`
`European Journal of Pharmaceutical Sciences 15 (2002) 115-133
`
`E U R O P I A N J O U R N A L O F
`
`PMEimmi
`Stlliffl
`
`www.elsevier.nl / locate / ejps
`
`Review
`Practical aspects of lyophilization using non-aqueous co-solvent systems
`Dirk L. Teagarden*, David S. Baker
`Department of Sterile Products Development. Pharmacia Corporation, 7000 Portage Road, Kalamazoo, MI 49001-0199, USA
`
`Received 25 June 2001; received in revised form 5 November 2001; accepted 11 November 2001
`
`Abstract
`
`Non-aqueous co-solvent systems have been evaluated for their potential use in the freeze-drying of pharmaceutical products. The
`advantages of using these non-aqueous solvent systems include: increased drug wetting or solubility, increased sublimation rates,
`increased pre-dried bulk solution or dried product stability, decreased reconstitution time, and enhancement of sterility assurance of the
`pre-dried bulk
`solution.
`
`Conversely, the potential disadvantages and issues which must be evaluated include:
`the proper safe handling and
`storage of flammable and/or
`
`explosive solvents,
`the special facilities or equipment which may be required,
`the control of residual solvent
`levels, the toxicity of the remaining solvent, qualification of an appropriate GMP purity, the overall cost benefit to use of the solvent, and
`the potential increased regulatory scrutiny. The co-solvent system that has been most extensively evaluated was the re/t-butanol/water
`combination. The /ert-butanol possesses a high vapor pressure, freezes completely in most commercial freeze-dryers, readily sublimes
`during primary drying, can increase sublimation rates, and has low toxicity. This
`co-solvent
`
`system been used in the manufacture of a has
`
`marketed
`injectable pharmaceutical
`
`product. When using this solvent system, both formulation and process control required optimization
`to maximize drying rates and to minimize residual solvent levels at the end of drying. Other co-solvent systems which do not freeze
`completely
`in commercial
`
`freeze-dryers were more difficult to use and often resulted
`in unacceptable
`freeze-dried
`
`cakes. use appears Their
`
`limited to levels of not more than 10%. © 2002 Elsevier Science B.V. All rights reserved.
`
`Keywords: Non-aqueous; Solvent; Freeze-drying; rerr-Butanol
`
`1. Introduction
`
`solutions to produce an
`Freeze drying of pharmaceutical
`elegant stable powder has been a standard practice em
`ployed to manufacture many marketed pharmaceutical
`injectable products. The overwhelming majority of these
`products are lyophilized from simple aqueous solutions.
`Water is typically the only solvent of significant quantity
`that is present which must be removed from the solution
`via the freeze-drying process. However, it is not unusual
`for small quantities of organic solvents to be present in
`either the active pharmaceutical ingredient or one of the
`excipients. These low levels of organic solvent are com
`monly found because they may be carried through as part
`of the manufacture of these individual components since
`the ingredient may be precipitated, crystallized, or spray
`dried from organic solvents. Therefore, most freeze-dried
`products may be dried from solutions which contain low
`levels of organic solvents. However, in addition to freeze-
`
`*Correspondmg author. Tel.: +1-616-833-6649: fax: +1-616-833-
`6935.
`E-mail address: dirk.l.teagarden@pharmacia.com (D.L. Teagarden).
`
`drying solutions with organic solvent impurities present,
`there may be instances where freeze-drying from organic
`solvents or mixtures of water and organic solvent may
`offer the formulation scientist advantages over simply
`drying from an aqueous solution. An example of at least
`one pharmaceutical product on the market which has
`utilized an organic co-solvent system during freeze-drying
`is CAVERJECT8 Sterile Powder (Teagarden et al.,
`1998a,b). This particular product has been successfully
`manufactured by freeze-drying from a 20% v/v tert-
`butanol/water co-solvent system.
`There are many reasons why it may be beneficial to both
`product quality and process optimization to select a
`lyophilization process which employs a strictly organic or
`organic/water co-solvent system. A list of some of these
`potential advantages
`
`includes: increases rate of sublimation
`and decreases drying time, increases chemical stability of
`the pre-dried bulk solution, increases chemical stability of
`the dried product, facilitates manufacture of bulk solution
`by increasing drug wettability and solubility in solution,
`improves reconstitution characteristics (e.g. decreases re-
`constitution time), and enhances sterility assurance for
`pre-dried bulk solution. However, the development sci-
`
`- see front matter © 2002 Elsevier Science B.V. All rights reserved.
