WO1999030688A1 - Methods of lyophilizing solutions - Google Patents

Abstract

A method of lyophilizing a solution comprising the steps of freezing the solution to a temperature at or below the lower of its eutectic temperature or its glass transition temperature and, in a first drying stage, removing at least a portion of the solvent by sublimation, characterized in that the solution contains an accelerant excipient to enhance the rate of solvent sublimation.

Description

WO 99/30688 - - 1 - PCT/GB98/03747

METHODS OF LYOPHILIZING SOLUTIONS

Description

The present invention relates to methods of lyophilizing or freeze-drying solutions and to products obtainable by such methods. In particular, the present invention relates to methods of lyophilizing solutions by including substances that enhance the rate of solvent sublimation and/or that obviate the need for subsequent drying steps.

In a number of different aspeαs of the present invention, there are provided those methods and products obtained or obtainable by such methods as defined and described in the attached claims.

Preferred embodiments of the invention in any of its various aspects are as defined in the sub-claims.

OVERVIEW

Freeze-drying has established itself as the standard method for the stabilisation of many drug substances in the solid state. Compared to other drying processes, the technology is capital-, labour- and energy-intensive. Its optimisation depends on a complex interplay of several formulation and processing parameters; these have been highlighted in several recent product and process development studies (1, 2). Particular attention is currently being paid to the mechanisms and control of deleterious physical and/ or chemical changes that might occur during freeze concentration of an initially dilute solution, during the ice sublimation phase, and thereafter. In the extreme, such changes render a dried product slow to rehydrate or even lead to partial or total inactivation, denaturation or modification. Some investigations have concentrated on the use of suitable additives, specifically to enable process cycle times to be reduced, with resulting economies.

It is by now well understood that for lyophilisation a solution needs to be cooled to the lower of its characteristic eutectic temperature T_c or glass temperature T_g', below which further cooling does not affect phase composition (3). As a first working approximation most formulations do not show eutectic phase separation and in practice the temperature of the amorphous freeze concentrate must not be allowed to rise above T ' during the primary drying stage, if structural collapse and/or chemical deterioration are to be avoided. The product developer would thus choose excipients that possess high T ' values, enabling the sublimation process to be performed at as high a temperature, and thus as short a time, as practicable.

One way of enhancing the sublimation rate of ice is by the modification of its crystal habit. Usually the most common form, plates of dendritic ice, is obtained during slow cooling. Ice crystallisation can be substantially modified by low concentrations of certain additives, to give a more open ice structure which would facilitate sublimation (4,5).

In this context, special benefits have been claimed for tertiary butanol (TBA) as an additive to accelerate the primary drying (ice sublimation) stage of the lyophilisation process (6,7). In ternary mixtures of water-sucrose-TBA, no special interactions between sucrose and TBA have been noted, implying that TBA does not affect the T ' value (-32 °C) usually ascribed to aqueous solutions of sucrose (11). This suggests that TBA is excluded from the amorphous freeze concentrate and crystallises completely from the frozen mixture. Because of its high vapour pressure, it is expected that it would to be sublimed, along with the ice, and without leaving a residue in the dried, amorphous sucrose cake. The specific effeα of TBA, on the system, is said to be related solely to its ability to modify the ice crystal habit. This leads to the growth of needle-shaped crystals, producing a significant increase in the specific surface area available for sublimation, and a concomitant decrease in the resistance of the dried cake to mass transfer. This enables the drying temperature to be raised and the sublimation cycle to be substantially shortened.

Contrary evidence (8), on the other hand, indicates that TBA does indeed affect T ' of freeze-concentrated sucrose solutions, without measurably affeαing the water content of the freeze concentrate, thus implying that at least some TBA is retained in the concentrated, amorphous phase in the form of a solid solution. In this respeα residual TBA may be considered disadvantageous for many pharmaceutical, diagnostic or food products; for example parenteral produαs intended for human injeαion. In this study (8) the effect of volatile excipients, thought likely to undergo euteαic crystallisation, was investigated as a . WO 99/30688 - - 3 - PCT/GB98/03747

route for increasing the rate of ice sublimation. Although enhanced sublimation rates were found, there was some evidence that the mechanism was not wholly via the euteαic crystallisation of volatile species.

Many freeze dried produαs contain sugars added either as bulking agents or as stabilisers. There are only very limited experimental data available concerning the solid/liquid phase behaviour of aqueous sugar solutions containing third components that might, under some circumstances, display euteαic phase separation. Experimental difficulties encountered in the study of phase changes with long relaxation times (relative to the experimental time scale) are well illustrated by the recently reported investigations of aqueous sucrose solutions with glycine (9,10) or sodium chloride (11) as third components. Depending on cooling/heating rates and solute composition ratios in the ternary mixtures, solute crystallisation may occur during cooling, during rewarming, or not at all on the experimental time scale. More complex mixtures would be expeαed to have similar properties.

might be expeαed, annealing treatments also affeα the observed phase behaviour and the phase ratio of crystalline.amorprious material. It is emphasised, therefore, that the experiments reported below cannot be regarded as comprehensive. They can only provide indications of processes that take place on the time scales probed by the experiments.

Despite the paucity of data and limited understanding of mechanisms, the modification of struαure to allow the more rapid removal of water (as vapour) from a crystal mass or glassy residue, may equally facilitate the re-entry of water into a freeze-dried cake and the dissolution of the produα. Acceleration of freeze-drying, may therefore also facilitate the produαion of a rapidly dissolving dosage form.

STATE OF THE ART

Current best praαice (3) teaches that the freeze-drying process consists of three distinα stages:

1) Concentration of the produα by freezing (removal of liquid water).

2) Sublimation of the ice so formed.

3) Removal from the produα of residual unfrozen water by diffusion/desorption/evaporation. Depending on its chemical composition, the dried produα may be

a) wholly or partially crystalline or b) amorphous

The significant properties of the produα that determine the correα setting of the process parameters are summarised below.

1) If one or more of the components can become subjeα to precipitation during freezing, the euteαic temperature T_c must be established, i.e. the lowest temperature at which any of the precipitated components (including ice) melt and re-dissolve back into the produα.

2) If the freeze-concentrated produα remains homogenous, freezing slows down and eventually ceases, in real time. The temperature at which the degree of freeze-concentration reaches its maximum value is referred to as the glass temperature T ' of the freeze-concentrated produα, which then still contains

W_g'g unfrozen water/g solids. At the glass transition temperature the physical properties change from those of a viscous liquid to those of a brittle solid. Water aαs as a plasticiser, depressing the glass transition temperature of the produα.

A rigorous differentiation is made between two apparent thermal transitions, called T ' and T ". The lower temperature T_g" is associated with a marked viscosity change (a glass is defined as a solution with a viscosity of 10¹⁴ Pa.s). Immediately above this temperature, however, the viscosity is still sufficiently high to severely inhibit diffusion and viscous flow. T ', the higher temperature transition, is the point at which the viscosity is reduced for mechanical flow (collapse) to occur on this time scale of observation. For most praαical purposes a temperature just below T ' is sufficient for effeαive freeze-drying. For extended cycles, lowering the temperature to just below T_g" (which itself would extend drying time) would ensure maximum storage stability. 3) For long-term shelf stability at ambient temperature, the dried produα must have a glass temperature in excess of the storage temperature. T_g of the dried produα depends on formulation details and the residual moisture content (W . Knowledge of these quantities is particularly important for the design of accelerated stability tests and for prediαions of produα shelf life.

4) The process efficiency, i.e. ice sublimation and secondary drying rates and the quality of the dried produα are determined by several faαors, among them: a) Volume of water to be removed. b) Total solids content and chemical composition of the produα. c) Liquid fill depth (i.e. container geometry). d) Temperature difference between produα and condenser. e) Chamber pressure f) Efficiency of the freeze dryer.

METHODS

Materials

In general analytical grade (or equivalent) laboratory chemicals were used. Specific sources are given below

Ammonium acetate Sigma A-8920 lot 54H11271

Ammonium bicarbonate Sigma A-6141 lot 15H0338

Eletriptan.HBr Pfizer lot R206

Formic acid, ammonium salt Sigma F-2004 lot 26H0044 α-Laαose monohydrate (Milk sugar) Sigma L-3625 lot 22H0100

PVP Type K17 (DAB / Ph.Eur./ US ( BASF gift batch unknown

Sildenafil Pfizer lot R2

Sucrose Prolabo Normapur AR lot 92126 Differential scanning calorimetry

Differential scanning calorimetry (DSC) was used to monitor the behaviour of the preparations. All samples were sealed in stainless steel calorimeter pans and loaded into the Perkin-Elmer model DSC-7 calorimeter equilibrated at 30 °C. A computerised data logging/ handling system (Perkin-Elmer PETA software, 1995) was used to examine the power-time curves obtained. For T_g' measurements, samples were typically cooled at 5 °C/min to -65 °C before heating (again at 5 °C/min) to 30 °C. Dried samples were cooled at 10 °C/min to -25 °C/-30 °C before heating at 10 °C/min.

For slow cooling experiments, samples were cooled at 0.3 °C/minute and warmed at 5 °C/min. In some cases, to aid throughput, samples were cooled to -10 or -15 °C at5 °C/min prior to the 0.3 °/min slow cooling protocol. In all cases freezing (which must precede euteαic crystallisation) occurred during the slow cooling regime.

Moisture analysis

Samples for moisture analysis were transferred and sealed into tared vials under a nitrogen atmosphere and then weighed. Moisture analysis was performed by coulometric Karl

Fischer titration (using a Mitsubishi CA-05 Moisture Meter with Hydranal AG, anolyte, and Hydranal CG, catholyte solutions. Hydranal reagents are produced by Riedel de Haen AG, D-3016 Seelze 1, Germany). The following standard protocol was used:

1) Approximately 6 μl of water were added to 20 ml of f ormamide, which was then sealed in an amber glass vial.

2) Replicate 0.5 ml portions were injeαed into the instrument reaαion cell using a 1 ml

Hamilton Gastight Syringe #1001, until three consistent readings were obtained.

3) Two millilitres of formamide were then injeαed into each sealed sample vial and into two sealed, but empty, blank vials. The vials were then heated at 90 °C for 10 minutes to ensure dissolution. 4) Replicate 0.25 ml portions of each sample solution were injeαed into the reaαion cell (Hamilton Gastight Syringe #1001).

5) The moisture content of the sample is the difference between the moisture content of the sample solution and of the blank vials.

Freeze-drying

Freeze-drying cycles were developed using "best praαice" as outlined in reference (3). All cycles were optimised for freezing behaviour, solids content, fill depth and container geometry.

EXPERIMENTAL

1. Choice of model systems

2. Freezing behaviour of excipient / salt mixtures 2A Excipients 2.B Salt solutions 2.C Excipient / salt mixtures 2.D Summary of experimental results.

