WO2005055985A1 - Metered dose inhalation preparations of therapeutic drugs - Google Patents

Abstract

Hydrophobic drugs have substantially greater stability in the presence of hydrofluoroalkane propellants for dispensing from metered dose inhalers, when formulated with PVP and a polyhydroxylated polyalkene, such as PVA.

Description

METERED DOSE INHALATION PREPARATIONS OF THERAPEUTIC DRUGS

The present invention relates to preparations of therapeutic materials, stabilised with polyvinyl alcohol, for pulmonary delivery, and methods for their preparation.

A metered dose inhaler (MDI) can simply be described as a system consisting of a therapeutic agent suspended or dissolved in a pharmaceutically acceptable, highly volatile propellant (classically a chlorofluorocarbon, CFC) (Noakes, 2002). This, together with a delivery device, which holds the formulation and is used to actuate the dose from a metering valve, forms a simple means of delivering respiratory drugs to the lungs (Smyth, 2003). Although both the formulation additives and devices have been continuously evolving over the last 50 years, the basic concept, of delivering respirable particles to the lungs using a volatile inert solvent, remains unchanged. The MDI's simple design has made this delivery system cost effective to manufacture and easy to use, and has resulted in it becoming the most popular mechanism to deliver respiratory drugs to the lungs today (Ross and Gabrio, 1999).

Beclomethasone dipropionate (BDP) was initially marketed for the treatment of asthma in the form of a suspension based, metered dose inhaler (MDI), using a mixture of chlorofluorocarbon based propellants, one such formulation being Becotide®. However, as CFC compounds are phased out, this and other products are progressively having to be reformulated.

Currently, there are only two propellants available for commercial use in MDIs, both of which are hydrofluoroalkanes; HFA 134a and HFA 227 [1,1,1,2- tetrafluoroethane (HFA134a) and 1,1,1,2,3,3,3-heρtafluoropropane]. The preferred formulation of BDP with either of the HFAs in a suspension based MDI, as a direct combination, is not possible, due to physical and/or chemical instability. In addition, the surfactants classically used to stabilise CFC MDIs are not soluble in HFAs and, so, are ineffective. For example, oleic acid, sorbitan trioleate and lecithin, the three most commonly used surfactants in CFC formulations, and which also have current regulatory approval for pulmonary delivery, each show less than 0.02% solubility in either of the HFA propellants (Vervaet and Byron, 1999). Since the introduction of the HFAs, two main strategies to formulate therapeutic agents with these propellants (McDonald and Martin, 2000) have been followed. The first is to employ a pharmaceutically acceptable co-solvent to the HFA MDI (Brambilla et al., 1999; Ganderton et al., 2002) to increase the solubility of the drug, excipients, or both, within the HFA propellant. Selection of the correct co-solvent allows either use of the traditional MDI stabilising excipients with HFAs or, in the case of complete miscibility of the therapeutic agent in the propellant, development of a solution MDI. Ethanol is the first co-solvent to be used in this context, and has been shown to increase the solubility of both the traditional MDI stabilising excipients and hydrophobic therapeutic agents within HFA 134a. Incorporation of this co-solvent with BDP in HFA 134a has resulted in the first commercially successful HFA based solution MDI, QNAR®, first marketed in 1998 (Schultz and Schultz, 1995).

A solution MDI, such as QNAR®, generates respirable particles in a different manner to more traditional suspension formulations. Within a suspension, MDI particles of a defined size have already been manufactured, and simply require safe storage and delivery by the device. However, a solution uses the design of the device and the energy created by the evaporating solvent to form the particles upon actuation of the metering valve. The size of the particles ejected from a solution MDI are, therefore, heavily dependent on the actuation orifice diameter and the device design (Lewis et al., 1998). Several research groups have demonstrated that optimisation of these two parameters can potentially produce a dramatic increase in the delivery efficiency of the MDI compared to suspension based formulations (LeBelle et al., 1996; Stein, 1999). QNAR® is defined in its product specification as delivering 60% of its total dose with particles with a mass medium aerodynamic diameter (MMAD) of less than 3.3 μm, which is significantly smaller than conventional suspension based MDIs. The smaller particles manufactured post-actuation by this formulation/device combination is thought to be the reason behind the greater lung deposition.

Although QNAR® is a much more efficient formulation and delivers a high proportion of its metered dose to the lung, there are still major fundamental flaws with this formulation approach, including a lack of specific drug targeting, reduced chemical and physical stability and a loss of control over the dissolution rate (Leach et al., 2002). The lack of control over the specific targeting within the deep lung has recently been studied by Hochhaus et al (1998). This group described mean pulmonary residence time as the major influence on pulmonary targeting of steroids. They showed that solution MDIs had a much lower pulmonary residence time compared to suspension formulations, and suggested that this could result in a lack of lung steroid receptor specificity, hence an increased chance of side effects (Hochhaus et al, 1998).

However, by far the most difficult problem to overcome when manufacturing solution MDIs is the reduction in chemical stability (Sonie et al, 1992). Blondino and Byron (1998) investigated the effects of a solution formulation on the chemical stability of a model drug, acetylsalicylic acid. Results from this work indicated that inclusion of a co-solvent to enhance the drug-excipient-propellant compatibility also increased the chemical degradation of the drug (Blondino and Byron, 1998). This was found to be dependent on the concentration of surfactant. Furthermore, within a solution formulation the drug is exposed to the significant levels of dissolved water taken up in the HFA propellant (Nervaet and Byron, 1999). This can also induce chemical degradation.

Thus, manufacturing an MDI formulation as a solution removes the prime advantage of the dosage form, which is to provide a protective apolar environment, which enhances both chemical and physical stability.

The second approach taken to formulate HFA based MDIs is to suspend the therapeutic agent within the propellant. A suspension MDI relies on minimising the compound's solubility in the HFA, whilst maximising the physical compatibility of the particulate interactions. The availability of only two HFA propellants means that matching the physical properties of the raw drug with the HFA propellants to achieve a stable suspension requires alteration of the chemical properties of the drug, or the addition of further formulation excipients. Fluticasone propionate (Flixotide Evohaler®), salbutamol sulphate (Nentolin Evohaler®) and salmeterol xinafoate (combination product Serevent®) are all formulated in HFA 134a as suspensions. Each of these therapeutic agents is directly compatible with HFAs and, hence, the reformulated version of the product is a direct copy of the CFC formulation. For some of these compounds it was necessary to develop a new variation in the salt form to allow direct combination with the propellant. However, other respiratory drugs, such as BDP, budesonide, nedocromil sodium, and sodium cromoghcate, are physically or chemically incompatible in HFA propellants alone, even with simple changes to the chemical nature of the compound, and therefore require additional stabilisation using a surface-active agent. Although direct combination of raw drug and HFA propellant has successfully produced some commercial formulations, there are still currently marketed drugs that cannot be reformulated in this manner.

An ideal excipient to stabilise an HFA formulation should be chemically inert, biologically compatible, manufactured commercially, have a quick excretion or degradation pathway and enhance the formulation stability and/or delivery. Several classes of compounds have been suggested to meet these criteria, including oligolactic acids, acyl amide-acids and mono-functionalised (M) polyethylene glycols, to identify but a few. As yet, none of these excipients has made it to market in a commercial formulation.

Although there is a broad base of surface active agents from which to choose for a potentially new formulation additive, it is important that the stabilising excipient bridges the incompatibility gap between the active agent and the propellant. Historically, this has been difficult to achieve, as it requires the stabilising excipient to have strong interactions with both the solute and the solvent which, in MDIs, can have very different physical and chemical characteristics. The stabilising agent must be soluble enough in the propellant to exist in an extended conformation whilst still being physically drawn to the surface of the drug to allow adsorption and, hence, protection thereof. If the stabilising excipient is too soluble in the propellant, it will not adsorb to the surface of the drug. Likewise, if it is not soluble enough in the propellant, then it will aggregate and confer no stabilisation properties. Furthermore, the stabilising excipient should allow the drug to remain fully insoluble in the propellant as, in partially soluble systems, Oswald Ripening will occur, causing the drug to cake on standing, resulting in heterogeneous formulation and inconsistent drug dosing.

One possible method to optimise the interaction of stabilising excipients with both the therapeutic agent and the HFA solvent is to "fix" a microfine coating of the excipient onto the surface of the respirable particles, prior to dispersion in the propellant. This can be achieved using a range of manufacturing techniques, including spray-drying, freeze drying, spray-freeze drying, supercritical fluid technology and potentially electrospray atomisation. However, spray-drying is currently the most developed of these manufacturing techniques, and allows the stabilising excipients to concentrate at the drug surface. Upon interaction with the water/air interface during the manufacturing process, the surfactants will naturally arrange to attain a maximum reduction in the surface free energy. The final product will consist of a therapeutic agent of a respirable size that is coated with a uniform microfilm of surfactant. This system can be improved if amphiphilic surface active molecules are used as the surfactants. Using a molecule with a dual functionality will promote the internalisation of most compatible functionality with the therapeutic agent, leaving the opposing chemical moiety externalised. Upon dispersion in the HFA propellant a physically stable suspension is achieved, if the surfactant's externalised moieties are compatible with the HFA, regardless of the internal functionality or therapeutic agent.

Although in previous work trehalose has been shown to stabilise the physical characteristics of therapeutics in microparticulates, these studies have been predominately focused on spray-drying compounds from a solution (Cardona et al., 1997; Davidson et al., 2003; Liao et al., 2002; Librizzi et al., 1999; Zhang and Zografi, 2001). Spray-drying a suspension does not allow the production of a homogenous dispersion of the drug and excipients within the microparticulate, it simply facilitates the accumulation of excipients at the surface of the suspended material, such as BDP.