`0928-0987/02/$
`PI1: S0928-0987(01 )00221-4
`
`AGILA ET AL - EXHIBIT 1006
`
`
`
`116
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences 15 (2002) 115—133
`
`entist must be aware that use of these organic/water
`co-solvent systems can cause a multitude of problems. A
`list of some of these potential disadvantages includes:
`toxicity concerns, operator safety concerns due to high
`degree of flammability or explosion potential, lack of
`compendia grades or monographs, may require special
`manufacturing facilities/equipment or storage areas, pos
`sess difficult handling properties, requires high purity
`solvent with known impurities at low levels, must reach
`acceptable residual solvent in final product, high cost to
`use, potential for splash/spattering of product in vial neck,
`and lack of regulatory familiarity. One should remember
`that successful sterile formulations should always employ
`an understanding of the fundamental interrelationships
`between the formulation, the process, and the package. The
`knowledge gained from the interrelationships enables
`optimization of the formulation which can be processed
`and packaged at a production scale. These same principles
`still apply to the use of organic solvents in freeze-drying.
`The advantages and disadvantages of their use must be
`carefully weighed before they are chosen to be used in the
`manufacture of a pharmaceutical product, especially one
`that is an injectable dosage form.
`A list of some of the solvents which have been tested in
`freeze-drying studies is provided in Table 1. Included in
`this table is a summary of some of the critical physical/
`chemical properties for each of these solvents. Several
`pharmaceutical products or drugs in various stages of
`formulation and/or clinical development have been manu
`
`factured via a process which required freeze-drying from
`organic co-solvent systems. These types of solvent systems
`were chosen for many of the reasons described above.
`Table 2 contains a list of examples of a few drug
`preparations which have been evaluated.
`Additional uses for the technique of freeze-drying from
`organic co-solvent systems, other than in the manufacture
`of pharmaceuticals, include the preparation of biological
`specimens or the preparative isolation of lecithin. The
`biological specimens can be prepared by lyophilization
`from organic co-solvent systems in order to improve
`specimen preservation for scanning electron microscopy
`examinations (Inoue and Osatake, 1988; Akahori et al.,
`1988; Hojo, 1996). ferf-Butanol appears to be the major
`organic solvent selected for this use. The surface structure
`of the specimen remains intact when employing rapid
`freezing followed by freeze-drying from an appropriate
`organic solvent such as ferf-butanol (Herman and Miiller,
`1997). Lecithin can be prepared in a solvent-free form via
`lyophilization from cyclohexane (Radin, 1978).
`
`2. Facilitating manufacture of bulk solution
`
`The first step in the manufacture of almost all freeze-
`dried products is the formation of a solution of the
`ingredients to be dried. Typically these solutions are sterile
`filtered, aseptically filled into containers, and freeze-dried.
`Some hydrophobic ingredients (e.g. the bulk drug or
`
`Table 1
`Properties of organic solvents evaluated in freeze-drying
`
`Solvent
`
`Solubility Vapor
`in water
`pressure
`(mmHg
`{%*)
`at 20 0C)
`
`Freezing Boiling
`point
`point
`(0C)
`(0C)
`
`Flammability
`
`Flash point
`(0F/0C)
`
`Autoignition Lower flamm.
`temperature
`limit (in air vol.%)
`(0F/0C)
`
`Upper flamm.
`limit (in air vol.%)
`
`tert-Butanol
`Ethanol
`//-Propanol
`ra-Butanol
`Isopropanol
`Ethyl acetate
`Dimethyl carbonate
`Acetonitrile
`Dichloromethane
`Methyl ethyl ketone
`Methyl isobutyl ketone
`Acetone
`1-Pentanol
`Methyl acetate
`Methanol
`Carbon tetrachloride
`Dimethyl sulfoxide
`Hexafluoroacetone
`Chlorobutanol
`Dimethyl sulfone
`Acetic acid
`Cyclohexane
`
`a 100% = miscible.