2.D.1 Excipients only 2.D.2 Salts only 2.D.3 Excipient salt mixtures 2.D.3.a Ammonium bicarbonate solutions

2.D.3.a.i Sucrose 2-D.3-a.ii PVP 2.D.3.a.iii Lactose 2.D.3.b Ammonium acetate solutions 2.D.3.b.i Sucrose

2.D.3.b.ii PVP 2.D.3.b.iii Lactose WO 99/30688 - . 8 - PCT/GB98/03747

2.D.3.C Ammonium formate solutions 2.D.3.c.i Sucrose

2.D.3.c.iii Lactose 5 3. Sublimation of samples

3 A Sucrose / Ambic Sublimation 3.B PVP / Ambic Formulation 3.C Lactose / Ambic Formulation 3.D Sucrose / AmAc Formulation w 3.E PVP / AmAc Formulation

3.F Lactose / AmAc Formulation

4. DSC Analysis of Sublimed Samples

5. Observations of samples removed from the dryer to measure sublimation ra

6. Sublimation of 0.1M salts containing formulations 15 6.A Freezing behaviour

6A.1 Sucrose 6A.2 PVP 6A.3 Lactose 6.B Sublimation of 0.1M Salt containing solutions 20 6.B.1 Overview

6.B.2 By category 6.B.2.I. Sucrose 6.B.2.2. PVP 6.B.2.3. Lactose 25 6.B.2.4. Ammonium bicarbonate

6.B.2.5. Ammonium acetate 6.B.2.6. Ammonium formate 6.B.2.7 Analysis of salt loss following primary and secondary drying.

7. Selection of formulations for further study

30 8. Freezing behaviour of Eletriptan and Sildenafil 8A Eletriptan 8.B Sildenafil 9. Freezing behaviour of formulated Eletriptan and Sildenafil 9.A Eletriptan / PVP formulations 9.A.1 Ammonium bicarbonate 9A.2 Ammonium formate 9.B.3 No salt; Eletriptan 5 mg/ml in 5% PVP 9.B Sildenafil / PVP formulations

9.B.1 Ammonium bicarbonate 9.B.2 Ammonium formate 9.B.3 No salt; Sildenafil 5 mg/ml in 5% PVP 9.C Summary of glass temperatures 10. Sublimation rates of formulated Eletriptan and Sildenafil 11. Preparative drying of Eletriptan and Sildenafil formulations 11A Eletriptan 11.B Sildenafil ll.C Characterisation of dried products ll.C.l Mechanical stability of samples subject to primary drying only

11.C.2 Physico-chemical characterisation ll.C.2.1 Eletriptan formulations ll.C.2.2 Sildenafil formulations ll.C.2.3 PVP solution 11.C.3 Dissolution studies

11.C.4 Handling properties

WO 99/30688 . ₁₀ - PCT/GB98/03747 .

EXPERIMENTAL

1. Choice of model systems Preliminary work investigating the effeαs of 0.1M salts (acetate, formate, bicarbonate) and TBA on ice sublimination rates from 8.5% laαose, PVP and sucrose is described.

Methods

The Perkin-Elmer DSC-2 instrument used in this work was fitted with Auto-scanning and subambient temperature accessories. Sample masses used were of the order of 15 mg. The DARES data colleαion and handling system was used for recording and processing DSC power-time curves. This system enables the normal sensitivity of the instrument to be increased by three orders of magnitude (Hatley et al., Thermochim. Aαa 156, 247- 257(1989)). Test solutions were cooled to 210-220K, and recording of DSC traces was started at 220-230K. The melting points of ice (273.2K) and indium (429.8K) were used for temperature calibration. Scanning rates for cooling and heating were 5 K/min, unless otherwise stated. A standard DSC-7 instrument was used for a limited number of experiments.

All transition temperatures were determined as the onsets of discontinuities in the heat flow curves, rather than the mid-points. The reasons are both theoretical and praαical: the exaα shape of the heat flow curve, and particularly its width, depend on the thermal history of the sample, e.g. cooling/heating rates and annealing details. The onset of a transition is the only point on the scan which is sensibly independent of the scanning details. It is more easily defined, and in praαice it can be related, albeit empirically, to the temperature of the centre of the transition, so that appropriate correαions can be applied (Yu et al., J. Chem. Soc. Faraday Trans. 91, 1511-1517(1995)).

The effeαs of ammonium salts with relatively high vapour pressures ("volatile salts") on several formulations was studied by measuring the initial rates of ice sublimation under WO 99/30688 - _ _{1 1} _ PCT/GB98/03747

freeze-drying conditions. 8.5% solutions of sucrose, Polyvinylpyrrolidone (PNP) and laαose were made up with additional 0.1 M ammonium acetate, formate or bicarbonate or 5% TBA. Drying was performed on 1 ml sample solutions in 5 ml vials, in a custom-built research lyophiliser which enables the temperature and chamber pressure to be sensitively and independently controlled. Care was taken to perform all operations according to "best praαice" (3). Solutions with and without the four additives, were made up in vials by weight and frozen. Their Tg' values were determined, as described above. Ice sublimation was performed at 4 deg below the lowest Tg' value found for each series of formulations. The chamber pressure was set at 0. 1 mb for all sublimation experiments. During the initial seven hours of the primary drying phase, samples were periodically withdrawn from the freeze- drier and the mass losses were determined.

The freeze-drying experiments were repeated with sucrose solutions containing the same four additives, but at a uniform sublimation temperature of 228K: well below the lowest Tg' value of any of the preparations used. Nials were cooled to, and maintained at 228K overnight before drying was commenced. After drying for 435 min, the samples were removed from the freeze-drier, refrozen, and their Tg' values redetermined.

RESULTS Ice sublimation

The ability of TBA to enhance the ice sublimation rate has been ascribed solely to its own high vapour pressure and its ability to modify the ice crystal habit (6,7). Comparative sublimation rates achieved by the formulation of 8.5% solutions of sucrose, PNP or laαose, each containing TBA or one of several other volatile excipients are shown in the Table 1 (overpage) as percent weight losses with time, and graphically in Figures la-c. The corresponding Tg' values of the freeze concentrates are shown in Table 2. Table 1A, Sublimation rates of frozen 8.5% excipient solutions of sucrose, PW and laαose, made up in 0.1 M ammonium acetate, bicarbonate and formate and in 5% TBA. Sublimation was carried out at temperatures 5 degrees below the respeαive Tg' values of the mixtures (see Table 2).

Sucrose

Time (min) acetate bicarbonate formate 5%TBA no additive

75 5.71 4.78 5.84 6.03 6.5

135 13.4 10.9 10.8 12.8 9.9

195 14.7 14.1 17.5 15.4 12.9

255 23.3 17.5 27.4 24.6 10.5

315 29.8 25.4 28.0 24.6 18.5

375 41.6 27.6 35.0 32.0 19.1

435 39.1 36.3 36.3 35.8 21.7

PVP

Time (min) acetate bicarbona .te formate 5% TBA no additive

50 2.8 3.0 2.6 5.0 3.2

110 6.7 6.6 6.2 10.4 6.1

170 10.2 11.8 10.3 16.2 9.3

230 11.1 18.3 12.6 23.8 12.9

290 13.0 21.8 13.6 26.3 15.6

350 18.2 22.6 14.9 30.0 18.1

410 27.4 34.3 27.8 39.3 21.8 Table 1A continued

Lactose

Time (min) acetate bicarbonate formate 5% TBA no additive

5 75 2.0 1.0 1.7 2.7 1.5

140 6.2 4.5 4.6 5.3 4.2

210 6.7 6.6 7.1 8.5 5.0

27.5 10.7 8.5 9.4 11.6 7.2

345 16.2 12.0 13.8 15.6 8.9

'0 410 28.3 14.4 14.0 21.7 13.8

47.5 20.8 15.2 18.7 26.3 15.4

TABLE IB

Sucrose data

15

Time min Acetate Bicarb Fomate TBA None

65 2.2 1.7 2.3 2.7 1.2

135 6.9 4.4 4.2 5.5 2.8

195 7.5 6.2 7.5 7.8 5

245 11.2 10.2 10.2 10.2 8.8

315 12.6 12.8 13 14.9 11.3

375 24 13.9 15.4 16.8 12

435 26.2 18 19.3 21.6 13.2

25 WO 99/30688 - . 14 . PCT/GB98/03747

Table IB continued PVP Data

Time min Acetate Bicarb Formate TBA None

50 2.8 3 2.6 5 3.2

110 6.7 6.6 6.2 10.4 6.1

170 11.1 11.8 10.3 16.2 9.3

230 10.2 18.3 12.6 23.8 12.9

290 13 21.8 13.6 26.3 15.6

350 18.2 22.6 14.9 30 18.1

410 27.4 34.3 27.8 39.3 21.8

Lactose data

Time min Acetate Bicarb Formate TBA None

75 2 1 1.7 2.7 1.5

140 6.2 4.5 4.6 5.3 4.2

210 6.7 6.6 7.1 8.5 5

275 10.7 8.5 9.4 11.6 7.2

345 16.2 12 13.8 15.6 8.9

410 18.3 15.2 14 21.7 13.8

475 20.8 14.4 18.7 26.3 15.4

The graphical representation is given in Figure IB WO 99/30688 - . ₁₅ . PCT/GB98/03747

Table 2. Glass transition temperatures (K) of maximally freeze concentrated solutions employed in the sublimation experiments summarised in Table 1 acetate bicarbonate formate TBA None sucrose 232.5 235.7 232.8 236.1 239.4

PVP 237.9 245.2 240.8 239.2 251.9 laαose 236.9 240.0 236.9 237.4 243.3

DISCUSSION All salt additives are seen to enhance the sublimation rate over that of the frozen solution which contains no additive. In the sucrose-based formulations the salts performed mostly better than TBA, but no significant difference between the sublimation rates (at least over the initial seven hours) could be established for PVP or laαose-based formulations.

Some vials, containing partly dried sucrose formulations with the various additives were removed from the freeze-drier after 435 mm and the contents transferred to DSC pans which were then sealed. They were refrozen and scanned, as described above, to establish Tg' values. In no case could a shift in Tg' be deteαed, indicating that the composition of the solid solution phase had not changed during the period of sublimation. Although all the additives used are classified as "volatile", it appears that during the initial stages of drying, covered by our experiments, only ice was removed from the preparation. This interpretation is further confirmed by the absence of any deteαable phase separation (crystallisation) with the preparations containing salts, indicating that under the experimental conditions, the salts are part of the vitreous phase from which evaporation would be expeαed to be extremely slow.

The results highlight some novel and unexpeαed aspeαs of the solution and phase behaviour of binary and ternary mixtures of the systems water-sucrose-X, where X is TBA or a salt. Some generalisations can be made about the role of TBA and "volatile" salts in affeαing the rate of ice sublimation from glassy freeze-drying formulations.

