations in the height of the imaged surface, as calculated by the AFM software using the following equations:

(105) $R_a=\frac{1}{n_p}\underset{i=1}{\overset{n}{.Math.}}.Math.y_i.Math.$$R_q=\sqrt{\frac{1}{n_p}\underset{i=1}{\overset{n}{.Math.}}{y}_i^2}$
where n.sub.p is the number of points in the image and y.sub.i is the distance of point i from the centre line.

(106) The specific surface areas of the FP samples were measured using a Gemini 2360 surface area analyser (Micromeritics Instrument Corporation, Norcross, USA). A five-point BET nitrogen adsorption analysis was carried out after degassing the samples for 24 hours in a FlowPrep 060 degasser (Micromeritics Instrument Corporation, Norcross, USA).

(107) The results are summarised in Table 2 below:

(108) TABLE-US-00002 TABLE 2 Sample Ra( nm) Rq( nm) Surface Area (m.sup.2/g) Micronized 30.97 (12.25) 45.07 (11.76) 6.55 Sample 3 53.79 (2.11) 72.11 (1.35) 10.79 Sample 4 11.20 (1.55) 16.52 (3.13) 7.49

(109) Surface roughness analysis of the samples suggested that sample 3 possessed the greatest surface roughness, whereas sample 4 (not prepared by the current invention) was the smoothest. Sample 3 had a greater surface area than the other samples, which may be related to the materials roughness. The R.sub.a and R.sub.q values, and the surface area of sample 3 are greater than the micronized sample and sample 4.

(110) The standard deviation of the R.sub.a and R.sub.q values for the micronized sample is much greater than samples 3 and 4. This could indicate a greater variance of surface roughness for micronized sample than samples 3 and 4.

(111) FIGS. 32 and 33 show AFM contour plots for surface roughness for samples 3 and 4 respectively. These show that sample 4 is much smoother than sample 3. The difference in the contour plots shows that sample 3 has a different surface roughness to sample 4 and this is reflected in the R.sub.a and R.sub.q values.

Example 8

(112) A solution of Fenoterol hydrobromide (10 g) in methanol (200 mL) was prepared then and spray-dry with Buchi-B290 using twin-fluid nozzle with 0.7 mm orifice with a supporting nitrogen flow rate of 35-40 m.sup.3/h (100% Aspirator), at flow rate of 9 mL/min (30% Pump) and nozzle clean setting 2. Inlet temperature is 78 C. and outlet temperature 38 C. Diisopropyl ether (300 mL) was charged to a stirred 500 mL maximum volume ultrasonic vessel connected to the bottom of the B-290 cyclone and thermoregulated at 5 C. The spray dried product was collected into the ultrasonic vessel operating at 40 W continuous power for 2 hr, following the addition of the first particles of amorphous Fenoterol hydrobromide. The particles were recovered by spray drying the suspension with a Buchi-B290 as above with inlet temperature is 110 C. and outlet temperature 50 C. The data and particle SEM images for this example are shown in FIGS. 17-20.

(113) FIG. 17 shows an SEM image of commercial micronised Fenoterol hydrobromide. FIG. 18 shows the particle size distribution data for samples of particles processed by this invention namely GR005/180/A4 and GR005/179/C. FIG. 19 shows a comparative Fine Particle Fraction (FPF) data using a commercial HandiHaler inhalation device whereas test rig. Comparing the row for FPF [%] on FIGS. 19 and 20, the increase in FPF varied between 30 and 117%.

(114) FIG. 21 shows the dispersive surface coverage v. surface energy for micronised Budesonide and Budesonide particles of the present invention. IGC was used to measure surface energy of the particles of this invention. IGC can be carried out with two sets of conditions. At finite dilution the adsorption isotherms can be derived from peak profiles and used to calculate adsorption energy distributions. Secondly at infinite dilution amount of solutes close to the detection limit of the instrument are injected and in this case the solute-solute interactions are small and only solute-sorbent interactions influence the measured retention time. Only limited adsorbance (coverage) at particularly high energy sites are analysed as shown on the left of FIG. 21. As the amount of solute is increased to finite dilution ultimately 100% coverage is achieved giving rise to adsorption on all sites of the particles regardless of varying surface energy. The particles of the invention are characterised by having isoenergetic distribution of surface energy as shown quite clearly in FIG. 21. The surface energy is very similar and near identical at both finite and infinite dilution for particles prepared by the preferred method of this inventions, whereas typical micronized particles show dramatic variances at finite and infinite dilution.