Suspension stabilisation of aqueous formulations using non-ionic polymers such as PNA and PNP has been previously described by several groups (Bagchi, 1973; Croot et al., 1995; Srinivasa et al., 2003; Zerfa and Brooks, 1998). Furthermore, PNA in combination with other polyols has also been shown to successfully stabilise proteins in HFA 134a (Yong-Hong Liao, 2002). However, PNP was shown to be less compatible with HFAs (Dawson, 1997). Dawson et al. (1997) directly combined PNP with necoramdil sodium in HFA 227a. However, the formulations produced by this group only produced low quantities of the fine powder fraction.

Polyvinyl alcohol (PNA) has been well characterised in an aqueous environment and is known to undergo significant conformational changes to minimise surface free energy (Nguyen, 1996). The polymer contains two main functional groups, a hydrophilic alcohol group and a hydrophobic acetate group, which can orientate in numerous positions to facilitate absorption at various interfaces (Boury et al., 1995). The ratio of these two functional groups (percentage vinyl hydrolysis), and the molecular weight of the polyvinyl alcohol, influences the thickness and characteristics of the absorbed layer (Chibowski et al., 2000).

WO 01/58425 discloses preparation of drugs, such as BDP, with PNA for dry powder inhalers. PNA was used to mitigate aggregation of the particles.

WO 95/15151 discloses pharmaceutical formulations for aerosol delivery and comprising the therapeutic agent in combination with a protective colloid, which may include PNA, and an HFA.

To date, the only commercially available HFA based BDP MDI remains QNAR^®, incorporating ethanol as a co-solvent to produce a solution HFA MDI. There are no HFA suspension-based products. Although solution MDIs have been shown to deliver a high proportion of the actuated dose to the deep lung, this is due to the production of smaller particles by the inhaler. Formulation of an inhaled drug as a solution MDI results in a loss of drug targeting, as controlled release profiles cannot be attained, and there is also increased susceptibility to chemical degradation, a lack of control on the physical characteristics of the drug, and a heavy dependence on the MDI device.

Therefore, there is a requirement to develop alternative approaches to reformulate HFA based MDIs.

Surprisingly, it has now been found that hydrophobic drugs, such as BDP, have substantially greater stability in the presence of HFAs, when formulated both with polyhydroxylated polyalkenes, such as PNA, and also polyvinylpyrrolidone (PVP).

Accordingly, in a first aspect, the present invention provides a formulation of a therapeutic substance suitable for delivery to a patient by a metered dose inhalation device, the formulation comprising a substantially dry powder preparation of the substance with a stabilising amount of polyvinylpyrrolidone and a polyhydroxylated polyalkene, in combination with one or more propellants therefor. In an alternative aspect, the present invention provides a formulation of a therapeutic substance suitable for delivery to a patient by a metered dose inhalation device, the substance being formulated in one or more propellants and/or cosolvent, characterised in that the therapeutic substance is first prepared as a substantially dry powder in the presence of polyvinylpyrrolidone and a polyhydroxylated polyalkene, prior to formulation with propellant.

It is preferred that the therapeutic substance be a hydrophobic drug. For convenience, reference will generally be had, herein, to beclomethasone, or beclomethasone dipropionate, or BDP, but it will be appreciated that such reference also incorporates reference to all suitable therapeutic substances, and especially hydrophobic drugs, unless otherwise stated, or apparent.

Suitable therapeutic substances include, for example: the corticosteroids, such as BDP, budesonide, flunisolide, triamcinolone acetonide, and fluticasone dipropionate; anticholinergic drugs, such as ipratropium bromide; the leukotrienes, such as montelukast, zarfirlukast; cannabiods; and antiemetcs, such as scopolamine.

It is surprising that such water soluble synthetic polymers as PNA and PNP possess the ability to enhance the physical and chemical stability of MDI inhaler formulations of hydrophobic drugs in HFAs.

In particular, it is surprising that hydrophobic drugs can be stabilised in hydrophobic propellants by excipients which are highly hydrophilic.

It is generally preferred that one, or both, of the polymers should generally be capable of forming strong interactions with both the therapeutic agent and the propellant in which it is suspended. The polymer(s) can act to protect the therapeutic agent during spray-drying. They may also act to improve the physical stability within the formulation suspension and may influence the dissolution of the drug on entry into the respiratory tract.

The BDP HFA MDI suspensions of the present invention match the delivery efficiency of the CFC product, Becotide. Although QNAR appears to deliver higher FPF's, it will be appreciated that good delivery efficiency is not the only criterion, and that, in an ideal drug delivery system, other aspects of the drug's delivery needs to be controlled, so that physical and chemical stability, specific site of delivery and release rate from the formulation, must also be taken into account. With a solution MDI, there is a lack, both of control of the chemical and physical stability, and of the release of the therapeutic agent, so that an increase in pulmonary residence time cannot be achieved, and specific drug targeting is not possible. With the suspension based formulations of the present invention, control of all these characteristics is now possible for hydrophobic drugs.

Preferred hydrophobic drugs include: the corticosteroids, such as BDP, budesonide, flunisolide, triamcinolone acetonide, and fluticasone dipropionate; anticholinergic drugs, such as ipratropium bromide; and the leukotrienes, and generally include any generally hydrophobic drug capable of having a therapeutic effect via respiratory, nasal or generally naso-pharyngeal surface membrane administration from a pressurised propellant. The hydrophobic drug may act in situ, or systemically.

The term "hydrophobic", in relation to drugs, is taken to mean those drugs that cannot readily be formulated in water without the use of a co-solvent and, in this respect, a drug that has a high partition coefficient (log P > 1.5) may be considered to be hydrophobic.

In particular, we have now found that BDP, for example, can be formulated with PNP and a polyhydroxylated polyalkene in an MDI to retain structural integrity during the production of respirable particles and formulating the particles with HFA propellant.

It is an advantage of the present invention that, by combination with PNP and a suitably substituted polyhydroxylated polyalkene, the therapeutic substances are now able to be provided in formulations which are stable, even in the presence of haloalkane propellants. It is a particular advantage that such stability is demonstrated in the presence of HFAs, but it will be appreciated that such stability is also demonstrated in the presence of other propellants, such as CFCs, and alkanes, such as butane and propane or combinations of said propellants.

Preferred propellants are the haloalkanes, and it is preferably envisaged that HFAs are used as propellants for MDIs in formulations of the present invention. However, it will be appreciated that the invention also extends to the use of CFCs and other alkanes, for example. The backbone of the propellant will generally be an alkane, whether substituted or unsubstituted, and may be straight or branched. Where branched, it is preferred that there only be one branch. Straight chains of the lower alkanes are preferred, especially C_2-4.

The preferred HFAs for use in the present invention are HFA- 134a and HFA-227.

The properties of polymers such as PNP may usefully be reported in terms of the Fikentscher K-value, derived from solution viscosity measurements, generally at 25 °C. The relationship between the viscosity in water at 25°C, the K-value, and the approximate molecular weight of PVP is shown in Table 1, below.

Table 1 Relationship of viscosity, K- Value, and approximate molecular weight for PVP

Although PNPs with K values of up to 120 and beyond are known, it is generally preferred to employ those with K values of up to 50, while studies have indicated that PNP having a K value of less than 30 is generally safe for inhalation. PVP having a K value of no more than 20 is most preferred, for the purposes of the present invention. In a preferred embodiment, Povidone K15 is employed in the present invention. It will be appreciated that the K value is not a guarantee of the uniformity of the molecular weight of the individual PVP molecules, but that the K value provides a guide to the average MW. Suitable polyhydroxylated polyalkenes for use in the present invention preferably have the structure

-(CH₂-CHOR)_n-

where R is the same or different from one monomeric unit to the next, and is hydrogen, lower alkyl, lower alkenyl, lower alkanoyl, lower alkenoyl or is a bridging group between adjacent monomers, such as a lower diacyl group. By "lower" is meant 1 to 6 carbon atoms, other than the carbonyl carbon, where present, with 1 to 4 being more preferred, and 1 or 2 being more preferred.

Examples of suitable polyhydroxylated polyalkenes include PVA, PVAc (polyvinylalcohol and polyvinylacetate, respectively), polyvinyl alcohol-co-vinyl acetate (PVAA), poly( vinyl butyral) and poly(vinyl alcohol-co-ethylene).

PVA is generally prepared by the hydrolysis of PVAc, and the level of hydrolysis may be as low as about 40% through to substantially complete hydrolysis, such as 98% or higher. Low levels of hydrolysis correspond to lower levels of hydrophilicity/higher levels of hydrophobicity, which can affect the formulations of the present invention. While levels of 98% hydrolysis are useful, it is generally preferred that the level of hydrolysis be in the region of 50 to 90%, with a level of about 80% being a preferred embodiment. Thus, preferred formulations are where the PVA is a hydrolysate of PVAc, and the level of hydrolysis is between 40% and 100%, preferably 50 and 90%. More preferably, hydrolysis is preferred to be 70% or above, and is preferably between 80% and 90%, especially where the primary propellant is HFA 134a. Hydrolysis of 98% provides good results with HFA 227, as do ranges down to 70%, although usefulness drops off below about 80% hydrolysis.

The size of the polyhydroxylated polyalkene compounds is not critical to the present invention, and PVA may range from a molecular weight of 9kDa through to about 500kDa, with 9kDa to 50kDa being more preferred. Where PVA is used as the sole polyhydroxylated polyalkene, then a preferred molecular weight is in the region of lOkDa. It will be appreciated that molecular weights for the polyhydroxylated polyalkenes are necessarily highly approximate, as the methods for their preparation necessarily result in a spread of molecular sizes. PVP/PVA copolymers are also available, and may be employed in the present invention, as a substitute for either or both of PVA and PVP. For example, Plasdone^® copolyvidonum is a synthetic water-soluble copolymer consisting of N-vinyl-2- pyrrolidone and vinyl acetate in a random 60:40 ratio, and is also known as Copolyvidonum Ph Eur, Copolyvidon DAB, and Copolyvidone JSPI, BP. The K-value for Plasdone S-630 copolyvidonum is specified as being between 25.4 and 34.2, and is similar to Plasdone K-29/32 povidone.