The weight loss results in Table 1 demonstrate that addition of the volatile solutes can considerably enhance the rate of ice sublimation from all three excipient solutions, but no clear patterns could be observed which would allow a ranking order to be established. The prior cooling rate probably influences the subsequent effeα of the volatile additive on the sublimation. The results for laαose are particularly erratic, suggesting that partial, but uncontrollable crystallisation of the sugar may be responsible for the scatter. It is clear, however, that with salts in combination with sucrose, at least, the mass loss during the initial few hours can be ascribed exclusively to ice sublimation, rather then removal of the additive.

The mechanism of sublimation rate enhancement by volatile salts is not clear. It bears no resemblance to the similar effeα by TBA (6, 7) or (12), where the former is ascribed to a modification of the ice crystal habit and the latter to the mechanical support provided by the in situ crystallisation of sodium chloride from the amorphous freeze concentrate during the removal of ice, thus enabling the drying process to be carried out at temperatures somewhat in excess of Tg'. In addition it avoids the problem of residual TBA in dried pharmaceutical preparations. The results demonstrate that several volatile buffer salts perform at least as well as, or in some cases better than TBA in enhancing drying rates. They, or even salts known not to be volatile, such as sodium chloride, or crystallisable amino acids (Hadey et al., Thermochim. Aαa 156, 247-257(1989); Yu et al., J. Chem. Soc. Faraday Trans. 91, 1511- 1517(1995)) might therefore be preferred to TBA as sublimation accelerators in parenteral produαs, if for no other reasons than regulatory and safety concerns. TBA appears to be a better accelerant when used with PW, with bicarbonate being superior to the other salts. In frozen sucrose solutions, all additives tested had a benefit, but TBA was not superior to either acetate or formate. In the laαose formulations acetate and TBA would seem to be the preferred additives. The freezing behaviour of the different formulations were not studied in detail, nor at cooling rates appropriate to cooling in a commercial freeze-dryer. The indications were, however, that all the additives were included in the freeze-concentrate and (at least over the brief initial sublimation period tested) mass loss was due solely to ice sublimation.

For the investigation undertaken here, freezing behaviour was examined in greater detail and differing salt concentrations were examined.

2. Freezing behaviour of excipient / salt mixtures

2A Excipients

Typically, neither laαose, PW nor sucrose forms a euteαic mixture on cooling. When cooled sufficiently the residual unfrozen solutions form a glass with the following charaαeristic glass temperatures.

Solution' T ' ²

25% PW (K17) -24.5 °C

12.5% Laαose³ -28.3 °C 25% Sucrose -31.6 °C

1. The value of T ' is independent of initial concentration.

2. As, measured at a cooling rate of 5 °C/min, these excipients are known not to crystallise, the freezing behaviour during slow cooling was not examined.

3. The solubility of laαose in water at room temperature is less than 25 g/100 ml. - lo -

2.B Salt solutions

Euteαic crystallisation is dependant upon (stochastic) crystal nucleation and upon

(Arrhenius) crystal growth. The freezing behaviour of the salts, therefore, is expeαed to depend both on concentration (as well as sample volume) and cooling rate. Crystallisation may occur during freezing (see Fig 2).

Alternatively crystallisation may occur on subsequent warming or not at all. See Figure 3, which shows two traces for the warming of the same 1M ammonium formate solution. Neither experiment showed euteαic crystallisation on cooling (at 5 °C/rrin) to -60 °C. In one instance (fig 3A) crystallisation occurred on warming; a crystallisation exotherm and a euteαic melting endotherm may be seen. In the second example (fig 3B) no crystallisation occurred during warming of the frozen solution.

The euteαic temperatures of the salt solutions used could not readily be found in the literature. We estimate the euteαic temperature of ammonium bicarbonate to be in the range -5 to -6 °C. Although euteαic solidification could be seen on cooling, we could not resolve the euteαic melt form the ice melting endotherm (see Fig 4, next page). Ammonium formate was found to have T_c = -36 °C (Fig. 3 above). In this study we were unable to measure a euteαic temperature for a solution of 2 M ammonium acetate, even after holding for 18 hours at -60 °C. Although there is some evidence for vitrification (see Fig 5 A, T_g = - 47 °C), this is approaching the limits of instrument stability. A comparison of power-time curves with that recorded for ammonium bicarbonate (no melting below T_e) suggests that ice melting begins below -60 °C, i.e. no crystallisation or vitrification has occurred. Analysis by cryogenic DSC may clarify the behaviour of this solution. For the purposes of this study it may be concluded that ammonium acetate solutions do not show euteαic crystallisation over the temperature ranges of interest.

2.C Excipient / salt mixtures

In general, adding a non crystallisable excipient to a salt solution would be expeαed to inhibit euteαic solidification, especially at higher excipien : salt ratio's. For example, a solution of ammonium bicarbonate has the following % by weight of solute.

Concentration %w/w ammonium bicarbonate

1 0.M 7.9 %

0.8 M 6.3 %

0.6 M 5.7 %

0.4 M 3.2 % 0.2 M 1.6 %

0.1 M 0.8 %

By way of example, the freezing behaviour of a solution of 5% sucrose with varying ammonium bicarbonate concentrations is given in Fig 6 (next page). Over the range 0 to 0.6 M ammonium bicarbonate in 5% sucrose, the glass transition temperature is depressed by the increasing amounts of salt, as given below

Sucrose Ammonium bicarbonate T_£'

5 % Nil -32 °C 5 % 0.1 M -39 °C

5 % 0.2 M -43 °C

5 % 0.4 M -48 °C

5 % 0.6 M -52 °C

5 % 0.8 M -36 °C

In the 5% sucrose 0.4 M ammonium bicarbonate solution, a crystallisation exotherm (trough) may be seen (Fig 6A, next page) at circa -22 °C. The ammonium bicarbonate has crystallised when warmed above the glass temperature, but did not do so on cooling.

Increasing the ammonium bicarbonate to 0.6 M causes the ammonium bicarbonate to crystallise more readily when warmed above T_g'. In this instance, some crystallisation was also seen on cooling (see Fig 7, below). When the ammonium bicarbonate concentration is 0.8 M, then euteαic crystallisation is completed during cooling (see Fig 8, below). Only the salt, andnot the sucrose crystallises. Although no further crystallisation occurs on warming, not all the salt is removed from solution, and some ammonium bicarbonate remains within the freeze-concentrated glass, causing a depression of T_g' (see fig 6 next page).

The observed behaviour is therefore a balance of probability and time. Low temperatures increase the probability of crystal nucleation and higher temperatures increase therate of crystal growth. Thermal history, and hence cooling rates, therefore may affeα the freezing behaviour. In our analyses of excipient salt mixtures, we have where necessary included analysis at a cooling rate of 0.3 °C/min, typical of many commercial freeze-dryers.

2.D Summary of experimental results.

The following abbreviations have been used in the tables

Ambic for ammonium bicarbonate.

AmAc for ammonium acetate.

AmForm for ammonium formate (formic acid, ammonium salt).

A standard excipient content of 5% was chosen for the following reasons

5% solids is the minimum needed for mechanical stability of the dried produα. All solutions would be in the approximate range 5-10%w/v solution and thus are not expeαed to show undue inhibition of sublimation.

2.D.1 Excipients only

Glass temperatures on

Normal Cooling Slow cooling

25% Sucrose -31.6 °C Not determined (n.d.) 25% PW -24.5 °C n.d. 12.5% Laαose -28.3 °C n.d.

As these excipients are known not to crystallise, freezing behaviour during slow cooling was not examined.

2.D.2 Salts only

A study of the freezing behaviour of the salt solutions was not a major aim of this projeα. Accordingly this was not studied extensively.

Eutectic solidification on Normal cooling Slow cooling

1 M Ambic Yes n.d.

T_e not resolved from ice melting

I M AmAc No(T'_g at -46°C?) slow crystallisation on cooling (T. at -47°C?)

2 M AmAc No (T ' = -47 °C?) No, even after slow cooling and 18 h at -60 °C

1 M AmForm Variable, crystallisation on n.d. warming T„ = -35 °C 2.D.3 Excipient salt mixtures

Note: references below to crystallisation refer only to the salt component. The excipient forms a freeze-concentrated glass and T ' may be depressed by the inclusion of some residual salt.

2.D.3.a Ammonium bicarbonate solutions

2.D.3.a.i Sucrose

Glass temperatures on

Normal Cooling Slow cooling

5% sucrose with

0.1 M Ambic -39.0 °C 38.9 °C

0.2 M Ambic -43.2 °C 42.9 °C cooling exotherm *

0.4 M Ambic -47.8 °C 47.6 °C crystallises on warming crystallises on warming

0.6 M Ambic -52.0 °C -35.4 °C slight euteαic crystallisation euteαic crystallisation crystallises on warming

0.8 M Ambic -35.6 °C euteαic crystallisation

* see discussion of PW / Ambic below. .D.3.a.ii PVP

Glass temperatures on

Normal Cooling Slow cooling

5% PW with

0.1 M Ambic -32.1 °C -27.7 °C crystallises on warming complex euteαic see below cooling exotherm

0.2 M Ambic -28.6 °C -28.2 °C euteαic crystallisation euteαic crystallisation cooling exotherm

0.4 M Ambic -29.4 °C -30.9 °C euteαic crystallisation euteαic crystallisation

0.6 M Ambic -31.2 °C -32.5 °C euteαic crystallisation euteαic crystallisation

0.8 M Ambic -33.1 °C n.d. euteαic crystallisation

Ammonium bicarbonate crystallises quite readily from a 5% PW solution, compared to a solution in 5% sucrose. Crystallisation from a 5% PW / 0.1 M ammonium bicarbonate solution, cooled at 0.3 °C/min shows two crystallisation exotherms but the origin of these are not understood. Figures A.l and A.2 (Appendix A) shows a comparison of the DSC cooling traces for the 0.1 M and 0.2 M salt formulations. Both of these formulations show a continuing output of heat as the temperature is lowered, suggesting further slow crystallisation. This is absent from the cooling traces for 5% PW solutions with higher concentrations of ammonium bicarbonate, but was noted for some other mixtures (marked "cooling exotherm" in the tables. This denotes an output of heat on cooling but not a "typical" crystallisation trough).

Although euteαic crystallisation does happen on cooling, the depression of T ', as the concentration of ammonium bicarbonate increases, refleαs an increasing amount of residual salt in the freeze-concentrate. The differences in T_g' found, for a given formulation, with the two cooling rates suggests similar small differences in the composition of the amorphous phase. Crystal growth at the lower cooling rate (and therefore higher temperature) would be expeαed to be more complete, resulting in a higher T ', although this does not appear to be the case.

The DSC traces of the PW / ammonium bicarbonate solutions show the best resolution of the ammonium bicarbonate / water euteαic melt from the bulk ice melting endotherm. Although in this ternary system, the partly resolved euteαic melt cannot be assumed to be the ammonium bicarbonate / water binary euteαic, it is unlikely that any PW crystallisation occurs. The estimated value for T_e (see Fig A.3) is -5.8 °C. This agrees closely with the previously estimated value (Fig. 4 above).