`b 25 0C.
`
`100
`100
`100
`7.7
`100
`8.7
`9.5
`100
`1.3
`27.0
`2.0
`100
`2.7
`25
`100
`0.08
`100
`100
`0.8
`100
`100
`0.008
`
`26.8
`41.0
`14.5
`5.6
`31.0
`64.7
`72
`69.8
`343.9
`76.2
`5.1
`160.5
`1.8
`148.7
`87.9
`78.9
`0.5
`5.0b
`
`11.6
`66.4
`
`24.0
`114
`127
`90
`89.5
`84
`2
`48
`97
`87
`80
`94
`78
`98
`98
`23
`18.4
`129
`97
`107
`16.2
`6.5
`
`82
`78.5
`97.1
`117.5
`81
`77.1
`90
`80.1
`40
`79.6
`117
`56.2
`138
`57
`65
`76
`189
`26
`167
`248
`118.5
`81
`
`52/11
`62/16
`59/15
`95/35
`53.6/12
`24/-4
`65/18
`45/8
`None
`26/-3
`56/13
`1/ —17
`120/49
`15/ — 9
`52/11
`None
`188/87
`None
`>212/ > 100
`290/143
`103/39
`- 1 / - 1 8
`
`892/478
`793/423
`760/404
`689/365
`750/398
`800/426
`
`975/524
`1033/556
`885/474
`860/460
`1000/538
`572/300
`935/502
`835/446
`None
`572/300
`None
`
`2.4
`3.3
`2.1
`1.4
`2.5
`2.2
`4.2
`4.4
`14
`1.7
`1.2
`2.6
`1.2
`3.1
`6.0
`None
`3.5
`None
`
`960/516
`500/260
`
`6.6
`1
`
`8.0
`19
`13.5
`11.2
`12
`11.5
`12.9
`16.0
`22
`10.1
`8
`12.8
`10
`16
`36
`None
`42
`None
`
`19.3
`9
`
`
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences IS (2002) 1IS—133
`
`117
`
`Table 2
`Examples of drug preparations freeze-dried from co-solvents
`
`Drug
`
`Alprostadil (CAVERJECT® S.Po.)
`Aplidine
`Amoxicillin sodium
`Gentamicin sulfate
`A^-Cyclodexyl-A^-methyl-4-(2-oxo-
`1,2,3,5- tetrahy droimidazo- [2,1 -b]
`quinazolin-7-yl)oxybutyramide with
`ascorbic acid
`Cyclohexane-l,2-diamine Pt(II)
`complex
`Annamycin
`
`Cephalothin sodium
`Cephalothin sodium
`
`Prednisolone acetate /polyglycolic
`acid
`Gabexate mesylate
`Piraubidin hydrochloride
`Progesterone, coronene, fluasterone,
`phenytoin
`Fructose-1,6-diphosphate
`Poly(lactide-co-glycolide)
`Dioleoylphosphatidylcholine and
`dioleoylphophatidylglycerol
`Vecuroniumbromide
`Bovine pancreatic trypsin inhibitor
`
`Co-solvent system
`
`20% v/v terf-butanol/water
`40% v/v terf-butanol/water
`20% v/v terf-butanol/water
`terf-Butanol/water
`50% v/v terf-butanol/water
`
`terf-Butanol
`
`sulfoxide/
`
`terf-Butanol/dimethyl
`water
`5% w/w isopropyl alcohol/water
`4% ethanol, 4% methanol or 4%
`acetone/water
`Carbon tetrachloride/
`hexafluoroacetone sesquihydrate
`Ethanol/water
`Ethanol/water
`Chlorobutanol hemihydrate / Dimethyl
`sulfone
`terf-Butanol/water
`Acetic acid
`Cyclohexane
`
`Acetonitrile
`Dimethyl sulfoxide/1% water
`
`Reference
`
`Teagarden et al, 1998a
`Nuijen et al., 2000
`Tico Grau et al., 1988
`Baldi et al., 1994
`Benjamin and Visor, 1989
`
`Tanno et al., 1990
`
`Zou et al, 1999
`
`Koyama et al, 1988
`Cise and Roy, 1981
`
`DeLuca et al, 1989a
`
`Kami jo et al., 1987
`Kaneko et al., 1993
`Tesconi et al, 1999
`
`Sullivan and Marangos, 1998
`Meredith et al., 1996
`Feigner and Eppstein, 1991
`
`Jansen, 1997
`Desai and Klibanov, 1995
`
`excipients) may be difficult to wet and get into solution or
`may require large amounts of water to adequately solubil-
`ize. The use of organic co-solvents can greatly facilitate
`the wetting of the hydrophobic substance, decrease the
`time to achieve a solution or uniform dispersion, and
`decrease the amount of solvent which needs to be removed
`during the drying process. All of these attributes can
`potentially have a positive effect on the consistency and