Suitable amounts of each of the PVP and the polyhydroxylated polyalkenes excipients range from about 1% to about 200%, preferably 5% to about 200%, by weight of the therapeutic substance, although there is little advantage to be seen in the provision of large amounts of either. In general, a suitable amount of each excipient, or excipient type where more than one polyhydroxylated polyalkene is used, for example, is between about 1% and about 60%, preferably between about 10% and about 50%, by weight of the therapeutic substance, with a range of about 20% to about 40% being preferred.

Prior to formulation with the haloalkane propellant, it is preferred to blend the therapeutic agent with the PVP and polyhydroxylated polyalkene in an aqueous vehicle, before drying. The aqueous vehicle may be any suitable, and will typically be selected from saline or a suitable buffer, such as phosphate buffered saline (PBS), although deionised water may also be used, if desired.

It will be appreciated that some formulations may comprise two or more populations of particles for administration. In such instances, the quality of PNP and polyhydroxylated polyalkenes may be selected as appropriate to each substance, and combined with propellant once prepared. It is also possible that, where there are two or more active substances, any two or more may be formulated together.

The powdered products resulting from the drying of the aqueous preparation may be achieved by any suitable drying process, including freeze-drying, spray-drying, spray-freeze-drying, supercritical drying, co-precipitation and air-drying. Of these, spray-drying and spray-freeze-drying are preferred, as these result in fine powders which generally require no further processing. However, if required, the dried products may be further processed to reduce the size of the resulting particles to an appropriate level. In particular, it is preferred that the aerodynamic diameter of the particles of the powder used in the formulations of the present invention is between about 1 μm and 50 μm, more particularly between about 1 μm and 12 μm, and even more particularly between about 1 μm and 10 μm. The dried powder is then brought into contact with the propellants under conditions suitable for storing in a reservoir useful in an MDI.

It is a particular advantage of the present invention that the stability of the particles prepared as described above is considerably greater than anything provided in the art, and preferred formulations of the present invention comprise only the active ingredient(s), PVP, polyhydroxylated polyalkene(s), and propellant(s). Thus, formulations of the present invention provide long-term stability of activity of the therapeutic substance, as well as ensuring consistency of dosing with time.

It will be appreciated that the present invention further provides a powdered formulation of a therapeutic agent, PVP and a polyhydroxylated polyalkene suitable for incorporation with a haloalkane propellant for dispensing from a metered dose inhaler.

The present invention further provides a metered dose inhalation device provided with a reservoir comprising a haloalkane propellant prepared with a therapeutic substance, PVP, and a polyhydroxylated polyalkene.

Doses delivered by the MDIs of the present invention will be readily determined by those skilled in the art and as appropriate to the condition to be treated. In general, doses will vary with the size and age of the patient and can be readily determined by calculating the concentration of the active ingredient in the propellant preparation.

In the accompanying Examples, we have demonstrated that the grade of both PVA and PVP has some influence on BDP stabilisation in HFA propellants. For example, formulations containing more greatly hydrolysed PVA, in the range of 80- 98%, together with PVP having a lower, and generally preferred, K-value of around K15, were generally more effective in delivering a high stage 2 deposition within the twin-stage impinger, so that such formulations could be considered to be more efficient.

Less hydrolysed PVA, such as 70 % hydrolysis, combined with lower K-value PVP, such as K15, provided 100%) release of the drug in the fastest time frame in the dissolution studies, thereby also providing an advantage. However, increasing the molecular weight of PVA, when combined with lower K-value PVP, such as K15, lead to a more controlled BDP release rate, whist still delivering a large fraction of the dose to stage 2 of the twin-stage impinger.

BDP formulated with PVA 80% hydrolysed and PVP K15 delivered the highest stage 2 dose in the twin-stage impinger using HFA 134a as the propellant, whilst PVA 98% hydrolysed was found to be the most effective grade of polymer to combine with PVPK15 to suspend BDP in HFA 227. It was difficult to distinguish between PVA 80% and 88% when using HFA 227, as they were both effective in stabilising the BDP (as illustrated by a high stage 2 twin-stage deposition). Increasing the molecular weight of PVP or PVA did not appear to improve efficiency of delivery or enhance dissolution of the BDP from the formulations.

The present invention will be further illustrated with regard to the accompanying Figures, referred to in the Examples, in which:

Figure 1 shows a powder X-Ray diffraction pattern from the raw micronised BDP;

Figure 2 shows dissolution profiles of the coated BDP particles compared to the raw drug in simulated lung fluid (■) BDP PVA80% formulation (A) BDP + PVA + PVP +

HA formulation (♦) raw BDP;

Figure 3 shows the impaction data for the five BDP MDI formulations and 2 controls determined in vitro using a twin-stage impinger;

Figure 4 shows the in vitro deposition profile of five BDP HFA 134a MDIs, testing the effects of varying the grade of polymer utilised as a stabilizer in the formulations;

Figure 5 shows the in vitro deposition profile of five BDP HFA 227 MDIs testing the effects of additional stabilizing excipients;

Figure 6 shows the in vitro deposition profile of six BDP HFA 227 MDIs, testing the effects of varying the grade of polymer utilised as a stabilizer in the formulations;

Figure 7 shows the dissolution of three forms of BDP microparticles within simulated lung fluid over a 24 hour time period;

Figure 8 shows the dissolution of three BDP microparticles containing three grades of

PVA varying in molecular weight within simulated lung fluid; Figure 9 shows the dissolution of four BDP microparticles containing three grades of PVA varying in percentage hydrolysis, within simulated lung fluid; and Figure 10 shows the dissolution of four BDP microparticles containing three grades of PVA varying in percentage hydrolysis, within simulated lung fluid.

The present invention will now be further illustrated by the following, non- limiting Examples.

EXAMPLE 1 METHODS

Manufacturing formulations

All the formulations were manufactured using the Bucchi 191 mini spray dryer (Bucchi, Darmstadt, Germany). The aspiration rate was set as 70%, the material feed rate was 4 ml min^"1 and the Inlet temperature was set to 180°C. The feed suspension was pumped through a spray atomisation nozzle which combined the liquid with an 80 ml hr^"1 airflow. The outlet temperature was consistently found to be in the range 84 - 86°C.

The spray-drying suspensions were produced by adding the PVA to 500 ml of deionised water. This solution was heated to 50°C and stirred for approximately 20 min using a heated stirrer (Stuart Scientific, Redhill, Surrey, UK). When the polymer was completely dissolved, the micronised beclomethasone 17,21 -dipropionate (Airflow Co., Buckinghamshire, UK) was added. A stable suspension was achieved by stirring continuously for a further 20 min, upon completion of which the remaining excipients were added. The final suspension was stirred for another 20 min and this was continued as the mixture was pumped through the spray dryer. Four formulations were manufactured in total as detailed in Table 2, below. The PVP was K15 grade and the PVA 80% hydrolysed molecular weight (M_w) 8,000-10,000 (Sigma Aldrich, Gillingham, UK). The hyaluronic acid (HA) was added as a viscosity modifying agent and had a 400,000 M_w. The PVA 40% hydrolysed was supplied by Polysciences, Warrington, USA and had a M_w 23,000. The product from the spray-drying process was collected and weighed into a glass vial. The samples were stored under silica desiccation at room temperature.

Table 2 Compositions of the BDP spray-dried formulations (Total quantities in 500 ml water).

Excipients BDP PVA 80% BDP + PVA BDP + PVA + PVP BDP PVA + PVP + HA 40%

Beclomethasone 5.0 g 5.0 g 5.0 g 5.0 g

Polyvinyl alcohol 0.5 g ^a 0.5 g ^a 3.0 g ^a 3.0 g ^b

Polyvinyl Pyrrolidone - 3.0 g 0.5 g -

Hyaluronic acid - - 0.5g - ^a Polyvinyl alcohol 8-10,000 M_w 80% hydrolysed. Polyvinyl alcohol 40% hydrolysed M_w 23,000.

The metered dose inhalers were manufactured by adding approximately 50.0 mg of the spray-dried powder into a PET canister (BesPack, Kings Lynn, UK). A 25 μL canister valve (BesPack, Kings Lynn, UK) was crimped in place, using the pamasol MDI filler (Pamasol, Pfaffikon, Switzerland) and 20.0 g of HFA 134a (Dupont, Willington, Germany) was pressure filled into the can via the valve. The formulation was then sonicated in an ultrasonication bath (Decon, Hove, UK) for 1 minute, to ensure particle separation, and stored, valve up, at room temperature for 24 hours.

Powder X-ray diffraction

The samples were run on a Siemens D500 refractometer (Siemens, Worcester, UK) using Cu-Kα. The diffraction pattern was collected between theta values of 3 - 60°, the step time was 4 s per 0.020°. The wave length was 1.54. Raw BDP was used as the control and 3 formulations were selected that would provide the greatest resolution for the experiment. Four resolved peaks, with identical theta values, were selected from each diffraction pattern and fitted with a Lorentz mathematical model using Origin^® software. The area under the fitted curve was integrated. Peak areas were compared to the control to obtain % crystallinity.