2.D.3.a.iii Lactose

Glass temperatures on

Normal Cooling Slow cooling

5% laαose with

0.1 M Ambic -32.4 °C -32.5 °C cooling exotherm

0.2 M Ambic -38.7 °C -38.5 °C cooling exotherm

0.4 M Ambic -44.9 °C -43.8 °C some crystallisation on warming some crystallisation on warming

0.6 M Ambic -48.6 °C -31.8 °C crystallises on warming euteαic crystallisation

0.8 M Ambic -32.9 °C n.d. complex euteαic crystallises on warming (below T_g

The behaviour of laαose / ammonium bicarbonate was broadly similar to the behaviour of the sucrose formulations. Cooling of the 0.8 M ammonium bicarbonate solution showed a complex euteαic crystallisation (see Fig A.4) similar to that recorded for the 5% PW / 0.1 M ammonium bicarbonate solution. On warming this solution, there is an apparently anomalous crystallisation at a temperature below the glass transition temperature (see Fig A.5). It should be noted, however, that the DSC instrument scans in time; with incomplete crystallisation on cooling, the glass temperature may have been below -50 °C, and the recorded glass temperature of -34.5 °C refleαs the composition of the amorphous phase after this further crystallisation.

2.D.3.b Ammonium acetate solutions

In the following tables some values are recorded simply as unreliable. With several formulations, there was no indication of crystallisation, and second order (step) transitions could only be found on great enlargement of the power-time trace. In several cases, transitions in the region of -47 °C, with a ΔCp value in the range 5-10 x 10^"3 J/g °C, were found. Enlargement of the power-time curves recorded for 25% PW and for 12.5% laαose (Fig. 9 below) revealed similar apparent transitions, which are therefore regarded as instrument artefaαs (note this artefaα could not be confirmed with the 25% sucrose data as the region of interests co-incided with the T_g" transition). For comparison, the recorded values of ΔCp at the known glass transitions were: PW 0.596 J/g °C and laαose 0.501 J/g °C. Analysis by DSC at ultra low temperatures is recommended if resolution of this problem is required, but note that with freeze-drying temperatures at or below -50 °C, cycle times may be impraαicably long.

2.D.3.b.i Sucrose

Glass temperatures on

Normal Cooling Slow cooling

5% sucrose with

0.1 M AmAc -42.4 °C

0.2 M AmAc unreliable unreliable cooling exotherm

0.4 M AmAc unreliable n.d.

0.6 M AmAc unreliable n.d.

0.8 M AmAc unreliable unreliable cooling exotherm

2.D.3.b.ii PVP

Glass temperatures on

Normal cooling Slow cooling

5% PW with

0.1 M AmAc -42.8 °C -43.2 °C cooling exotherm

0.2 M AmAc -51.4 °C n.d. 0.4 M AmAc unreliable n.d.

0.6 M AmAc unreliable n.(

0.8 M AmAc unreliable n.d.

The recorded glass temperatures for the 0.1 M and 0.2 M solutions are consistent with glass temperatures below the recorded range for formulations containing greater amounts of the acetate salt.

2--D.3-b.iii Lactose

Glass temperatures on Normal cooling Slow cooling

5% laαose with

0.1 M AmAc -38.9 °C n.d. some crystallisation on cooling

0.2 M AmAc -46.7 °C n.d.

0.4 M AmAc unreliable n.d. 0.6 M AmAc unreliable n.d.

0.8 M AmAc unreliable n.d.

The glass temperature of the 0.2 M ammonium acetate formulation is consistent with a depression of T„' below the recorded range when the salt concentration is increased. In the case of the 0.2 M formulation the value of ΔCp, 0.137 J/g °C, allowed a reliable assignment of the transition. There is evidence for euteαic crystallisation of some salt during cooling of the 0.1 M acetate formulation (see Fig. A6). Unusually, however, there seems to be no crystallisation from formulations containing greater amounts of the salt (see also 5% laαose / 0.1 M ammonium formate, below).

2.D.3.C Ammonium formate solutions

2.D.3.c.i Sucrose

Glass temperatures on

Normal Cooling Slow cooling

5% sucrose with

O.l M AmForm -43.7 °C -43.1 °C cooling exotherm

0.2 M AmForm n.d. n.d.

0.4 M AmForm n.d. n.d.

0.6 M AmForm n.d. n.d.

0.8 M AmForm n.d. n.d.

Glass temperatures on

Normal cooling Slow cooling

5% PW with

O.l M AmForm -40.8 °C -39.9 °C

0.2 M AmForm circa -53 °C n.d.

Close to starting transient 0.4 M AmForm <-55 °C ^■ n.d. Below range

0.6 M AmForm <-55 °C n.d. Below range

0.8 M AmForm <-55 °C n.d. Below range

The plasticising effect of ammonium formate can be seen by comparing the T ' values for PW, and for PW containing 0.1 M and 0.2 M ammonium formate.

2.D.3.c.iii Lactose

Glass temperatures on

Normal Cooling Slow cooling

5% laαose with

0.1 M AmForm -38.8 °C -40 °C some crystallisation on cooling no crystallisation

0.2 M AmForm n.d. n.d.

0.4 M AmForm n.d. n.d.

0.6 M AmForm n.d. n.d.

0.8 M AmForm n.d. n.d.

The 5% laαose / 0.1M ammonium formate sample showed a small amount of crystallisation on cooling at a rate of 5 °C/min (see Fig. A.7). The absence of any crystallisation exotherm for the same sample when cooled at 0.3° C/min emphasizes the random nature of this event. 3. Sublimation of samples

Freeze drying experiments were carried out using 1 ml samples in 1.6 cm i.d. vials. Sublimation times estimated to be in the region of 24 h (5% solids, i.e. low salts) to 36 h (5% excipient plus 0.4 - 0.6 M salts) to upwards of 48 h (5% solids 0.8 M salts). Aαual sublimation times will vary with solids content and temperature

For comparison 0.1 M Ambic is 0.8 % solids 0.2 M Ambic is 1.6 % solids

0.4 M Ambic is 3.2 % solids 0.6 M Ambic is 5.7 % solids 0.8 M Ambic is 6.3 % solids

The time course data points were not taken at regular intervals, typically overnight steps are longer. The plotted time courses of sublimation are extrapolated from the available data with this limitation.

3.A Sucrose / Ambic Sublimation

Primary drying at -46 to -47 °C (see Figure 9C).

The 1 M ammonium bicarbonate solution dries more quickly than any other, but this may simply refleα the (lowest) solids content. Over the total period the 0.8 and 0.6 M salt with sucrose are similar to sucrose alone. The 0.2 and 0.4 M salt additions accelerate drying compared to sucrose alone. Surprisingly, the DSC analysis would prediα that only the 0.8 and 0.6 M salt with sucrose contain a volatile euteαic solid.

3.B PVP / Ambic Formulation

Primary drying at -37 to -38 °C (see Figure 9D). The DSC analysis suggests that all the salt inclusion formulations (0.2 to 0.8 M) would form euteαic mixtures on cooling. Relative to PW alone, there is no rate enhancement; the 0.4 M formulation dries more slowly than the 0.6 and 0.8 M mixtures, which are similar. The 0.2 M formulation behaves similarly to PW alone. With the exception of the 0.4 M formulation these results would appear to show only the expeαed decrease in sublimation rate with increasing solids content.

3.C Lactose / Ambic Formulation

Primary drying at -46 °C (see Figure 9E).

The laαose / Ambic formulations show the same features as the sucrose mixtures. The formulations expeαed to contain volatile euteαic solids, dry more slowly than laαose alone. The 0.2 and 0.4 M compositions, show an enhancement of sublimation rate, compared to laαose alone, despite having an increase solids content.

3.D Sucrose / AmAc Formulation

Primary drying at -46 °C (see Figure 9F)

Note: this data set includes an additional formulation containing 0.1 M ammonium acetate in 5% sucrose.

With the exception of the 0.1 M ammonium acetate formulation, in all cases drying was carried out above T ' which was below the region accessible to the freeze-drying equipment. All formulations, except 5% sucrose 0.8 M ammonium acetate dry more rapidly than 5% sucrose alone. Since, however, we may be comparing evaporation rate in many cases (with the associated evaporative cooling) the results are not considered a reliable guide. In praαice, freeze-drying any similar formulation with an acetate content greater than 0.1 M would not be recommended. In consideration of the same argument, the ammonium formate salts were not studied, except at the 0.1 M concentration (see seαion 6 below). 3.E PVP / AmAc Formulation

Primary drying at -46 °C (see Figure 9G).

The results from this drying are erratic. Given, however, that in all cases drying was carried out above any presumed T ', then we may be observing evaporation coupled with physical collapse. In the cases of the highest salt concentration (0.6 and 0.8 M salt) then wholesale collapse would be expeαed to limit the drying rate and thus may explain the apparent lack of drying.

3.F Lactose / AmAc Formulation

In consideration of the results found with sucrose and PW formulations in the presence of ammonium acetate, it was felt that performing another sublimation, at a temperature in excess of T_g', was not warranted. Given the unexpeαed, inverse relationship between drying time and the presence of volatile euteαic solids, an investigation of excipients containing 0.1 M salt additions offered more scope for success.

4. DSC Analysis of Sublimed Samples

The original hypothesis, that a volatile euteαic solid would enhance the rate of sublimation, by decreasing the resistance to drying, as the euteαic solid itself sublimed, does not seem to be supported by available data. One consequence of this hypothesis, would be removal of the euteαic solid, and thus if the (partially) dried material was reconstituted, then a change in T_g' would be expeαed. Samples removed from the dryer at the end of ice sublimation were therefore reconstituted and analysed by DSC. The data shown represent analysis at a cooling and warming rate of 5 °C. Sucrose / Ambic

Tg' °C

Salt content Before Sublimation After Sublimation

°C °C

0.2 M -43 -42 0.4 M -50 -48 0.6 M -52 -52 0.8 M -36 -38

PVP / Ambic

Tg' °C

Salt content Before Sublimation After Sublimation

°C °C

0.2 M -29 -27 0.4 M -29 -28 0.6 M -31 -29 0.8 M -33 -32 - WO 99/30688 - . 34 - PCT/GB98/03747

Laαose / Ambic

Tg' °C

Salt content Before Sublimation After Sublimation °C °C

0.2 M -39 -35

0.4 M -45 -43

0.6 M -49 -47 0.8 M -33 -33

In almost all cases there is little change indicating that substantial salt remains after sublimation, at the lowest temperate in the drying cycle. These samples have not been subjeαed to 2° drying and the results do not preclude the loss of volatile materials during warming. If there is any trend then those samples which contain the least salt, show the greatest, albeit small, differences.