`ease of product manufacture. Several examples of this
`increased drug solubility in the presence of organic co-
`solvents targeted for lyophilization include: (1) alprostadil
`formulated in a ferf-butanol/ water solution (Teagarden et
`al., 1998a) and (2) cardiotonic phosphodiesterase inhibitors
`complexed with vitamins formulated in a ferf-butanol/
`water solution, however, other alcohols such as ethanol,
`w-propanol, or isopropyl alcohol are also claimed to
`provide further increases in solubility (Benjamin and
`Visor, 1989) and (3) aplidine formulated in ferf-butanol/
`water solution (Nuijen et al., 2000). The actual ferf-
`butanol concentration (i.e. 40% v/v in water) selected for
`aplidine produced a greater than 40-fold increase in
`solubility compared to that in pure water.
`
`3. Stabilization of bulk solution
`
`A major challenge in developing a sterile injectable
`product can be its instability in solution. Most freeze-dried
`products are developed as this dosage form in order to
`
`circumvent poor stability. The manufacture of a freeze-
`dried product necessitates that the product is usually first
`manufactured as a solution, filtered to sterilize, aseptically
`filled, and finally lyophilized to remove the solvents. All of
`these unit operations require that the product be held in the
`solution state for a defined period of time. However, as the
`product is held in the solution phase it can experience
`various levels of degradation which are dependent on the
`kinetics of the degradation mechanism. The presence of the
`various levels of organic solvent can have a profound
`effect on the chemical stability. Those drug candidates
`which are very labile in aqueous solutions may require the
`added stability to achieve an acceptable level of degra
`dation during manufacture. Early efforts to freeze-dry an
`anti-neoplastic
`agent
`(l,3-bis(2-chloroethyl)-l-nitroso-
`urea) from an ethanol/water solution were initiated be
`cause of the rapid degradation in aqueous solution and
`improved solution stability in ethanol/water solutions
`(Flamberg et al., 1970). Unfortunately freeze-drying this
`product in the ethanol/water co-solvent system proved to
`be unsuccessful due to potency losses and unacceptable
`clarity. Flamberg et al. suggested that an alternative
`process to freeze-drying solvent systems containing etha
`nol would be to use low temperature vacuum drying.
`However, alprostadil has been successfully freeze-dried
`from a ferf-butanol/water solution. The first-order degra
`dation rate constant of alprostadil in 20% v/v ferf-butanol/
`water (k=0.0011 day 1 at 25 0C) was significantly reduced
`compared to water buffered at the same pH value (k=
`
`
`
`118
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences 15 (2002) 115—133
`
`at 25 0C). These data are consistent with the
`0.0041 day
`claims of extraordinary stability of prostaglandins in tert-
`butanol (Monkhouse, 1975). This decreased degradation
`rate enables the manufacturing unit operations to be
`performed at ambient conditions without requiring cooling
`of the solution during manufacture. Additionally, it adds
`flexibility in scheduling these various operations because
`the solution degradation is now minimized.