Particle size analysis

The spray-dried powders were assessed using the Model 26C4L Malvern laser diffraction particle size analyser (Malvern Instruments Ltd, Malvern, UK). The Malvern was set up using the liquid dispersion system. A Span 80 (Sigma Aldrich, Gillingham, UK), 1% cyclohexane (Merck, Poole, UK) mixture saturated with micronised beclomethasone 17,21 -dipropionate was used as the dispersion media. Samples were prepared by sonicating 2 mg of sample in 2 ml of the dispersion media for 40 min. The particle size was measured using the 63 mm (0.5 - 110 μm) lens set at focal length of 14.5 cm, whilst stirring the cell on 75% of full power. The samples were added dropwise in to the stirred cell until the desired obscuration was achieved. Each sample was measured in triplicate, and 3 batches from each sample were analysed.

Thermogravimetric analysis

A TGA 50 thermobalance (Mettler, Beaumont Leys, UK), in combination with a TCI 5 controller (Mettler, Beaumont Leys, UK) and an M5 micro balance (Mettler, Beaumont Leys, UK) was used to assess the volatile content of the manufactured powders. The heating rate of the thermogrametric balance was set at 5°C per minute over a range of 30°- 150°C. A blank run, using identical parameters, was performed prior to each sample. The % weight loss was calculated between the limits of 30°- 110°C for each sample.

Dissolution Studies

The dissolution medium was based on work by Gambel to mimic in vivo lung fluid (Gambel, 1967). Six dissolution stations were run concurrently containing 900 ml of medium which, consisted of 0.0116 moles L^"1 NaCl; 0.027 NaHCO₃; 0.005 glycine; 0.001 L-cysteine; 0.0002 Na citrate; 0.0002 CaCl₂; 0.0005 H₂SO₄; 0.0012 NaH₂P0₄; 1% SDS (all supplied by Sigma Aldrich, Gillingham, UK). The media were adjusted to pH 7.4 with NaCl. The dissolution experiment was carried out in a DT6 dissolution apparatus (Copley, Nottingham, UK) using the paddles designed to specifications listed in the British Pharmacopoeia. Approximately 3-4 mg of each of the samples was loaded directly onto the paddles using double sided sticky tape. Once the water bath had equilibrated to 37°C the paddles were lowered into the media and rotated at 900 rpm. Samples were withdrawn from the dissolution baths at 0, 10, 20, 30 , 40, 50, 60, 120, 180 minute time points. The samples were filtered through 0.2 μm PVDF filters (Whatman, Maidstone, UK) and stored at room temperature for subsequent HPLC analysis. The 1 ml sample was replaced by 1ml of dissolution medium after each measurement. This was compensated for in the final concentration calculations.

Twin-Stage Impinger

The twin-stage impinger (Radleys, Saffron, UK) was set up as per the United States Pharmacopoeia specification. A 70/30 acetonitrile HPLC grade (Merck labs, Darmstadt, Germany) water solution was used as the solvent, both within the apparatus and as a washing agent. The airflow was set to 60 ml min^"1 and the inhaler was actuated 20 times. Between each actuation there was a five second pause with the pump running. The pump was then stopped, the canister removed and shaken for 5 s before the sequence repeated. Each of the stages was washed individually upon completion of the 20 canister actuations. The device was washed into a 50 ml volumetric with stages 1 and stage 2 being washed into 100 ml volumetric flasks. The resulting solutions were analysed using HPLC. All twin-stage runs were completed in triplicate. The Becotide 50^® and QVAR^® 50 formulations were used as received (Allen and Hanbury. Uxbridge, UK and 3M Healthcare, Loughborough, UK).

The liquid chromatography system used for HPLC analysis of BDP consisted of an isocratic Pu 980 Pump (Jasco, Great Dunmow, UK) set at 1.0 ml min^"1, an AS 950 autosampler fitted with a 100 μL injection loop (Jasco, Great Dunmow, UK), a CI-10B integrator (LDC/Milton Roy, Stone, UK) and a chart printer (LDC Milton Roy, Stone, UK). BDP was detected using a 975 UV/VIS detector (Jasco, Great Dunmow, UK) set at 254 run. The column was a Cι₈ 150 mm x 3 μm (Hichrome. Theale, UK).

To prepare these standards, 10 mg of BDP was weighed out accurately into a glass weighing boat. This was added to a 100 ml volumetric flask and made up to volume with mobile phase. The precise weight of the BDP was calculated by subtracting the mass of the empty weighing boat from its original mass including the BDP. A set of serial dilutions was then performed in thelOO μg ml^"1 to 0.5 μg ml^"1 range producing 5 calibration standards. The standards were run in series, preceded by a blank, which consisted of just mobile phase. Two injections were made of each concentration. The samples produced by the twin-stage impinger were run in sequence after the calibration curve.

RESULTS

Spray-dried material characterisation

All the formulations were successfully manufactured using the Bucchi spray dryer.

Table 3

Malvern particle size measurements of the manufactured formulations (n=3) mean ± standard deviation

Formulation D[v, 0.1] μm D[v, 0.5] μm D[v, 0.9] μm

BDP 1.83 ± 0.21 3.13 ± 0.15 4.90 ± 035^™"

BDP PVA 80% 2.54 ± 0.02 4.22 ± 0.04 7.42 ± 0.24

BDP + PVA + PVP 2.40 ± 0.02 3.37 ± 0.02 4.83 ± 0.01

BDP + PVA + PVP + HA 2.36 ± 0.02 4.02 ± 0.15 6.89 ± 0.22

BDP PVA 40% 2.57 ± 0.01 4.42 ± 0.03 7.99 ± 0.07

The particle size measurements of the spray-dried material indicated that all of the batches were of a suitable respirable size, i.e. less than 10 μm, as shown in Table 3, below. Although each of the batches were similar in terms of particle size, they were all larger than the initial raw BDP. The smallest mean particle size (3.37 ± 0.02 μm) was produced using the PVA / PVP polymer combination, and the largest by the combination of BDP with PVA 40% (4.42 ± 0.03 μm).

The diffraction pattern obtained from each of the four samples analysed showed very sharp, intense peaks, many of which were completely resolved. Figure 1 shows a powder X-Ray diffraction pattern from the raw micronised BDP. This type of diffraction pattern, is indicative of a highly crystalline material. However, processing the formulations using spray-drying did reduce the material's relative crystallinity (Table 4).

Table 4

Relative crystalline content of the spray-dried formulations raw BDP used as the control

Sample Peak l Peak 2 Peak 3 Peak4 Average (%) (%) (%) (%)

BDP PVA 80% 83.80 80.69 69.78 80.74 78.75 ± 6.15

BDP + HA + PVA + 49.78 22.89 59.91 48.26 45.21 ± 15.75 PVP

BDP + PVP + PVA 78.40 68.61 72.44 76.42 73.97 ± 4.35

The BDP PVA 80% formulation and the BDP + PVA + PVP formulations conferred the largest protection during the spray-drying process, both batches retaining over 70% of the BDP's crystallinity, as shown in Table 4. There was no significant difference (p > 0.05, ANOVA) between the crystallinity of these two polymer combinations. Addition of HA to the BDP + PVA + PVP formulation dramatically reduced the percentage crystallinity of the material to just over 45 %. The volatile content of raw BDP was relatively low. However, spray-drying the steroid with excipients lowered this volatile content even further. The formulation containing PVA 40% hydrolysed had the lowest volatile content which was significantly different (p < 0.05, ANOVA) from the rest of the compounds tested. The BDP + PVA + PVP formulation and the BBDP PVA 80% formulation which had the highest volatile content were not significantly (p > 0.05, ANOVA) different to the raw BDP. Results are shown in Table 5.

Table 5

TGA data showing the percentage volatile content within the raw drug, BDP and each spray-dried powder. Mean ± standard deviation (n=3)

Sample BDP BDP PVA BDP + PVA BDP PVA BDP + PVA 40% + PVP 80% HA + PVP

% weight loss 1 2.84 0.89 1.60 2.84 2.75

% weight loss 2 3.02 0.90 1.54 2.84 2.70

% weight loss 3 3.13 0.85 1.56 3.10 2.61 Average 3.00 ± 0.14 0.88± 0.02 1.57 ± 0.03 2.93 ± 0.15 2.70 ± 0.07

Unsurprisingly, the crystallinity of the manufactured particles seemed to influence the dissolution profiles of the BDP. The raw crystalline micronised drug produced a relatively consistent release over the 3 hour time period whilst the most amorphous material (the BDP + PVA + PVP + HA formulation) released 100% of the drug within 40 min. The BDP PVA 80% formulation produced an initial burst release within the first 5 min, followed by a more gradual release over the subsequent 2hrs. The results are shown in Figure 2, which shows dissolution profiles of the coated BDP particles compared to the raw drug in simulated lung fluid. ANOVA was used to test the discrimination of these results and there was a significant difference between the three data sets up to the 60 minute time point (p < 0.05). MDI formulation deposition studies The four manufactured MDI formulations were compared to both the raw BDP, suspended in HFA 134a, and to two commercial formulations, Becotide 50®, a CFC based inhaler formulation, and QVAR®, a HFA solution formulation, using the USP twin-stage impinger. Figure 3 shows the impaction data for the five BDP MDI formulations and 2 controls determined in vitro using a twin-stage impinger. With this equipment, the fine particle fraction is defined as the particles collected on stage 2 of the device. Stage 2 has a size cut-off MMAD of < 6.4 μm. The QVAR^® formulation produced the highest fine particle fraction, as expected, with just over 70% of the formulation reaching stage 2 of the impinger. Becotide 50^® formulation produced the second highest particle fraction, with almost 50% of the delivered dose reaching stage 2 of the apparatus. However, using ANOVA, it was shown that there was no significant difference between the stage 2 deposition of the Becotide 50^® formulation, the BDP + PVA + PVP formulation and the BDP + PVA + PVP + HA formulation.