A simplistic model of the variation of T ' with composition would show an approximate linear relationship between T ' and the mass fraαion of the solids (expressed as mass solid A/ (mass solid A + mass solid B) without regard for the water). For these compositions the mass fraαions (Ambic/ (PW+ Ambic) are

Salt content Mass fraαion Ambic %

0.2 M 24.0 %

0.4 M 38.7 %

0.6 M 48.7 %

0.8 M 58.8 %

and thus if the rate of loss of volatiles was a funαion of temperature (and surface area) but not original content, then a difference would be more apparent at lower concentrations. . WO 99/30688 - 35 - PCT/GB98/03747

5. Observations of samples removed from the dryer to measure sublimation rate

Following ice sublimation, samples without any further secondary drying at higher temperatures will contain close to w ' g water per g of produα. On warming to room temperature, such samples are not expeαed to remain in a glassy state. For most of the samples analysed, the expeαed collapse was observed. In three cases, however, on the completion of primary drying the product renamed physically sta vhenvxrmed to at ambient temperature and did not show any collapse.

The formulations were:

5% Sucrose 0.4 M Ambic

5% PW

5% PW 0.6 M Ambic

The photograph (see Figure 9FT) shows three vials from the 5% sucrose / Ambic series, 7a, 7b and 7c after primary drying for 51 hours. The compositions are 7a - 5% sucrose

7b - 5% sucrose / 0.2 M Ambic 7c - 5% sucrose / 0.4 M Ambic. Vial 7a has dissolved, 7b has a collapsed cake, but 7c has an apparently normal dry cake. Surprisingly the 0.6 and 0.8 M homologues did not show good cake structure, presumably due to differences in the w ' content.

In the case of the 5% PVP 0.6 M ammonium bicarbonate and, possibly, the 5% sucrose 0.4 M ammonium bicarbonate, this physical stability may result from a supporting mass of salt crystals. This type of system, in which a collapsed amorphous state is supported on a crystal matrix has been described for the sucrose / NaCl / water system (reference 12). Such a system may only exist within narrow limits of composition as collapse may still be expeαed WO 99/30688 - - 36 - PCT/GB98/03747

if sufficient unfrozen water is released from the glass on warming, which then dissolves the crystalline salt.

This rationalisation cannot apply to the 5% PW sample and it is difficult from the observed freezing behaviour to account for the "solid" product from 5% sucrose 0.4 M Ambic (euteαic only seen on warming). It is noted that the sucrose 0.4 M Ambic sample has the fastest drying of the sucrose / Ambic data set. It is quite likely, at least in the case of 5% PW, that this phenomenon has not been observed simply because removing samples before secondary drying is contrary to accepted praαice.

Regardless of the underlying cause, such systems offer significant benefits in decreasing produαion time. This is modelled below for a hypothetical produα with T ' = -30 °C, based upon a 1 ml sample in a vial of 1.8 cm internal diameter.

Standard Cycle Primary drying only cycle

Freezing 3 h 3 h

1° drying (at -35 °C) 5 h 5 h

2° drying 13 h Hold 2 h 2 h

Overall 23 h 10 h

6. Sublimation of 0.1 M salts containing formulations

In consideration of the results with the ammonium acetate and bicarbonate solutions, it was decided that lower concentration (i.e. 0.1 M) salt containing solutions should be further investigated. Some data was already available (ref. 8) and initial results showed that accelerated sublimation was only seen at lower salt concentrations. It is possible that at a greater salt content the inhibition caused by the increased solids more than balances any acceleration found. In addition, it was also decided to extend the salts studied to include ammonium formate. Freeze drying experiments were carried out using 1 ml samples in 1.6 cm i.d. vials. 6.A Freezing behaviour

This data has been presented above, but it reproduced below for ease of reference.

6A.1 Sucrose

Glass temperatures on

Normal Cooling Slow cooling

5% sucrose with

0.1 M Ambic -39.0 °C -38.9 °C

0.1 M AmAc -42.4 °C n.d.

O.l M AmForm -43.7 °C -43.1 °C cooling exotherm

6A.2 PVP

Glass temperatures on

Normal cooling Slow cooling

5% PW with

0.1 M Ambic -32.1 °C -27.7 °C crystallises on warming complex euteαic cooling exotherm

0.1 M AmAc -42.8 °C -43.2 °C cooling exotherm

O.l M AmForm - -40.8 °C -39.9 °C A.3 Lactose

Glass temperatures on

Normal cooling Slow cooling

5% laαose with

0.1 M Ambic -32.4 °C -32.5 °C cooling exotherm

0.1 M AmAc -38.9 °C n.d. some crystallisation on cooling

O.l M AmForm - 38.8 °C -40 °C some crystallisation on cooling no crystallisation

6.B Sublimation of 0.1 M Salt containing solutions

6.B.1 Overview

Primary drying at -46 °C (see Figure 91).

The sucrose / ammonium acetate data set shown previously contained the 0.1 M salt result for comparison. This 0.1 M result, although showing a higher sublimation rate than sucrose alone, had a lower rate than the 0.2 and 0.4 M homologues. The principal aim of this projeα, however, is not to deueiop faster drying cydes, but to facilitate rapid rehydration. Comparative drying rates are intended as a guide to changes that might facilitate dissolution.

The results present a spread of rates. Of all the salts, the bicarbonate series seem to have the slowest sublimation rates. Overlapping this set are laαose/formate, laαose/acetate, WO 99/30688 - ^{" "} PCT/GB98/03747

sucrose/acetate and PW/formate. The highest sublimation rates are shown by sucrose/formate and sucrose/acetate.

The rate of sublimation from the ammonium bicarbonate set, was very low (see individual data sets below and compare with salt series homologues above). The reason for this is not understood. All vials of the 0.1 M salt series were dried in the chamber together, but preferential removal of the most volatile salts seems unlikely as the previous data suggests little if any loss of "volatile" material during primary drying. Excipient only vials (subjeαed to primary drying only) from the full sucrose/ cetate, sucrose/ Ambic, laαose/Ambic and PW/ Ambic, data sets (above) were reconstituted and T ' measured. No significant difference was found between the observed and expeαed values, suggesting that gas phase transfer of "volatile" salt does not occur to any measurable extent. There is, however some indication that (at a fixed solids content) the rate of sublimation decreases as (T-T ') increases.

Some samples from the ammonium bicarbonate and ammonium formate series were subjeαed to secondary drying and, following reconstitution, T_g' measured to assess loss of salt. (NB. residual mass is unlikely to be reliable as the residual water content (w is not known and may vary with residual salt content).

6.B.2 By category

6.B.2.I. Sucrose (see Figure 9J).

The drying rate in this case increases as (T-T_g') decreases (where T is the drying temperature).

6.B.2.2. PVP (see Figure 9K).

Only the ammonium bicarbonate mixture is expeαed to containing a euteαic solid. Again there is an approximate inverse relationship between drying rate and (T-T ').

6.B.2.3. Lactose (see Figure 9L). Very little, if any euteαic crystallisation is expeαed. The rates are all similar, if anything the bicarbonate mixture is the least porous (i.e. least resistance to sublimation). It may be coincidence that the acetate and formate formulations have similar T ' .

6.B.2.4. Ammonium bicarbonate (see Figure 9M).

For this series of a given salt, there is not an inverse relationship between (T-T_g') and drying rate. All rates are similar and only the PW formulation would exhibit euteαic crystallisation of the bicarbonate salt. In retrospeα, with this series of vials there was very little sublimation at all (see 6.B.1 overview, above).

6.B.2.5. Ammonium acetate (see Figure 9N).

In this series the PW sample seems to have greatest prospeα of an open structure.

6.B.2.6. Ammonium formate (see Figure 90).

The drying rates are broadly similar for these mixtures.

6.B.2.7 Analysis of salt loss following primary and secondary drying.

Data from the salt series results suggested that after primary drying little, if any salt was lost from the samples. Primary drying is the lowest temperature segment of a "normal" cycle, and by analogy to the residual water content, the sample would be expeαed to contain the most amount of residual "volatile" salt. Some samples were therefore subjeαed to a secondary drying protocol, with a final temperature of 30 °C. These were then reconstituted with distilled water and T_g' measured. A loss of salt would be indicated by a shift in T_g' towards the excipient only value. Tg^{, 0}C

Before Sublimation After Sublimation Excipient only

5% Sucrose with -32 °C

0.1 M Ambic -39.0 -36.9

O.l M AmForm -43.6 -44.4

5% PW with -24.5 °C

0.1 M Ambic -32.2 -24.7

O.l M AmForm -40.8 -27.6

5% Laαose with -28 °C

0.1 M Ambic -32.4 -33.2

O.l M AmForm -38.8 -39.2

Only in the case of the PW formulations has the salt been substantially removed. It is expeαed a PW / 0.1 M acetate solution will behave similarly. The loss of ammonium bicarbonate from the 5% PW / 0.1 M Ambic formulation suggests that inhibition of salt sublimation was not the rate limiting f aαor causing the unexpeαedly low rate of ice sublimation from the Ambic series. 7. Selection of formulations for further study

The rate of sublimation, from a frozen sample, is principally governed by the resistance within the produα cake itself. Thus studying sublimation rates should be a guide to resistance both to dehydration and to rehydration. From our knowledge of the principles of freeze-drying and considering the data above, there would seem to be a balance of three faαors governing choice of suitable model systems :-

1. The rate of ice sublimation is a funαion of temperature. 2. For a given salt additive, at a fixed drying temperature, there may be an inverse relationship between sublimation rate and (T-T_g'). 3. Residual salt left in the produα may be undesirable.

One the basis of point 3, then only the PW formulations contain little or no residual salt. It is likely that this is lost during secondary drying and not primary drying. There is little to choose between the PW / formate and PW / acetate formulations at the 0.1 M level. Since the formate solutions have a higher T_g', then the overall rate of ice sublimation at an optimum drying temperature would be higher. A similar argument applies to the PW / Ambic formulations, although the rate of sublimation at -46 °C was low, at an optimised drying temperature of

-37 °C, the temperature effeα alone would increase the rate of ice sublimation by approximately 150%. The higher T ' temperature will also be of value, when formulating with drug substances. Any plasticising aαive materials would depress T ' and with the formate salt may therefore lead to undesirable but necessary long cycle times.

Thus both the formate and bicarbonate formulations of PW were seleαed as candidate formulations.

8. Freezing behaviour of Eletriptan and Sildenafil

Neither of these compounds were particularly soluble. Eletriptan.HBr was barely soluble at 10 mg /ml, even with gentle warming. Sildenafil could only be dissolved in acidic solution. WO 99/30688 - . 43 - PCT/GB98/03747

For the purposes of these experiment, Sildenafil solutions were acidified with hydrochloric acid (which is volatile and will be lost from lyopr-ilised samples, which may then be redissolved in an appropriate buffer (e.g. citrate)). It was not possible to unambiguously assign values of T ' for either Eletriptan or Sildenafil. The aim if this projett was to investigate the properties of formulated produαs, and thus aqueous solutions of the drug substances were not studied in detail.

8A Eletriptan

A 10 mg/ml solution, in water was prepared with gentle heating. The material was not fully soluble and was centrifuged, before analysis, to give a saturated, nominally 10 mg/ml, solution.