`The formulation of trecetilide fumarate, a sterile inject
`able in clinical development for treatment of arrhythmias,
`also involved freeze-drying from a ferf-butanol/water
`mixture (Baker, 1998). Kinetic analysis showed solution
`degradation occurred by a process of defluorination
`through SNj substitution and K elimination, both proceed
`ing through the same carbonium ion intermediate. Since
`factors such as ionic strength, buffer type, solution pH, and
`drug and buffer concentrations did not significantly affect
`degradation rate, destabilization of the fluoride leaving
`group was one of the few methods left to control this
`reaction. Use of tertiary butyl alcohol as a co-solvent
`slowed solution state degradation by a factor of approxi
`mately 4-5. This significantly increased the probability of
`being able to scale up the manufacturing process while
`maintaining tight control of the level of degradation. The
`rate constant (k) for drug degradation was decreased
`substantially as the ferf-butanol content was increased. The
`work required to separate two charges to infinite distance
`is related to the function (1 —1/Er) where Er is the relative
`permittivity of the medium. The linear relationship ob
`served between log k and this function (Fig. 1) indicates
`that the decreased relative permittivity of the solvent
`system (i.e. the increased work required to remove the
`fluoride group) was the major effect for the improved
`
`solution state stability of trecetilide fumarate. The de
`creased ability of ferf-butanol (relative to water) to solvate
`and stabilize the two ions appeared to be less of a factor.
`The use of ferf-butanol allowed formulation and filling on
`a production scale over a 24-h period for this compound.
`The resulting freeze-dried product was predicted to have
`an acceptable shelf-life of at least 2 years at ambient
`temperature. This was a dramatic improvement compared
`to a frozen aqueous solution which had to be stored at —80
`0C and required use within 3 h of thawing and admixture
`preparation. This type of effect would be expected to be
`observed for many other drug products which are degraded
`in the presence of water.
`
`4. Impact on the freeze-drying process
`
`4.1. Effect on freezing
`
`The first stage of freeze-drying involves freezing the
`solution to remove solvent (typically water) from the drug
`and excipients through the formation of ice. The resulting
`semi-frozen system is cooled further to transform all
`components into a frozen state. A selected time/tempera
`ture profile is achieved by placing the solution, which is
`commonly held in glass vials or syringes, onto cooled
`shelves. Suspended impurities in the solution or imperfec
`tions in the walls of the container initiate heterogeneous
`nucleation during freezing. This event almost always
`involves supercooling where upon crystallization occurs
`below the equilibrium freezing point of the solution.
`Consequently, when freezing does occur, crystal growth
`tends to be rapid and results in a complex mixture of
`
`y = 260.67x-258.74
`R2 = 0.9847
`
`•
`
`•
`
`-1.1
`
`c -1.3
`03
`(f)
`c
`O -1-5
`G)
`C O
`*-1.7
`
`CO
`"§ -1.9
`CD
`0) a -2.1 -
`
`Ui
`
`-2.3
`
`•
`
`-2.5
`0.9835
`
`0.9840
`
`0.9845
`
`0.9850
`
`0.9860
`
`0.9865
`
`0.9870
`
`0.9875
`
`0.9855
`(1-1/Er)
`
`Fig. 1. Effect of relative permittivity (Er) on the solution degradation rate constant for trecetilide fumarate, a compound undergoing SN1 substitution and
`E1 elimination through the same carbonium ion intermediate.
`
`
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences 15 (2002) 115—133
`
`119
`
`crystalline, amorphous and metastable materials. The
`impact of the presence of organic solvents on the various
`phases of freeze-drying has been discussed in detail
`(Seager et al., 1978, 1985; Seager, 1978, 1979a,b). Not
`surprisingly, the type and concentration of the organic
`solvent present affects the freezing characteristics of the
`solution prior to initiation of drying. The resulting frozen
`or semi-frozen solution significantly impacts the crystal
`habit of the ice, the drying rates, the collapse temperatures,
`the appearance of the dried cake, the surface area of the
`dried cake, and reconstitution properties, etc. The choice of
`solvent can also affect the degree of crystallinity of the
`drug. It has been demonstrated that incorporation of
`isopropyl alcohol readily results in highly crystalline
`cefazolin sodium (Koyama et al., 1988). Scaling up this
`process required incorporation of a heat treatment step to
`insure complete crystallization of the drug. Conversely, use
`of co-solvents can sometimes have deleterious effects
`during freezing. The use of volatile organic solvents has
`been reported to result in drug precipitation in the latter
`parts of freezing due to solvent evaporation. This can lead
`to an increase in drug concentration above its saturation
`level (Seager, 1979b). Care should be taken to select
`excipient concentrations such as buffer salts so that they do
`not exceed their saturation solubility. This is particularly
`important for phosphate buffers since they have very low
`solubility products with certain cations such as aluminum,
`calcium or iron (Hasegawa et al., 1982a,b,c, 1983). As a
`result, salt precipitation can produce a haze upon reconsti
`tution. This problem can be exacerbated in co-solvent
`systems due to the decreased solubility and higher associa
`tion constants for such systems.