Conversely, incorporation of raw BDP within an HFA 134a MDI formulation produced a low fine particle fraction (< 20%). The majority of the drug delivered by this formulation was deposited on stage 1 of the twin-stage impinger, suggesting that the BDP is agglomerating at some point, producing particles > 6.4 μm. A similar deposition profile was achieved with the BDP PVA 40% formulation, as well as with the BDP PVA 80%, both delivering less than 20% of actuated dose to the second stage of the impinger.

It can be seen that the addition of HA to the formulation caused a loss of almost 50% of the initial crystallinity of the drug. In comparison, using PVA as the stabilising polymer only resulted in a loss of 20% in the raw drug crystallinity. There was no significant difference between the measured crystallinity of the BDP PVA 80% formulation and the BDP + PVA + PVP formulation. The efficacy of PVA as a lyoprotectant has been previously reported (Guiavarc'h et al., 2003). Guiavarc'h et al. (2003) suggested that PVA functioned as a heat stabiliser in a similar manner to other polyols, and used the OH functionality to decrease the water activity at the interface with a therapeutic agent. While it might be expected that HA would counteract this stabilisation, the fact that the combination of PVA and PVP does not reduce the crystalline content is surprising, and cannot be explained with current knowledge.

Water has previously been shown to be influential in the stability of MDI suspensions (Kitahara et al, 1967; Williams and Hu, 2001). However, in the above Example, neither a relatively high nor a low water content seemed to influence suspension stability. This may be due to the hydrophobic nature of the therapeutic agent and the fact the volatile content within the tested formulations remained below 3% which, in absolute terms, is a low water content.

The commercial preparations, QVAR^® and Becotide^®, delivered similar fine particle fractions as detailed in previous studies (Barnes and Nash, 1996; LeBelle et al., 1996; Steckel and Muller, 1998). The efficient delivery of BDP from these formulations was expected, these two products being simply used as controls. However, the test BDP formulations showed vividly different deposition profiles. As all the MDI formulations were manufactured to a specification of < 4.5 μm, prior to dispersion in the HFA propellant, it can be assumed that the low fine particle fractions obtained with the raw BDP and the two PVA formulations (40% PVA and 80% PVA), were as a result of incompatibilities of the formulation with HFA 134a.

The physical stability of suspensions is a heavily researched area and is discussed in more detail in the referenced reviews(Bagchi, 1973; Eirich, 1977; Hiestand.E.N, 1964; van Mill et al., 1984). However, in simple terms a physically stable suspension can be defined as the condition in which the particles do not aggregate and in which they remain uniformly distributed throughout the dispersion. Applied to the current system, if a physically stable MDI formulation is attained, then a uniform dose of individual particles would be released by the metering valve. As the manufactured formulations contain particles that have a volume mean diameter of less than 4.5 μm prior to entering this HFA suspension, a stable system would deliver a high proportion of particles within this size range to the lung (tested using the TSI apparatus). Even taking into consideration the conversion between mean volume diameter (units used for particle size of the manufactured material) and MMAD (units used for particle size with the TSI), if the developed formulations are physically stable, then they should deposit a high proportion of their delivered dose to stage two of the TSI apparatus, which collects particles with a MMAD < 6.4 μm. The fact that this does not occur with the raw BDP and the two PVA formulations (40%PVA and 80%PVA) suggests that, in these systems, the particles are not physically stable. This lack of physical stability may result in the particles irreversibly binding to the walls of the container; forming agglomerates within the formulation canister or during ejection from the metering valve.

Physical instability was not seen with both the PVA PVP coated BDP formulation and the BDP+ PVA + PVP + HA formulation which produced fine particle fractions > 45%. The improved suspension stability of the PVA/PVP polymer combination (with or without HA) within the HFA propellant suggests that either PVA and or PVP are improving the particles' compatibility with HFA 134a. One or both of the polymers promote the uniform suspension of individual particles within the formulation, thereby enhancing the physical stability.

Dawson et al. (1997) directly combined PVP with necoramdil sodium in HFA 227a. However, the formulations produced by this group only produced stage 2 depositions within the twin-stage impinger < 25%. Therefore, an element in currently developed system must be enhancing this stabilising power. One important characteristic of the PVA, PVP polymer combination is that there is evidence that the two polymers can interact (Cassu and Felisberti, 1997). If this occurs, then it is likely that the PVA, which appears incompatible with HFA134a alone, is used as the inner layer to suspend the BDP in the aqueous environment, during the manufacture process. The outer PVP layer is formed during spray-drying, and interacts with HFA 134a to generate the physical suspension stability.

EXAMPLE 2

As in Example 1, the excipients were spray-dried with beclomethasone dipropionate (BDP) and pressure filled within HFA MDIs. Grades of PVA were selected in order to investigate the effects of both percent hydrolysis and molecular weight on the MDIs performance and dissolution. In addition, the molecular weight of PVP was varied to monitor its effects on the formulation. MATERIALS AND METHODS

Formulation manufacture

Micronised BDP was used as received (Airflow Co., UK). A binary BDP/HFA suspension formulation consisting of 50.0 mg of the micronised BDP and 20.0 g of HFA 134a (Solkane Solvay, UK) was manufactured using pressure filling on a Pamasol MDI filler (Pamasol, Switzerland). The formulation was made up in a clear PET canister with a 25 μL valve. Ultrasonication was applied to the BDP suspension for 1 min to ensure dispersion of the powder in the HFA.

The BDP MDI formulations of the invention were manufactured by spray-drying 1.0 g of BDP with the excipients listed in Table 6, below. The polyvinyl alcohol was dissolved in 100 ml of water at 80°C and the BDP suspended with the other excipients within this solution. PVA 70% hydrolysed, Mw 13,000; PVA 80% hydrolysed, Mw 8,000-10,000; 87-89%% hydrolysed, Mw 13,000-23,000; 87-89% hydrolysed, Mw 31,000-50,000; and, 87-89% hydrolysed, Mw 124,000-180,000, were obtained from Sigma Aldrich, UK. Polyvinylpyrrolidone (PVP) K15, Mw 10,000, and polyvinylpyrrolidone K90, Mw 360,000 was also purchased from Sigma Aldrich, UK. Hyaluronic acid (HA 400,000 M_w) was a donation from King's College London, UK, and the trehalose dihydrate was purchased from Sigma Aldrich, UK. The aqueous suspension of BPD was spray-dried on a 191 spray drier (Bucchi, Switzerland) using an inlet temperature of 180°C, material feed rate of 4 ml min^"1, atomisation flow of 70% and nozzle air flow of 800 ml min^"1. The novel HFA suspensions were manufactured by combining 50.0 mg of the dry product from the spray-dried method with 20.0 g of HFA134a (Solkane, UK) or 15.0g of HFA 227a (Solkane, UK) under pressure. In addition to the BDP formulations manufactured, a CFC suspension based MDI Becotide 50^® (Allen & Hanbury, UK) was used as received. Particle size analysis

The particle size of the dry powdered material was measured using a liquid stirring cell on a Model 26C4L particle size analyser (Malvern Instruments, UK). The optical bench was calibrated using a latex standard prior to use. A saturated cyclohexane (Merck, UK), 1% span 80 (Sigma Aldrich, UK) solution was used as the dispersion medium. The sizing method was validated according to ISO 13320 (1999) (data not shown) and employed a ³A power stirring rate, a sonication time of 40 min, 2000 sweeps, measurement path length of 14.5 mm and 63 mm lens. Three measurements were made of each sample and 3 samples were taken from each formulation using a standardised sampling procedure.

Impaction particle size analysis

The twin-stage impinger apparatus is used to model drug delivery in the lung, delivery to the second stage being indicative of ability to deliver to the lung. It was set up and run as per the British Pharmacopoeia (flow rate 60 ml min^"1). A total of 20 actuations were sprayed into the apparatus from each inhaler. Chemical analysis was performed on a Waters Integrated Millennium HPLC system (Waters, UK) using a C₁₈ 150 mm x 5μm Hichrome column (Hichrome, UK), an injection volume of 100 μL, a runtime of 7 min, a 70/30 acetonitrile : water mobile phase at room temperature. The study used filtered and degassed (0.2μm nylon filter, Whatman, UK) HPLC grade solvents (Merck labs, Germany). The washing solutions for the impinger equipment were equivalent to the mobile phase. The particles depositing on the second stage of this apparatus models the delivery of the drug to the deep lung which is its target site for BDP.

Dissolution of BDP formulations

The dissolution medium was based on work by Gambel to mimic in vivo lung fluid (Gambel, 1967). Six dissolution stations were run concurrently containing 900 ml of medium which, consisted of 0.0116 moles L^"1 NaCl; 0.027 NaHCO₃; 0.005 glycine; O.OOl L-cysteine; 0.0002 Na citrate; 0.0002 CaCl₂; 0.0005 H₂SO₄; 0.0012 NaH₂P0₄; 1% SOS (all supplied by Sigma Aldrich, Gillingham, UK). The medium was adjusted to PH 7.4 with NaOH. The dissolution experiment was carried out in a DT6 dissolution apparatus under non-sink conditions (Copley, Nottingham, UK) using the paddles designed to specifications listed in the British Pharmacopoeia. Approximately 3-4 mg °f each of the samples was loaded directly onto the paddles using double sided sticky tape. Once the water bath had equilibrated to 37°C, the paddles were lowered into the **iedia and rotated at 900 rpm. Samples were withdrawn from the dissolution baths at 0, 10, 20, 30 , 40, 50, 60, 120, 300, 480 and 1440 min time points (i.e. over a period of 24 hours). The samples were filtered through 0.2 μm PVDF filters (Whatman, Maidstone, lJK) and stored at room temperature for subsequent HPLC analysis. The volume of liquid removed was replaced with 1 ml of temperature equilibrated media after each Hieasurement. This was compensated for in the final concentration calculations.