There was no evidence of euteαic crystallisation on cooling. It was difficult to establish a glass transition on warming. Fig A.8 (Appendix A), shows a candidate transition at -15 °C. This assignment, however, must remain tentative as the change in heat capacity is small. The full power-time trace, recorded on warming (fig A.9) has an unusually sharp onset of the ice melting endotherm. Overall the evidence suggests that ice melting does not begin below -55 °C (i.e. T ' is above this), but, although it is unlikely, one cannot exclude the possibility of a euteαic crystallisation, at the poorly controlled lower end of the temperature scale (see behaviour of Eletriptan formulated with ammonium bicarbonate, below).

8.B Sildenafil

The Sildenafil was dissolved in 100 mM HC1 to give a 10 mg/ml solution. There was no euteαic crystallisation on cooling. On warming, no reliable transition could be assigned. An extremely weak transition is present at -34.5 °C but the magnitude of the change is equivalent to baseline artefaαs known at this great scale enhancement. It is only considered as a candidate as it would appear that ice melting starts in this region (see Fig A.10). . WO 99/30688 - . 44 . PCT/GB98/03747

9. Freezing behaviour of formulated Eletriptan and Sildenafil

9.A Eletriptan / PVP formulations

9 A.1 Ammonium bicarbonate

A 5 mg/ml solution of Eletriptan in 5% PW and 0.1 M ammonium bicarbonate, pH 8.5, was prepared and centrifuged to clarify (note it was not possible to prepare a 10 mg/ml solution). The freezing behaviour of this formulation was quite complex.

On cooling at 5 °C/min, there was no evidence of euteαic crystallisation, but the sample crystallised on warming. The full power-time curve is given in Fig A.ll. Enlarged segments of the DSC output are shown in Fig 10 (over page). There is a glass transition at -32 °C, followed by a crystallisation, and then by a melting endotherm (peak at -8 °C, onset ciwz -12 °C). This is presumably a euteαic melt (see later). Its identity is unknown but an Eletriptan / water binary euteαic is the most likely identification. Following this we can see the ammonium bicarbonate / water T_e and finally the bulk ice melting endotherm. As ammonium bicarbonate will aα as a plasticiser, this analysis gives a lower limit for the T ' of Eletriptan.

After cooling slowly at 0.3 °C/min, enlargement of the cooling trace showed some euteαic crystallisation (see Fig A.12, which also shows the cooling curve for 0.1 M Ambic / 5% PW, for comparison). From the behaviour of the PW / Ambic solution, crystallisation of ammonium bicarbonate would be expeαed. On warming (see Fig 11 below), the glass transition was found to have increased to -18 °C. This is consistent with the presence of some residual salt lowering the T ' of Eletriptan from cτrox -15 °C. The euteαic melting endotherm is again present.

In an attempt to clarify some of the observed changes, the sample was subjeαed to a set of heating and cooling cycles, with the objeα of discriminating between reversible (e.g. glass transitions) and irreversible (e.g. melting) changes. The regime was (rates are 5 °C unless otherwise given)

Regime Cycle segments

Cool to -65 °C at 0.3 °C/min Cool RC1

Warm to -15 °C Heat RH1

Cool to -25 °C Cool RC2

Warm to -7 °C Heat RH2

Cool to -15 °C Cool RC3 W Waarrmm ttoo + +3300 °°CC Heat RC4

There was no evidence of euteαic crystallisation on cooling and, on warming (Heat RH1), a glass transition and devitrification was seen, similar to the "normal" 5 °C/min cooled sample (Fig 10). T ' was found to be -33 °C. This trace (Heat RH1) is given in Figure A.13.

On the second heating cycle (which just melts the 12 °C cnset euteαic), see Fig 12, only the euteαic melt is seen (sample not cooled sufficiently to reach T '). The figure shows comparisons to the earlier normal and slowly cooled (euteαic crystallisation on cooling) samples.

As expeαed, the third warming (RH3), see Fig. 13 does not show the 12 °C cnset melt, but only the ammonium bicarbonate / euteαic melt and the bulk ice melting endotherm.

Two further experiments were run on the same sample. In one experiment the sample was cooled at 5 °C and annealed by warming to -25 °C. In the second, cooling was at 0.3 °C and annealing was at -20 °C for 15 minutes. Prior to annealing, both experiments gave T,.' = -32 °C and on subsequent warming both of these showed T_g' = -28 °C. (Fig A.14). Both also showed some evidence of slow crystallisation on cooling.

In summary, this formulation seems to show variable crystallisation. The glass temperature depends upon the degree of residual salt, but this seems difficult to control. Although both annealed samples showed the same T_g', this was unexpeαedly below that recorded in one of the slow cooling experiments. This in part may be due to the partial and variable WO 99/30688 . 4^ . PCT/GB98/03747

crystallisation of two components.

For the purposes of freeze-drying, the lowest T ' value of -32 °C can be used to determine satisfaαory cycle parameters, although the composition of the produα may be somewhat variable.

9A.2 Ammonium formate

A 10 mg/ml solution of Eletriptan.HBr in 5% PW, 0.1 M ammonium formate, pH 5.7, was prepared by warming. The solution was centrifuged after cooling to ensure a clear solution for analysis. There was no evidence of euteαic behaviour and a glass transition was found at -38 °C (see Fig A.15).

When the Eletriptan.HBr content is reduced, to 5 mg/ml, the glass transition temperature decreases to -40 °C. (Fig A.16).

9.A.3 No salt; Eletriptan 5 mg/ml in 5% PVP

As expeαed, this formulation does not show any euteαic crystallisation. The glass transition temperature (Fig. A.17) is -25 °C, close to that expected for PW type K17. Based upon the assignment of T_g' (Eletriptan) = -15 °C, a slight increase relative to T ' (PW K17), of -24.5 °C, would have been expeαed.

9.B Sildenafil / PVP formulations

Formulations were prepared by dilution of a stock solution of Sildenafil, 10 mg/ml, in 100 mM HC1 and are (final concentration) Sildenafil 5 mg/ml in 50 mM HO, 5% PW. WO 99/30688 - . _γ . PCT/GB98/03747

9.B.1 Ammonium bicarbonate

On preparation of this solution, there was an immediate and heavy precipitate (presumably of Sildenafil). Thus this formulation is of no praαical use for freeze-drying.

9.B.2 Ammonium formate

The freeze-thaw behaviour of 5 mg/ml Sildenafil, 5% PW, 0.1 M ammonium formate does not show euteαic crystallisation. The glass temperature (Fig. A.18) is - 39 °C.

9.B.3 No salt; Sildenafil 5 mg/ml in 5% PW

Although, no enhanced porosity of the cake is to be expeαed, this formulation may have value, if the need for secondary drying can be eliminated. There is no euteαic crystallisation, on cooling, and the glass temperature is -26 ° C (Fig. A.19). Note: the data in 9.B.2 and 9.B.3 show T ' values above that of 5% PW, 0.1 M ammonium formate (-41 °C) and below that of 5 % PW (-24.5 °C). Thus the T_g' of Sildenafil, in HO, must be between these limits.

9.C Summary of glass temperatures

Eletriptan Sildenafil

10 mg/ml 5 mg/ml 10 mg/ml¹ 5 mg/ml²

Aqueous solution -15 °C ³ 35 °C ³

5% PW -25 °C -26 °C

5% PW 0.1 M Ambic -32 °C insoluble

5% PW 0.1 M Amform -38 °C -40 °C -39°C

Notes: ' contains 100 mM HC1

^: contains 50 mM HCl

³ T ' is independent of initial dilution 10. Sublimation rates of formulated Eletriptan and Sildenafil

Freeze drying experiments were carried out using 1 ml samples in 1.6 cm i.d. vials. The following formulations were studied;

Drug PVP Salts Other

Eletriptan, 5 mg/ml 5%

Eletriptan, 5 mg/ml 5% 0.1 M Ambic

Eletriptan, 5 mg/ml 5% 0.1 M Ambic annealed at - 15 °C

Eletriptan, 5 mg/ml 5% O.l M AmForm

Sildenafil, 5 mg/ml 5%

Sildenafil, 5 mg/ml 5% 0.1 M AmForm

Annealing was carried out by cooling to below T ' and warming to the temperature shown. Primary drying was carried out at a produα temperature of -46 °C. The relative sublimation rates are shown overpage.

Although there is a marked difference between the rates of sublimation for Eletriptan / PW and Sildenafil / PW, the inclusion of the volatile salt leads to an increased sublimation rate. The difference between the two samples without salt, must arise from differences in produα cake morphology.

There is no significant difference between the annealed and non-annealed samples of Eletriptan / 5% PW / 0.1 M Ambic. In praαice therefore any crystallisation may go to completion during the (slow) cooling phase of the drying cycle. For the Eletriptan formulations, the formate salt causes the most acceleration (circa 20 %) of sublimation rate, and is thus the preferred candidate for rapid rehydration. WO 99/30688 - _ ₄₉ _ PCT/GB98/03747

The inclusion of the formate salt, also accelerates the rate of primary drying for the Sildenafil / PW formulations by 20%. (See Figure 14).

11. Preparative drying of Eletriptan and Sildenafil formulations

For preparative drying, 1 ml portions were dispensed into 1.8 cm i.d. vials. Annealing was carried out by cooling to below T ' and warming to the temperature indicated, before re- cooling.

11A Eletriptan

Samples of 5 mg/ml Eletriptan in 5% PW, 0.1 M ammonium bicarbonate were cooled to - 39 °C and then warmed to -25 °C. Samples of 5 mg/ml Eletriptan in 5% PW and in 5% PVP / 0.1 M ammonium formate were then introduced into the drier and all samples frozen to -43 °C i.e. 5 °C below the lowest T '. Calculated sublimation time, with no correαion for acceleration is 35 hours, secondary drying was limited to an increase of 5 °C/h, with a final drying temperature of 30 °C. Dried samples were stored at -40 °C.

ll.B Sildenafil

One ml samples of Sildenafil 5 mg/ml, 5% PW in 0.05 M HO and of Sildenafil 5 mg/ml, 5% PW, 0.1 M ammonium formate, in 0.05 M HCl were freeze dried together with samples of 5% PW in water. Primary drying was carried out at -44 °C (calculated drying time 38 h, assuming no acceleration), secondary drying was limited to an increase of 5 °C/h,with a final drying temperature of 30 °C. Samples were subjeαed to an extended primary drying time of 116 hours (to ensure equilibrium) and some samples removed before the secondary drying period. At the end of the drying cycle, samples were stored at -40 °C. ll.C Characterisation of dried products

ll.C.l Mechanical stability of samples subject to primary drying only

The following vials were removed from the dryer at before secondary drying

Sildenafil/PW 5 vials

Sildenafil/PW/AmForm 5 vials

PW 3 vials (none subjeαed to further drying)

It was noted that cracking of the cake had occurred during the primary drying of the Sildenafil / PW and PW only samples, during primary drying. None of the samples collapsed immediately upon removal from the dryer, but by the time vial capping was complete some shrinkage was evident in the Sildenafil / PW / AmForm samples.