`The size and shape of the ice crystals has been found to
`vary with different organic solvents. The presence of high
`melting point solvents such as ferf-butanol results in
`solvent crystallizing between the ice matrix as the tempera
`ture is decreased. The presence of the ferf-butanol altered
`the crystal habit of the ice as it formed. The size of the ice
`crystals (i.e. large vs. fine) changed depending on the
`quantity of ferf-butanol present in the system. Thermal
`analysis studies (via Differential Scanning Calorimetry and
`freeze-dry microscopy) have been used to evaluate the
`various stable and metastable states which form for ferf-
`butanol/water systems during freezing (Kasraian and
`DeLuca, 1995a). The DSC warming thermograms for the
`ferf-butanol/water mixtures are illustrated in Fig. 2. The
`authors were able to apply various annealing techniques to
`eliminate the metastable states and were able to construct
`the true phase diagram (Fig. 3). Although this phase
`diagram agreed well with other ferf-butanol/water phase
`diagrams reported in the literature (Ott et al., 1979;
`Woznyj and Liidemann, 1985), it was claimed that the
`slight differences could be explained by the presence of
`metastable events which thermal treatments eliminated.
`These data suggested that ferf-butanol levels in the range
`of 3-19% caused the ice to form needle-shaped crystals.
`
`they
`As these large needle-shaped crystals sublimed
`created a more porous, less resistant matrix, which facili
`tates drying. Other solvents such as dimethyl carbonate
`also appear to freeze under production freeze-dryer con
`ditions producing eutectic solids, solid organic solvent
`mixed with ice, drug, and excipients. However, most of the
`organic solvents investigated (Seager et al., 1985) such as
`methanol, ethanol, w-propanol, w-butanol, acetonitrile,
`methyl ethyl ketone, dichloromethane, and methyl isobutyl
`ketone do not freeze in typical commercial freeze-dryers
`but remain as liquid residues within the frozen matrix.
`Solutions containing 8% ethyl acetate, 10% dimethyl
`carbonate, or 10% w-butanol appeared to dry rapidly.
`Solutions containing 10% ethanol, 10% w-propanol, or
`10% methanol appeared to dry slowly. Solutions con
`taining up to 20% ethanol experienced collapsed cakes and
`were near impossible to dry (Seager et al., 1985). Some of
`the more hydrophilic solvents such as ethanol and metha
`nol retained significant amounts of associated water which
`only partially froze as the temperature decreased. In those
`systems which completely froze (e.g. fert-butanol) the ice
`and frozen solvent grew upwards until reaching the solid
`surface and formed a eutectic skin. Hydrophilic solvents
`which retained large volumes of water formed thick liquid
`skins containing ice whereas less hydrophilic solvents
`containing less water formed thinner skins with less ice.
`The samples being dried need to be protected from radiant
`heat in order to prevent temperature fluctuations and non
`uniform drying. This is especially important since the heat
`of sublimation is significantly lower for the organic co-
`solvents compared to ice (Willemer, 1975). It should also
`be noted that the time between filling the co-solvent
`solution and the freezing of this solution should be
`carefully controlled. The volatility of the organic portion
`of the solution can be such that a significant portion of the
`organic solvent can be lost due to evaporation. One should
`be aware of the potential for a reflux type phenomenon
`when using highly volatile solvents such as ferf-butanol.
`This situation can happen when the evaporating tert-
`butanol condenses near the top of the vial and forms a
`stream of solvent returning to the solution. The dissolved
`substances in the solution can diffuse in this stream. After
`freeze-drying has been completed, the vial can contain
`spots of powder near the neck of the vial. The presence of
`dried powder near the neck of a vial is not desired because
`of both a poor appearance and the possibility of negatively
`impacting the seal with the rubber closure. This problem
`can be decreased by shortening the time period between
`the filling and the freezing of the solution.