Table 6 Composition of the novel spray-dried BDP MDI formulations

Formulation BDP PVA PVP HA Trehalose Polymer grades

BDP PVA l.Og 0.6 g - . _ PVA 80% PVP K15

BDPLW80K15 l.Og 0.6 g O.lg PNA 80% PNP Kl 5

BDPLW80K15T l.Og 0.6 g O.lg l.Og PNA 80% PVP Kl 5

BDPLW80K15HA l.Og 0.6 g O.lg O.lg - PVA 80% PVP Kl 5

BDPLW70K15 l.Og 0.6 g O.lg PVA 70% PVP Kl 5

BDPLW88K15 l.Og 0.6 g O.lg PVA 87-89% low M_w PVP K15

BDPMW88K15 l.Og 0.6 g O.lg PVA 87-89% med M_w PVP K15

BDPHW88K15 l.Og 0.6 g O.lg PVA 87-89% high M_w PVP K15

BDPHW88K90H l.Og 0.6 g 0.6 g PVA 87-89% high M_w PVP K90

BDPHW88K90L ^* l.Og 0.6 g O.lg PVA 87-89% high M_w PVP K90

BDPLW98K15 l.Og 0.6g O.lg PVA 98% PVP Kl 5 volume of water was increased to 300ml in this formulation to decrease the total solid content of the spray-dried suspension.

RESULTS

Formulation manufacture

The excipients were successfully combined with BDP to form a solid dosage form suitable for inhalation (i.e. with a particle size range < 10 μm) in all but one of the formulations, BDPHW88K90H (Table 7, below). Using a high concentration of PVP K90 produced a median particle diameter (Dv, 0.5) of 8.14 μm and 90% cumulative particle size (Dv, 0.9) of 18.34 μm. The large particles formed in BDPHW88K90H would probably deposit in the upper airways and are unlikely to be very suitable for respiratory delivery of BDP. The largest median particle size, apart from that obtained with BDPHW88K90H, was with BDPHW88K90L, at 4.58 μm, which was significantly larger (p > 0.05) than any of the other formulations. Thus, it appears that the molecular weight of PVP has a significant effect on the dimensions of the spray-dried particles. The smallest particles were found with the BDP, alone, at 3.13 μm.

Table 7 Particle size and manufacture yield of the novel BDP formulations (n=3, mean ± standard deviation)

Sample % Yield Dv, 0.1 Dv, 0.5 Dv, 0.9 % BDP contei (μm) (μm) (μm)

BDP n a 1.83 ± 0.21 3.13 ± 0.15 4.90 ± 0.36 100.00

BDPLW70K15 19.09 2.52 ± 0.03 4.02 ± 0.13 6.13 ± 0.21 64.68 ± 3.79

BDPLW80K15 14.62 2.40 ± 0.02 3.37 ± 0.02 4.83 ± 0.01 74.29 ± 3.80

BDPLW88K15 20.60 2.50 ± 0.08 3.89 ± 0.29 5.97 ± 0.56 64.39 ± 3.20

BDPMW88K15 17.44 2.46 ±0.10 3.74 ± 0.31 5.43 ±0.70 67.03 ± 7.95

BDPHW88K15 13.76 2.47 ±0.06 3.87 ± 0.16 5.33 ± 0.27 62.51 ± 4.51

BDPHW88K90H 25.90 3.23 ± 0.02 8.14 ±0.27 18.34 ± 0.37 38.21 ± 3.45

BDPHW88K90L 28.52 2.64 ± 0.05 4.58 ±0.17 9.50 ± 0.90 62.41 ± 3.56

BDPLW98K15 20.18 1.98 ± 0.30 3.22 ± 0.24 4.93 ±0.39 66.41 ± 6.28

While the use of the high molecular weight PVP increased the amount of product produced, it did not correlate with the content of therapeutic substance. The efficiency of the spray-drying process was not directly influenced by the particle size of the product, but it was dependent on the type and number of excipients added to the formulation. This was also true of the BDP content in the formulations. Adding higher molecular weight excipients reduced the BDP content of the inhalable powders, whilst the less excipients that were added, the higher the BDP content remained.

In vitro deposition

Adjusting the % hydrolysis of the PVA, or changing the molecular weight of either the PVA or PVP, appeared not to improve on the stage 2 deposition using HFA 134a. The results are shown in Figure 4, which shows the in vitro deposition profile of five BDP HFA 134a MDIs. Measurements were made using a twin-stage impinger. with formulation contents being detailed in Table 6 (n=3, error bars represent standard deviation). Only BDPLW98K15 displayed a significantly larger stage 2 deposition compared to BDP alone in HFA 134a (p < 0.05, ANOVA). Changing the polymer type increased the variability of the MDI deposition profile, for example the BDPMW88K15 displayed a standard deviation of just over 10% when formulated in HFA 134a.

Combining BDP with PVA 80% hydrolysed in HFA 227 had no significant effect on the stage 2 deposition in the twin-stage impinger (see Figure 5) compared to BDP suspended alone in HFA 227. Figure 5 shows the in vitro deposition profile of five BDP HFA 227 MDIs. Measurements were made using a twin-stage impinger, with formulation contents being detailed in Table 6 (n=3, error bars represent standard deviation). However, as in HFA 134a, adding a combination of PVA and PVP to the BDP microparticles in HFA 227 produced a stage 2 deposition of > 50%. The addition of HA had little effect on the deposition profile of the formulation but, addition of trehalose again reduced the stage 2 deposition to less than 50%.

Varying the percentage hydrolysis of the PVA in PVA/PVP stabilised BDP suspensions in HFA 227 did influence the deposition profiles of the MDIs. PVA 70% hydrolysed produced the lowest stage 2 deposition of 28%, which was significantly lower (p < 0.05, ANOVA) than both PVA 80 and 88%, both of which displayed depositions of approximately 52%. The greatest stage 2 deposition of any of the formulations was produced by BDP stabilised with a mixture of PVA 98% and PVPK15 with a stage 2 deposition of 54.49%.

Changing the molecular weight of PVA had no significant effect (p > 0.05, ANOVA) on the stage 2 deposition of the BDP in HFA 227. However, increasing the molecular weight of PVP from 10,000 to 360,000 reduced the stage 2 deposition from 52.25% to 18.53%.

Figure 6 shows the in vitro deposition profile of six BDP HFA 227 MDIs. Measurements were made using a twin-stage impinger. Formulation contents are detailed in Table 6 (n=3, error bars represent standard deviation). Dissolution

All the formulations seemed to show a two phase release, an initial burst when a proportion of BDP instantly dissolves in the simulated lung fluid and a secondary phase which represented a more steady state release. The BDP taken from the CFC based Becotide 50^® formulation produced a significantly larger burst phase compared to the raw BDP, but the subsequent dissolution profiles were almost identical, as shown in Figure 7, which illustrates the dissolution of three forms of BDP microparticles in simulated lung fluid over a 24 hour time period (n=3 error bars represent ± the standard deviation).

When the BDP was combined with any of the excipients and spray-dried, the burst phase was significantly larger, compared to BDP alone or BDP from the Becotide 50^® formulation. Complete dissolution occurred during the time frame of the experiment for the majority of formulations that contained one or more of the polymers. Figure 7 compares the raw BDP release profile with that of Becotide 50^® and BDP spray-dried with PVA 88% low molecular weight and PVP K15. At each of the time points, the 88% hydrolysed PVA formulated BDP exhibited a greater release compared to the raw BDP.

Figure 8 shows the dissolution of three BDP microparticles containing three grades of PVA varying in molecular weight in simulated lung fluid (n=3 error bars represent ± the standard deviation), and demonstrates that, increasing the molecular weight of PVA from 13,000-23,000 to 31,000-50,000, i.e. low to medium molecular weight, had little effect on the dissolution of the BDP from the microparticles. There was no significant difference (p > 0.05, ANOVA) between the low and medium molecular weight PVA microparticles at any time points taken during the dissolution run. However, increasing the molecular weight further, to 124,000 - 180,000, reduced the quantity of BDP released during the burst phase, and gave a more consistent release over the rest of the dissolution time frame. In contrast to varying the molecular weight of PVA, changing the molecular weight of PVP from 10,000 (K15) to 360,000 (K90) had little influence on either the drug's burst or steady-state release profile (detailed in Table 8). Changing the % hydrolysis of the PVA in the BDP microp article formulation did influence the release rate. Although PVA 70% hydrolysed had one of the lowest percentage release at T₁₀ at 66.6%, it was one of the first formulations to release 100% of its BDP dose at 180 min (see Figure 9). The dissolution profiles of the microparticles containing PVA 80% and 88% was very similar. Their initial burst release was not significantly different (p > 0.05, ANOVA), although PVA 88% did release 100% of the BDP before PVA 80% hydrolysed. PVA 98% hydrolysed had the lowest initial release of all the formulations and did not manage to release 100% of its BDP load before the experiment was terminated. Figure 9 shows the dissolution of four BDP microparticles, containing three grades of PVA varying in percentage hydrolysis, in simulated lung fluid (n=3 error bars represent ± the standard deviation).

The data from the dissolution profiles was linearised using the Higuchi diffusion model (Higuchi, 1967). However, as there appeared to be two phases in the release of the BDP from the microparticles, the burst phase was omitted, and only the state-state release phase was subjected to the Huguchi modelling. A typical example of the linearisation of the dissolution data can be seen in Figure 10, which shows the dissolution of four BDP microparticles, containing three grades of PVA varying in percentage hydrolysis, in simulated lung fluid (n=3 error bars represent ± the standard deviation). Whilst the Higuchi model was initially developed to model matrix drug release by using this model, it is not suggested that the release from the microparticulates is based on matrix release. The Higuchi plot was simply used to compare the steady-state release phase of the microparticles across the formulations.