Vials were initially stored for 24 h at 4 °C. No further collapse was evident.

Some vials were then stored for 22 hours at ambient temperature (20-28 °C). The Sildenafil / PW / AmForm samples all collapsed. No collapse was seen in the freeze dried PW and 2 (out of 4) Sildenafil / PW samples showed some collapse. The remaining two samples did not.

All samples were subsequently stored at -40 °C.

There would appear to be a good prognosis for developing stable formulations that do not require a full traditional freeze-drying cycle. 11.C.2 Physico-chemical characterisation

All fully dried formulations exhibited similar properties on heating. A low amplitude second order transition, in the region of 40 °C, followed by an exotherm, followed by a higher amplitude second order transition. Since this is common to both drug types, it must arise from a common feature. Although no entirely consistent pattern was seen, this first low amplitude transition was present in several, but not all second heatings. It is therefore not associated with collapse. This low amplitude transition was absent from the formulated samples subjeαed to primary drying only, yet the primary dried PW samples shows two quite distinα transitions. No entirely consistent hypothesis has been formulated but the phenomenon of the first transition followed by the apparent exotherm, may represent enthalpy relaxation of a population of low(er) molecular weight excipient in a matrix of high(er) molecular weight species as the viscosity changes from that of a true glass transition (10¹⁴ Pa.s) to that of struαural collapse (10⁷ Pa.s). This would, at least, be consistent with the absence of the "exotherm" from the second heat.

We have, in our experience, noted similar behaviour in several produαs freeze-dried in the presence of Byco, a gelatin hydrolysate, and have noted some evidence for it in at least one other PW formulation. Confidentiality obligations ensure that such observations remain essentially anecdotal, but this may be a feature of polydisperse, macromolecular excipients. Gel filtration analysis of PW solutions (F. Franks, pers. comm) has shown a biphasic molecular weight distribution. It must be noted, however, that no such "biphasic" behaviour is evident in a maximally freeze-concentrated solution, i.e. when T_g' is determined.

For the purposes of this report, we shall refer to the greater amplitude second order transition as T„ ll.C.2.1 Eletriptan formulations

DSC power-time curves can be found in Appendix A, figures A.20 to A.25. For reference purposes Fig 15 illustrates the typical features which are summarised below.

Formulation First Heat Second Heat Moisture

Transition 1 T_g Transition 1 T_g Content

Eletriptan /

PW 42.9°C 90.0°C 40.0°C 86.0°C 2.1% w/w

Eletriptan /

PW / 43.8°C 75.2°C 42.1°C 69.8°C 1.4% w/w

Eletriptan /

PW/ 36.1°C 88.7°C - 71.7°C 0.8% w/w

AmForm

ll.C.2.2 Sildenafil formulations

DSC Power-time curves may be found in Appendix A, figures A.26 to A.34.

Both of the samples of primary dried material, taken for DSC analysis, were samples that had collapsed during "storage". This would suggest that the lower temperature transition is related to the collapse phenomenon and/or changes that might take place in samples of viscosity ar a 10⁷ Pa.s. The presence of both transitions in the second heating of some Eletriptan formulations (see ll.c.2.1) excludes the possibility of a physical collapse related change.

Formulation First Heat Second Heat Moisture Transition 1 T„ Transition 1 T„

Sildenafil / PW l°drying only 24.9°C 24.7°C 14.3% w/w Fully dried 44.5°C 85.5°C 76.5°C 0.4% w/w

Slidenafil / PW Amform 1% drying only 10.5°C 5.7°C 12.4% w/w Fully dried 42.5°C 77.1°C 70.8°C 0.7% w/w

The glass temperatures of the primary dried samples, especially the Sildenafil / PW formulation, suggest that it may be possible to produce ambient temperature stable produαs either with limited secondary drying or simply by increasing the PW content.

ll.C.2-3 PVP solution

DSC Power-time curves may be found in Appendix A, figures A.34 and A.35. The sample analysed had not collapsed during storage, and surprisingly two transitions are evident.

Formulation First heat Second Heat Moisture

Transition 1 T„ Transition 1 T„

1° drying only 35.7 °C 62.5 °C 32.5 °C 14.1% w/w

The high glass temperature and lack of physical collapse of this sample during "storage" augers well for the production of stable PW formulations without the need for a full freeze- drying process.

11.C.3 Dissolution studies

Freeze-dried samples were removed from the freezer and allowed to warm to room temperature. One ml of water (Eletriptan samples) or 100 mM HG (Sildenafil samples) was added to each vial. The times taken for disintegration of the cake and clarification of the resultant solution (with gentle rocking) were noted. Sample Disintegration Clarification

Eletriptan / PW instant 20-30 sec.

Eletriptan / PW/ Ambic instant < 10 sec.

Eletriptan / PW / AmForm instant 7 sec.

Sildenafil / PW instant 15-20 sec

Sildenafil / PW/ AmForm instant 15-20 sec

11.C.4 Handling properties

In a fortuitous accident, one vial of Eletriptan / PVP/ AmForm was dropped and the vial broke. The cake survived the impaα without disruption, and further more was sufficiently robust to be handled. The cake disintegrated immediately on contaα with water.

DISCUSSION

The addition of inorganic salts increases the rate of ice sublimation from a frozen solution. This acceleration effeα can be of the order of a 20-25 % decrease in primary drying time. Ammonium acetate was found to be the least useful of the additives used, principally owing to the low temperatures believed to be required to freeze-such samples to a glass. Some of the sublimation trials showed erratic results, and these could arise from partial or erratic crystallisation of the solutes, either excipient (laαose) or additives. Contrary to expeαation, the formation of a volatile euteαic solid does not seem to be the mechanism bv which this acceleration is produced. Only in the case of PW formulations is there evidence of loss of salt during drying.

As acceleration is not associated with euteαic solidification there is no reason to use the higher salt concentrations initially studied. As it is considered desirable that little, if any, salt should remain within the final produα, then the sucrose and laαose formulations are less desirable than the PW formulations. Accordingly PW formulations containing 0.1 M ammonium bicarbonate or ammonium formate were chosen to study drug containing samples. Sublimation rate enhancement was found with the inclusion of salts, and in the case of PW / ammonium formate mixtures this was of the order of a 20% reduαion in primary drying time. The dried samples were readily rehydrated, although whether or not dissolution rates are significantly better than freeze-dried excipient mixtures (without salt) requires a quantitative measure rather than a subjeαive decision based on observation of dissolution times.

A full interpretation of the DSC charaαerisation of the dried produαs would require further direα study. From the results found, however, there is a good prognosis for long term ambient temperature stability of the produαs. Such stability can be improved, at a minimum cost by increasing the PW content.

Although the mechanism remains obscure, using the premise that accelerating ice sublimation will accelerated rehydration, it has been possible to make formulations with the desired charaαeristics (particularly the Eletriptan / PW / ammonium formate formulation). Work undertaken prior to this projeα, on acceleration only, has now been published (ref. 8).

These studies seleαed excipients, additives and formulations on a more or less arbitrary basis. We would recommend the extension of these studies to other materials and formulations.

Unexpeαedly, this study has shown that the freeze-drying of certain formulations does not require a full drying cycle and that primary drying alone is sufficient. This is both surprising and of significant potential. From the data presented here, the possibility of producing ambient temperature stable produαs, by such a rapid route, also exists. This combination of - D6 -

formulation and process could, for example halve produαion times for new produαs and thus double produαion capacity of extant equipment. We would recommend that Pfizer take appropriate steps to safeguard access to this development.

CONCLUSION

The Eletriptan / PW/ ammonium formate produα, has the charaαeristics of a stable rapidly dissolving dosage form that is mechanically stable. Although the handling properties are not charaαerised, the analogous Sildenafil formulation is similar. Such produαs could be freeze-dried in a blister pack (the use of plastic containers does not cause significant changes in freeze-drying behaviour) and could be "dispensed" direαly onto the tongue of a neo-natal, paediatric or geriatric patient. Produα collapse is sufficiently rapid to ensure that the medication cannot be unilaterally rejeαed. The ratios of excipient, accelerant and aαive material have not been optimised and thus the prognosis for improvement and extension of these properties to other drug substances and formulations is good.

WO 99/30688 - -.-, PCT/GB98/03747

- 7 -

REFERENCES

1. R.H.M. Hatley. F. Franks. S. Brown. G. Sandhu and M. Gray. Stabilisation of a pharmaceutical drug substance by freeze-drying: a case study. Drug Stability 1. 73-

85 (1996).

2. M.J. Pikal. Freeze-drying of proteins: Process, formulation and stability, ACS Symp. Ser. 567. 120-133 (1994).

3. F. Franks, Effective freeze-drying: A combination of physics, chemistry. engineering and economics, Proc. Inst. Refrigeration 91. 32-39 (1994).

4. A.P. MacKenzie. Non-equilibrium freezing behaviour of aqueous systems. Phil. Trans. Roy. Soc. Ser. B, 278. 167-188 (1977).

5. S. Ablett, M.J. Izzard and P.J. Lillford. Differential scanning calorimetry study of frozen sucrose and glycerol solutions. J. Chem. Soc. Faraday Trans. 88. 789-794 (1992).

6. K. Kasraian and P.P. DeLuca. The effect of tertiary butyl alcohol on the resistance of the dry product layer during primary drying, Pharm. Res. 12, 491-495(1995).

7. Anon. TebolR 99: Tertiary Butyl Alcohol in freeze-drying Applications. Arco Chemical Co., Newtown Square. PA (1996). 8. J.Oesterle, F. Franks and T. Auffret. The influence of tertiary butyl alcohol and volatile salts on the sublimation of ice from frozen sucrose solutions: Implications for freeze-drying. Pharmaceutical Development & Technology 3(2) 175-183 (1998).

9. T. Suzuki and F. Franks; Solid-liquid phase transitions and amorphous states in ternary sucrose-glycine-water mixtures, 3. Chem. Soc. Faraday Trans. 89, 2527-

2537 (1993).

10. E.Yu. Shalaev and A.N. Kanev. Study of the solid liquid state diagram of the water-glycine-sucrose system. Cryobiology 31. 374-382(1994).

1 1. E.Yu. Shalaev. F. Franks and P. Echlin. Crystalline and amorphous phases in the ternary system water-sucrose-sodium chloride. 3. Phys. Chem. 100. 1 144-1 152

(1996). 12 E. Yu. Shalaev and F. Franks, Changes in the physical state of model mixtures during freezing and drying: Impact on product quality, Cryobiology 33, 14-26 (1996).

- Do -

APPENDD

Power-time curves for the analyses described in the text Figure Title

A.1 Enlarged power-time curve for the slow cooling of 5% PW /

0.1 M ammonium bicarbonate. A.2 Enlarged power-time curve for the slow cooling of 5% PW /

0.2 M ammonium bicarbonate. A.3 Partial power-time curve for the warming of frozen 5% PW /

0.2 M ammonium bicarbonate. A.4 Enlarged power-time curve for the cooling of 5% laαose /

0.8 M ammonium bicarbonate. A.5 Enlarged power-time curve for the warming of frozen 5% laαose / 0.8 M ammonium bicarbonate.