`
`4.2. Acceleration of sublimation rate
`
`The freeze-drying process is a unit operation which
`typically involves a long and expensive process. Improve
`ments in the rate of mass transfer of solvent through the
`partially dried cake layer will increase the rate of sublima-
`
`
`
`120
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences 15 (2002) 115—133
`
`(a)
`
`I
`
`£
`
`-10
`
`-5
`
`0
`
`S
`
`Tcmpefature ("CJ
`
`D
`
`C
`
`(c)
`
`JK
`*o
`c
`UJ
`
`A
`
`B
`
`(b)
`
`g
`w
`JS
`
`"P
`tu
`
`-10
`
`(d)
`
`g
`I
`-s
`i5
`
`*5
`0
`Temperature fC)
`
`S
`
`-20
`
`-10
`
`0
`
`10
`
`-10
`
`-5
`
`0
`
`s
`
`Temperature fCJ
`
`Teroperatwc ("CJ
`
`(c) 50% w/w tert-butanol;
`(b) 20% w/w tert-butanol;
`Fig. 2. DSC warming thermograms for tert-butanol/water mixtures, (a) 15% w/w tert-butanol;
`Endotherm A: melting of metastable eutectic; Exotherm B: recrystallization of metastable eutectic form to stable form; Endotherm C: melting of eutectic;
`Endotherm D: melting of tert-butanol hydrate; (d) thermal treatment of 50% w/w solution at —7 0C to eliminate metastable states. (Kasraian and DeLuca,
`1995a).
`
`tion and hence decrease the time for the primary drying
`phase of the freeze-dry cycle. Mathematically it has been
`shown that:
`
`Sublimation rate = pressure difference/resistance
`
`The resistance term is an additive term which reflects
`the sum of the dried product resistance, the vial/stopper
`resistance, and the chamber resistance. However, the
`predominant resistance of the three terms is typically the
`product resistance which usually accounts for about 90%
`of the total resistance (Pikal, 1990). The mass transfer in
`the dry layer occurs via two general mechanisms: bulk
`flow (the movement of material in the direction of a
`pressure gradient, which may be molecular or viscous) or
`diffusive flow
`(the movement of material by molecular
`motion from higher concentration to lower concentration
`or partial pressure) (Nail and Gatlin, 1993). Typically the
`resistance to mass transfer increases with cake depth. What
`happens during the freezing phase can have a profound
`impact on the resistance of the dried cake to mass transfer
`
`during primary drying. Those co-solvents which can alter
`the ice crystal habit and size of the ice crystals, such as
`promoting the formation of large needle crystals, can
`dramatically increase the rate of bulk flow. This is because
`the permeability of the dried layer would increase pro
`portionally (to the third power for molecular flow and to
`the fourth power for viscous flow) to the average diameter
`of the pore created by the sublimation process (Nail and
`Gatlin, 1993). Many of the co-solvents selected for freeze-
`drying increase sublimation rate because they have higher
`vapor pressures than water and hence an expected larger
`driving force for sublimation because the latter depends on
`this vapor pressure difference (Wittaya-areekul, 1999).
`The potential acceleration of the freeze-drying rates of
`aqueous solutions of lactose and sucrose with 5 and 10%
`aqueous solutions of ferf-butanol has been studied (De
`Luca et al., 1989b). It was found that both lactose and
`sucrose solutions could be successfully freeze-dried in the
`presence of ferf-butanol at considerably higher shelf
`temperatures than corresponding aqueous solutions of
`either lactose or sucrose. The drying rates were signifi-
`
`
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences IS (2002) 1IS—133
`
`121
`
`11
`
`30
`
`25 -
`
`20 -
`
`15 -
`
`10 -
`
`o
`<D u
`Hi & a (L>
`
`&
`
`ice 4- l.s.
`
`"W
`
`ice +
`" eutectic A
`
`Eutectic A
`
`hydrate (s) +Ls.
`
`•w
`
`hydrate +
`eutectic A
`
`hydrate (s) + K";
`
`TEA + Ls.
`
`&
`
`. $4
`Eutectic B
`
`hydrate +
`eutectic B
`
`TBA +
`Eutectic B
`
`0
`
`10
`
`o
`
`10
`
`20
`70
`x=d.057
`x=0.36
`TBA Concentration (% w/w)
`Fig. 3. Phase diagram of the tert-buta.no\ (TBA)/water system plotted on % (w/w) basis. Mole fractions are listed for Eutectic A and B and the pure
`hydrate, 'x' is te/t-butanol
`concentration on mol fraction basis (Kasraian and DeLuca, 1995a).