Table 8 Release rate of the inhalable BDP microspheres using simulated lung fluid as the dissolution medium

Formulation Tι₀ Release Rate BDPLW70K15 66.6% 0.0065 BDPLW80K15 77.4% 0.0078 BDPLW88K15 80.9% 0.0091 BDPMW88K15 82.6% 0.0070 BDPHW88K15 65.9% 0.0097 BDPHW88K90 70.0% 0.0126 BDPLW98K15 55.4% 0.0049 BDP 23.6% 0.0167 Becotide 50^® 36.7% 0.0108

The steady-state release was more rapid in the raw BDP and Becotide 50 formulations, compared to the BDP/excipient combinations. However, this must be offset by the lower burst rate of these two formulations (see Table 8, above). There was no apparent link between the steady-state release rate and either the percentage hydrolysis of the PVA or its molecular weight. Increasing the molecular weight did increase the steady state release but, again, this was probably due to a lower burst rate. DISCUSSION

Although in previous work trehalose has been shown to stabilise the physical characteristics of therapeutics in microparticulates, these studies have been predominately focused on spray-drying compounds from a solution. Spray-drying a suspension does not allow the production of a homogenous dispersion of the drug and excipients within the microparticulate, it simply facilitates the accumulation of excipients at the surface of the suspended material, in this case BDP. The fact that trehalose did not aid the suspension of the raw drug in the MDI formulation, or the solid state characteristics of the BDP, implies that it may not have interacted with the BDP, resulting in a physical mix of the components.

Whilst PVA could suspend the BDP in an aqueous solution alone, it did not influence the stability of the BDP microparticles within the HFA systems. The twin- stage deposition data was the primary indicator of the suspension stability in this study as, with the exception of one formulation, the particles entering the HFA environment had a particle size < 10 μm and, therefore, if the suspension was physically stable it should emit a high fraction of its dose to stage 2 of the twin-stage impinger. Wright (1994) found that PVA/PVP co-polymer increased apolar suspension stability, marginally, when combined with a therapeutic in a physical mix, but in the present study the use of spray-drying to combine these two polymers with BDP had a dramatic effect, increasing the stage 2 deposition of the twin-stage impinger by over 100%, compared to BDP formulated alone in HFA propellant.

The grades of PVA are known to vary considerably in terms of physiochemical properties. In his comprehensive review of PVA, Pritchard (1970) showed that many of the physiochemical properties were determined both by molecular weight and the percentage hydrolysis of the polymer. The thermal properties, solid-state characteristics and solubility in a range of solvents were all shown to be dependent on the % hydrolysis and molecular weight of PVA. Although varying the grade of PVA had very little effect on the particle size of the manufactured product, it did have significant influence on the physical stability of the MDI suspension and dissolution profile. Increasing the molecular weight of PVP had a significant effect on the manufacture method and the final HFA MDI formulation. In the present study, including high levels of PVP in the initial spray-drying suspension increased the final particle size dramatically. Although, by reduction of the solid content in the suspension, this effect could be minimised, aggregation of the particles in the HFA suspension was a problem for the PVP K90 in both HFA propellants. The PVP component was crucial in the suspension stabilisation of the BDP microparticles as, without it, the formulations did not show any evidence of physical stabilisation in the HFA, as illustrated by the change in twin-stage impinger deposition (see Figure 3).

BDP is a highly hydrophobic steroid, with a log P of 4.27, that exhibits limited solubility in aqueous systems. For poorly water soluble drags, such as BDP, absorption in vivo is dependent on the rate of dissolution in the surrounding medium. As BDP is targeted in the conducting airways of the lung, it must dissolve in an aqueous environment, and be absorbed, before it is removed by the mucocilliary escalator i.e. within approximately 1-2 hours (Davies and Feddah, 2003).

Although dissolution testing is an official test defined in the British Pharmacopoeia (BP) (2002) for solid and semi-solid dosage forms, there is not a specific test defined for the in vitro simulation of dissolution in the lung. The dissolution apparatus defined in the BP was initially designed for oral dosage forms. However, not only is the composition of lung fluid radically different to gastric fluid, but the quantity of fluid that lines the lung epithelium is much smaller (Patton, 1997). Therefore, in this study, a dissolution medium was chosen that had previously been shown to model the electrolyte content of lung fluid, in vivo, and this was combined with a simple surface active agent to simulate the lung surfactant (Gambel, 1967). Non- sink conditions (defined as incorporating the therapeutic within the dissolution apparatus at > 10% but less than its maximum solubility) were employed, as it was considered that this would mimic conditions in the lung epithelia.

Compared to the BDP, either from the commercial MDI or the raw drug, the dissolution of all the microparticulate formulations of the invention was more rapid. The raw BDP and Becotide 50^® released approximately 40% of their total dose after 2 hours, compared to between 60-90% when additional excipients were added to the formulations. Boman et al. (2001), describes four potential mechanisms for the enhancement of dissolution, using water soluble polymers, eutectic formation, increased surface area due to precipitation of the carrier, solid solution formation, increased wettability due to intimate contact with the hydrophobic carrier, or a decrease in crystallinity. Of the explanations provided by Boman et al. (2001), the increased wettability of the steroid and a decrease in the crystallinity seem the most favourable reasons for the changes observed in dissolution profiles of the microparticulates. Although the Higuchi model was used to linearise the data in the dissolution studies, the dissolution mechanism was not investigated. It was not thought crucial to investigate the precise mechanism of release as, regardless of whether the polymer was controlling the release rate of the BDP or not, the speed of dissolution in non-sink conditions was being improved.

References

ISO 13320 - Particle size analysis - Lazer diffraction methods part 1 : General Principles. 1999. ISO Standards Authority. 2002, British Pharmacopeia, London, UK, Stationary Office.

Bagchi, P., 1973, Theory of stabilization of spherical colloidal particles by nonionic polymers: Journal of Colloid and Interface Science 47 pp. 86-99.

Barnes, A.R., S. Nash, 1996, Comparison of aerodynamic particle size from 250[mu]g per dose beclomethasone dipropionate metered-dose inhalers: International Journal of Pharmaceutics 139 pp. 255-258.

Blondino, F.E., P.R. Byron, 1998, Surfactant dissolution and water solubilization in chlorine- free liquified gas propellants: Drug Development and Industrial Pharmacy 24 pp. 935-945.

Boury, F., T. Ivanova, I. Panaiotov, J.E. Proust, A. Bois, J. Richou, 1995, Dynamic Properties of Poly(Dl-Lactide) and Polyvinyl- Alcohol Monolayers at the Air- Water and Dichloromethane Water Interfaces: Journal of Colloid and Interface Science 169 pp. 380-392.

Brambilla, G., D. Ganderton, R. Garzia, D. Lewis, B. Meakin, P. Ventura, 1999, Modulation of aerosol clouds produced by pressurised inhalation aerosols: International Journal of Pharmaceutics 186 pp. 53-61. Broman, E., C. Khoo, L.S. Taylor, 2001, A comparison of alternative polymer excipients and processing methods for making solid dispersions of a poorly water soluble drug: International Journal of Pharmaceutics 222 pp. 139-151.

Cardona, S., C. Schebor, M.P. Buera, M. Karel, J. Chirife, 1997, Thermal stability of invertase in reduced-moisture amorphous matrices in relation to glassy state and trehalose crystallization: Journal of Food Science 62 pp. 105-112.

Cassu, S.N., M.I. Felisberti, 1997, Poly(vinyl alcohol) and poly(vinyl pyrrolidone) blends: Miscibility, microheterogeneity and free volume change: Polymer 38 pp. 3907- 3911.

Chibowski, S., M. Paszkiewicz, M. Krupa, 2000, Investigation of the influence of the polyvinyl alcohol adsorption on the electrical properties of A12O3-solution interface, thickness of the adsorption layers of PVA: Powder Technology 107 pp. 251-255.

Croot, R.A., A.R. Goodall, S.D. Lubetkin, 1995, Solution and Adsorption Properties of Vinyl Alcohol/Vinyl Acetate Copolymers: Colloids and Surfaces A-Physicochemical and Engineering Aspects 98 pp. 137-145.

Davidson, I.G., E.J. Langner, S.N. Plowman, J.A. Blair, 2003, Release mechanism of insulin encapsulated in trehalose ester derivative microparticles delivered via inhalation: International Journal of Pharmaceutics 254 pp. 211-222.

Davies, Ν.M., M.R. Feddah, 2003, A novel method for assessing dissolution of aerosol inhaler products: International Journal of Pharmaceutics 255 pp. 175-187.

Dawson, L. The stabilisation of micronised drug dispersions in a hydrofluoroalkane propellant system. 1997.

Eirich, F.E., 1977, The conformational states of macromolecules adsorbed at solid- liquid interfaces: Journal of Colloid and Interface Science 58 pp. 423-436.

Gambel, JL, 1967, Chemical Anatomy, Phsiology, and Pathology of Extracellular Fluid, Boston, Harvard University Press.

Ganderton, D., D. Lewis, R. Davies, B. Meakin, G. Brambilla, T. Church, 2002, Modulite (R): a means of designing the aerosols generated by pressurized metered dose inhalers: Respiratory Medicine 96 pp. S3-S8.

Guiavarc'h, Y., D. Sila, T. Duvetter, A. Van Loey, M. Hendrickx, 2003, Influence of sugars and polyols on the thermal stability of purified tomato and cucumber pectinmethylesterases: a basis for TTI development: Enzyme and Microbial Technology 33 pp. 555.

Hiestand.E.N, 1964, Theory of coarse suspension formulation: Journal of Pharmaceutical Sciences 53 pp. 1-18.