A.6 Enlarged power-time curve for the cooling of 5% laαose /

0.1 M ammonium acetate. A.7 Enlarged power-time curve for the cooling of 5% laαose /

0.1 M ammonium formate. A.8 Enlarged power-time curve for the warming of a frozen

Eletriptan.FIBr solution. A.9 Full power-time curve for the warming of a frozen,

Eletriptan.HBr solution. A.10 Enlarged power-time curve for the warming of a frozen Sildenafil solution.

A.11 Power-time curve for the warming of frozen 5% PVP /

0.1 M Ambic / 5 mg/ml Eletriptan.FIBr. A.12 Enlarged power-time curve for the slow cooling of frozen 5% PW /

0.1 M Ambic / 5 mg/ml Eletriptan.HBr. A.13 Enlarged power-time curve for the warming of a slowly cooled 5% PW /

0.1 M Ambic / 5 mg/ml EletriptanHBr solution (Heat RH1). A.14 Enlarged power-time curve for the warming of annealed 5% PW /

0.1 M Ambic / 5 mg/ml Eletriptan.HBr solution. WO 99/30688 - _ -₉ _ PCT/GB98/03747

A.15 Enlarged power-time curve for the warming of 5% PW /

0.1 M ammonium formate / 10 mg/ml EletriptanJHBr solution. A.16 Enlarged power-time curve for the warming of 5% PW /

0.1 M ammonium formate / 5 mg/ml Eletriptan.HBr solution54 A.17 Enlarged power-time curve for the warming of 5% PW /

5 mg/ml Eletriptan.HBr solution. A.18 Enlarged power-time curve for the warming of 5% PW /

0.1 M ammonium formate / 5 mg/ml Sildenafil / 50 mM HQ A.19 Enlarged power-time curve for the warming of PW / 5 mg/ml Sildenafil / 50 mM HQ

A.20 Power-time curve for the first warming of freeze-dried 5% PW /

5 mg/ml Eletriptan.HBr. A.21 Power-time curve for the second warming of freeze-dried 5% PW /

5 mg/ml Eletriptan. HBr. A.22 Power-time curve for the first warming of freeze-dried 5% PW /

5 mg/ml Eletriptan HBr / 0.1 M ammonium bicarbonate. A.23 Power-time curve for the second warming of freeze-dried 5% PW /

5 mg/ml Eletriptan. HBr / 0.1 M ammonium bicarbonate. A.24 Power-time curve for the first warming of freeze-dried 5% PW / 5 mg/ml Eletriptan HBr / 0.1 M ammonium formate.

A.25 Power-time curve for the second warming of freeze-dried 5% PW /

5 mg/ml Eletriptan / 0.1 M ammonium formate. A.26 Power-time curve for the first warming of dried 5% PW /

5 mg/ml Sildenafil / 50 mM HCl (Primary dried). A.27 Power-time curve for the second warming of dried 5% PW /

5 mg/ml Sildenafil / 50 mM HCl (Primary dried). A.28 Power-time curve for the first warming of freeze-dried 5% PW /

5 mg/ml Sildenafil / 50 mM HQ (Fully dried). A.29 Power-time curve for the second warming of freeze-dried 5% PW / 5 mg/ml Sildenafil / 50 mM HCl (Fully dried) .

A.30 Power-time curve for the first warming of dried

5% PW / 5 mg/ml Sildenafil / 0.1 M ammonium formate / 50 mM HCl (Primary dried). A.31 Power-time curve for the second warming of dried

5% PW / 5 mg/ml Sildenafil / 0.1 M ammonium formate /

50 mM HCl (Primary dried). A.32 Power-time curve for the first warming of freeze-dried 5% PW /

5 mg/ml Sildenafil / 0.1 M ammonium formate / 50 mM HQ (Fully dried). A.33 Power-time curve for the second warming of freeze-dried 5% PW /

5 mg/ml Sildenafil / 0.1 M ammonium formate / 50 mM HQ (Fully dried). A.34 Power-time curve for the first warming of primary-dried 5% PW

A.35 Power-time curve for the second warming of primary-dried 5% PW.

Claims

Claims

1. A method of lyophilizing a solution comprising the steps of freezing the solution to a temperature at or below the lower of its eute╬▒ic temperature or its glass transition temperature and, in a first drying stage, removing at least a portion of the solvent by sublimation, chara╬▒erized in that the solution contains an accelerant excipient to enhance the rate of solvent sublimation.

2. A method as claimed in claim 1, wherein the solution comprises a substance capable of forming a glass on freeze concentration.

3. A method as claimed in claims 1 or 2, wherein the solution comprises at least one nonvolatile substance.

4. A method as claimed in any one of claims 1 to 3, wherein the solution comprises a pharmaceutically active substance.

5. A method as claimed in any one of the preceding claims, wherein substantially all of the solvent is removed during the sublimation step.

6. A method as claimed in any one of claims 1 to 4, wherein a portion of the solvent remains after the sublimation step.

7. A method as claimed in claim 6, further comprising a secondary drying stage after the sublimation step, to remove excess solvent.

8. A method as claimed in any one of the preceding claims, wherein the lyophilized solution is substantially stable at ambient temperature and/or pressure.

9. A method as claimed in 8, wherein the glass transition temperature of the lyophilized solution is higher than ambient temperature and/or pressure under desired storage conditions.

10. A method as claimed in any one of the preceding claims, wherein the solvent is aqueous and the solution undergoes ice sublimation during lyophilization.

11. A method as claimed in any one of the preceding claims, wherein the accelerant excipient, in whole or in part, undergoes eute╬▒ic crystallization on cooling.

12. A method as claimed in any one of claims 1 to 10, wherein the accelerant excipient, in whole or in part, is included in the amorphous state.

13. A method as claimed in any one of the preceding claims, wherein the accelerant excipient is at least partly removed during the sublimation step.

14. A method as claimed in claim 13, wherein the accelerant excipient is volatile under the conditions used for sublimation.

15. A method as claimed in any one of claims 1-12, wherein the accelerant excipient is not removed during the sublimation step.

16. A method as claimed in any one of the preceding claims, wherein the accelerant excipient is an inorganic salt.

17. A method as claimed in claim 16, wherein the accelerant excipient is an ammonium salt of a weak or poorly dissociated acid.

18. A method as claimed in claim 17, wherein the accelerant excipient is sele╬▒ed from the group consisting of ammonium formate, ammonium acetate and ammonium bicarbonate.

19. A method as claimed in any one of the preceding claims, wherein the accelerant excipient is present in combination with at least one compound seleαed from the group consisting of sucrose, PW and laαose. WO 99/30688 - . _ω . PCT/GB98/03747

20. A method of lyophilizing a solution for storage comprising the steps of freezing the solution to a temperature at or below the lower of its eute╬▒ic temperature or its glass transition temperature and removing solvent by sublimation, chara╬▒erized in that the solution comprises a substance capable of raising the glass transition temperature of the lyophilized solution above the ambient storage temperature.

21. A method as claimed in claim 20, wherein no additional drying step is required after the sublimation step.

22. A m╬▒hod as claimed in claims 20 or 21, wherein the substance is capable of forming a matrix to support the amorphous phase during sublimation.

23. A method as claimed in any one of claims 20 to 22, wherein the solution is aqueous and the solution undergoes ice sublimation during lyophilization.

24. A method as claimed in claim 23, wherein the solution contains a substance capable of forming a colloidal solution in water.

25. A method as claimed in any one of claims 20 to 24, wherein the solution contains a hydrophilic polymer.

26. A method as claimed in any one of claims 20 to 25, wherein the solution contains polyviπylpyrrolidone.

27. A method of lyophilizing an aqueous solution of a glass-forming substance comprising the steps of freezing the solution to a temperature at or below the lower of its eute╬▒ic temperature or its glass transition temperature and, in a first drying stage, removing at least a portion of solvent by sublimation, chara╬▒erized in that the solution contains an ammonium salt sele╬▒ed from the group consisting of ammonium formate, ammonium acetate and ammonium bicarbonate.

28. A method as claimed in claim 24, wherein the solution further comprises a bulking agent sele╬▒ed from the group consisting of polyvinylpyrrolidone, sucrose and la╬▒ose. WO 99/30688 - . ₆₄ . PCT/GB98/03747

29. A m╬▒hod of manufacturing a rapidly-dissolving dosage form of a pharmaceutically- a╬▒ive compound, which includes the step of fyophilizing a solution of the compound by a method according to any one of the preceding claims.

30. A m╬▒hod as claimed in claim 29, wherein the solution contains an accelerant excipient that is pharmalogically acceptable.

31. A method as claimed in claims 29 or 30, wherein the accelerant excipient is volatile.

32. A method as claimed in claim 31, wherein substantially all of the accelerant excipient is removed during the sublimation step.

33. A method as claimed in any one of claims 29 to 32, wherein the solution further comprises a bulking agent.

34. A method as claimed in claim 31, wherein the bulking agent is sele╬▒ed from the group consisting of sucrose, la╬▒ose and polyvinylpyrrolidone.

35. A method as claimed in any one of claims 29 to 32, further comprising the step of processing the lyophilized solution into the desired dosage form.

36. A method of preparing a sterile solution of a drug for oral or parenteral administration comprising adding sterile water to a sterile rapidly dissolving dosage form obtained by a method as claimed in any one of claims 29 to 35.

37. A method as claimed in any one of the preceding claims, wherein the solution is frozen to a temperature just below or about its glass transition temperature or its eutectic temperature.

38. A method as claimed in claim 37, wherein the solution is frozen at a temperature of within 15┬░C of the lower of its eute╬▒ic temperature and its glass transition temperature, preferably within 10┬░C, and most preferably within 5┬░C.

39. A lyophilized produ╬▒ obtainable by a method as claimed in any one of the preceding claims.

40. A lyophilized produ╬▒ as claimed in claim 39 being rapidly dissolvable.

41. A lyophilized produ╬▒ as claimed in claims 39 or 40, wherein the produ╬▒ does not require additional drying steps.

42. A lyophilized produ╬▒ as claimed in claim 41, wherein the glass transition temperature of the produ╬▒ is higher than ambient storage temperatures, preferably by at least 10┬░C.

43. A lyophilized produ╬▒ as claimed in any one of claims 39, 40 or 41, being a pharmaceutical composition.

44. A pharmaceutical composition as claimed in claim 43, comprising an a╬▒ive agent sele╬▒ed from the group consisting of Eletriptan, Sildenafil and Viagra.

45. A pharmaceutical composition as claimed in claim 44, further comprising at least one excipient sele╬▒ed from the group consisting of polyvinylpyrrolidone, sucrose, la╬▒ose, ammonium bicarbonate and ammonium acetate.