Higuchi, W., 1967, Diffusion Models Useful in Biopharmaceutics: Journal of Pharmaceutical Sciences 56 pp. 315-324. Hochhaus, G., S. Suarez, R. J. Gonzalez-Rothi, H. Schreier. Pulmonary targeting of inhaled glucocorticoids: How is it influenced by formulation. VI, pp 45-52. 1998.

Kitahara, A., K. Shuichi, H. Yamada, 1967, The effect of water on electrokinetic potential and stability of suspensions in nonpolar media: Journal of Colloid and Interface Science 25 pp. 490-495.

Leach, C.L., P.J. Davidson, B.E. Hasselquist, R.J. Boudreau, 2002, Lung deposition of hydrofluoroalkane-134a beclomethasone is greater than that of chlorofluorocarbon fluticasone and chlorofluorocarbon beclomethasone - A cross-over study in healthy volunteers: Chest 122 pp. 510-516.

LeBelle, M., R.K. Pike, S.J. Graham, E.D. Ormsby, H.A. Bogard, 1996, Metered-dose inhalers .1. Drug content and particle size distribution of beclomethasone dipropionate: Journal of Pharmaceutical and Biomedical Analysis 14 pp. 793-800.

Lewis, D., S. Johnson, B. J. Meakin, D. Ganderton, G. Brambilla, R. Garzia, P. Ventura. Effects of Actuator Orifice diameter on beclomethasone dipropionate delivery from a pMDI HFA solution formulation. VI, pp 363-364. 1998.

Liao, Y.H., M.B. Brown, T. Nazir, A. Quader, G.P. Martin, 2002, Effects of sucrose and trehalose on the preservation of the native structure of spray-dried lysozyme: Pharmaceutical Research 19 pp. 1847-1853.

Librizzi, F., E. Vitrano, L. Cordone, 1999, Dehydration and crystallization of trehalose and sucrose glasses containing carbonmonoxy-myoglobin: Biophysical Journal 76 pp. 2727-2734.

McDonald, K.J., G.P. Martin, 2000, Transition to CFC-free metered dose inhalers - into the new millennium: International Journal of Pharmaceutics 201 pp. 89-107.

Nguyen, D.T., 1996, Determination of equilibrium surface energy of adsorbed polyvinyl alcohol layers in water at 25 degrees C: Colloids and Surfaces A-Physicochemical and Engineering Aspects 116 pp. 145-160.

Noakes, T., 2002, Medical aerosol propellants: Journal of Fluorine Chemistry 118 pp. 35-45.

Patton, J.S., 1997, Deep-lung delivery of therapeutic proteins: Chemical Technology Review 27 pp. 34-38.

Pritchard.J.G, 1970, Poly(vinyl alcohol): Basic properties and uses., Mcdonald G.B. Science publishers.

Ross, D.L., B.J. Gabrio, 1999, Advances in metered dose inhaler technology with the devlopment of a chlorofluorocarbon free drug delivery system: Journal of Aerosol Medicine-Deposition Clearance and Effects in the Lung 12 pp. 151-160.

Schultz, R. K., D. W. Schultz. Beclomathsone solution aerosol formulations. Minnesota Minning and Manufacturing Company. 455,872[5,776,432]. 1995. Smyth, H.D.C., 2003, The influence of formulation variables on the performance of alternative propellant-driven metered dose inhalers: Advanced Drug Delivery Reviews 55 pp. 807-828.

Sonie, W.H., F.E. Blondino, P.R. Byron, 1992, Chemical Stabiliy pressurised inhalers formulated as solutions: Journal of Biological and Pharmaceutical Science 3 pp. 41-47.

Srinivasa, P.C., M.N. Ramesh, K.R. Kumar, R.N. Tharanathan, 2003, Properties and sorption studies of chitosan-polyvinyl alcohol blend films: Carbohydrate Polymers 53 pp. 431-438.

Steckel, H., B.W. Muller, 1998, Metered-dose inhaler formulations with beclomethasone-17,21- dipropionate using the ozone friendly propellant R 134a: European Journal of Pharmaceutics and Biopharmaceutics 46 pp. 77-83.

Stein, S.W., 1999, Size distribution measurements of metered dose inhalers using Andersen Mark II cascade impactors: International Journal of Pharmaceutics 186 pp. 43-52. van Mill, P.J.J.M., D.J.A. Crommelin, P.H. Wiersema, 1984, Stability of course suspensions in nonpolar media: Effect of gravity on the interaction between particles: Journal of Colloid and Interface Science 98 pp. 61-71.

Vervaet, C, P.R. Byron, 1999, Drug-surfactant-propellant interactions in HFA- formulations: International Journal of Pharmaceutics 186 pp. 13-30.

Williams, R.O., C.J. Hu, 2001, Influence of water on the solubility of two steroid drags in hydrofluoroalkane (HFA) propellants: Drag Development and Industrial Pharmacy 27 pp. 71-79.

Wright.P, 1994, Polymers for the colloidal stabilisation of drugs in HFAs: Respiratory Drag Delivery IV pp. 243-247.

Yong-Hong Liao. Studies on the Stabilisation and Formulation of Proteins for Airway Delivery. 2002. Kings College.

Zerfa, M., B.W. Brooks, 1998, Experimental investigation of PVA adsorption at the vinyl chloride water interface in monomer suspensions: Colloids and Surfaces A- Physicochemical and Engineering Aspects 132 pp. 267-273.

Zhang, J., G. Zografi, 2001, Water vapor absorption into amorphous sucrose-poly( vinyl pyrrolidone) and trehalose-poly( vinyl pyrrolidone) mixtures: Journal of Pharmaceutical Sciences 90 pp. 1375-1385.

Claims

CLAIMS:

1. A formulation of a therapeutic substance suitable for delivery to a patient by a metered dose inhalation device, the formulation comprising a substantially dry powder preparation of the substance with a stabilising amount of polyvinylpyrrolidone and a polyhydroxylated polyalkene, in combination with one or more propellants therefor.

2. A formulation according to claim 1, wherein the PVP has a K value of no greater than 50.

3. A formulation according to claim 2, wherein the PVP has a K value of no greater than 30.

4. A formulation according to claim 2, wherein the PVP has a K value of about 15.

5. A formulation according to any preceding claim, wherein the therapeutic substance is a hydrophobic drug.

6. A formulation according to claim 5, wherein the hydrophobic drug is selected from the group consisting of: corticosteroids; anticholinergic drags; leukotrienes; cannabiods; antiemetcs; and combinations thereof.

7. A formulation according to claim 6, wherein the corticosteroid is selected from the group consisting of: beclomethasone dipropionate, budesonide, flunisolide, triamcinolone acetonide, and fluticasone dipropionate, and mixtures thereof

8. A formulation according to any preceding claim, wherein the propellant is alkane based.

9. A formulation according to claim 8, wherein the propellant is at least one haloalkane.

10. A formulation according to claim 9, wherein the propellant is selected from HFA-134a and HFA-227.

11. A formulation according to any preceding claim, wherein at least one polyhydroxylated polyalkene has the general structure

-(CH₂-CHOR)_n-

where R is the same or different from one monomeric unit to the next, and is hydrogen, lower alkyl, lower alkenyl, lower alkanoyl, lower alenoyl or is a bridging group between adjacent monomers.

12. A formulation according to claim 11 , wherein, when R is not hydrogen, the number of carbon atoms, excluding any -CO- group, is between 1 and 6, inclusive.

13. A formulation according to claim 11 or 12, wherein the polyhydroxylated polyalkene is selected from polyvinylalcohol, polyvinylacetate, polyvinyl alcohol-co- vinyl acetate, poly(vinyl butyral), poly(vinyl alcohol-co-ethylene), and mixtures thereof.

14. A formulation according to claim 13, wherein the polyhydroxylated polyalkene is PVA.

15. A formulation according to claim 13 or 14, wherein the PVA is a hydrolysate of PVAc, the level of hydrolysis being above 70%.

16. A formulation according to claim 15, wherein the level of hydrolysis is above 80%.

17. A formulation according to claim 15 or 16, wherein the level of hydrolysis is between 80 and 90% and the majority of the propellant is HFA 134a.

18. A formulation according to any of claims 12 to 15, wherein the PVA has a molecular weight of between about 9kDa and 50kDa.

19. A formulation according to any preceding claim, wherein the polyhydroxylated polyalkenes are present in an amount of from about 1% to about 200% by weight of the therapeutic substance.

20. A formulation according to claim 19, wherein the polyhydroxylated polyalkene is present between about 1% and about 60%) by weight of the substance.

21. A method for the preparation of a formulation as defined in any preceding claim, comprising blending the hydrophobic drag with PVP and polyhydroxylated polyalkene substances in an aqueous vehicle, drying the resulting blend to a powder, and then formulating with propellant.

22. A method according to claim 21, wherein the aqueous vehicle is selected from saline, a suitable buffer, and deionised water.

23. A method according to claim 21 or 22, which comprises spray-drying the blend.

24. A powdered formulation of a therapeutic agent, PVP and a polyhydroxylated polyalkene, as defined in any of claims 1 to 20, which is suitable for incorporation with a haloalkane propellant for dispensing from a metered dose inhaler.

25. A powdered formulation according to claim 24, wherein the powder particles have an aerodynamic diameter of between about 0.1 μm and 100 μm.

26. A powdered formulation according to claim 25, wherein the powder particles have an aerodynamic diameter of between about 0.1 μm and 20 μm.

27. A powdered formulation according to claim 24, wherein the powder particles have an aerodynamic diameter of between about 1 μm and 50 μm.

28. A metered dose inhalation device provided with a reservoir comprising a formulation according to any of claims 1 to 20.