`
`30
`
`40
`
`50
`
`60
`
`80
`
`90
`x=0.68
`
`100
`
`cantly increased when using the fert-butanol as a co-
`solvent. The drying times were decreased to approximately
`half the time when drying sucrose in the presence of
`ferf-butanol. The collapse temperature for the frozen
`solutions appeared to increase when ferf-butanol was
`present. The ferf-butanol readily froze and remained frozen
`during the primary drying phase. The ferf-butanol sub
`limed during primary drying and created a porous structure
`which facilitated the mass transfer of water vapor due to
`decreased cake resistance. The resulting dried cakes ex
`hibited significantly increased surface areas. Freeze-drying
`of a similar amorphous carbohydrate such as a lactose base
`formulation from a ferf-butanol/water co-solvent (e.g.
`CAVERJECT Sterile Powder) also produces a very porous
`cake structure as illustrated by the SEM shown in Fig. 4.
`Alternatively, it was demonstrated that using other organic
`co-solvents at a 5% level (e.g. methanol, ethanol, iso-
`propanol, acetone, w-butanol, or dioxane) that do not freeze
`under operating conditions for conventional commercial
`freeze-dryers produced unacceptable freeze-dried cakes of
`either lactose or sucrose due to boiling of solvent and cake
`collapse.
`A more in-depth study was later completed evaluating
`the use of ferf-butanol as a mass transfer accelerator during
`
`the freeze-drying of a 5% w/v sucrose solution (Kasraian,
`1994; Kasraian and DeLuca, 1995b). Again it was demon
`strated that the primary drying phase (i.e. sublimation)
`proceeded more rapidly when 5% w/w ferf-butanol was
`present, thereby resulting in an approximately 10-fold
`reduction in drying time. This increased drying rate was
`caused by the formation of needle-shaped ice crystals
`which dramatically lowered the product resistance of the
`dried cake. The resulting surface area of the dried cake
`increased by approximately 13-fold when the 5% ferf-
`butanol was used. The presence of the ferf-butanol did not
`impact the collapse temperature, however, the rapid subli
`mation prevented the product from reaching the collapse
`temperature. The rationale for this was postulated to be
`because the water content in the partially dried layer
`decreased faster in the presence of the ferf-butanol, which
`resulted in an increased viscosity and thereby prevented
`collapse. The rate of sublimation of both the water and the
`ferf-butanol was impacted by the ratio of the two solvents.
`Water appeared to sublime faster at ratios of less than 20%
`w/w ferf-butanol/water. ferf-Butanol appeared to sublime
`faster at ratios of greater than 20% w/w ferf-butanol/
`water. Both solvents sublimed at equal rates at 20% w/w
`ferf-butanol/water. The latter data suggested a strong
`
`
`
`122
`
`D.L. Teagarden, D.S. Baker / European Journal of Pharmaceutical Sciences IS (2002) 1IS—133
`
`5
`Det WD
`sCC.V Spot Magn
`.00kV4.0
`lOOOx GSE 9.9
`£
`
`20 (im
`1.2 Torr Caverject Sterile Powder
`
`I
`
`Fig. 4. SEM picture of CAVERJECT Sterile Powder which has been lyophilized from a 20% v/v terf-butanol/water
`
`solution.
`
`association at this concentration. These data are consistent
`with the sublimation of water and ferf-butanol from the
`frozen matrix of CAVERJECT Sterile Powder during
`freeze-drying as illustrated in Fig. 5.
`Mixtures of ferf-butanol/water were also used to in
`crease the rate of freeze-drying for solutions for a model
`
`drug, gentamicin sulfate (Baldi et al., 1994). Simple
`freeze-drying of the gentamicin with the co-solvent system
`was insufficient to produce an acceptable cake. The
`investigators had to apply a statistical experimental design
`to achieve the final formulation. All components of the
`formulation studied, i.e. the active ingredient, ferf-butanol/
`
`20
`
`15
`
`10
`
`-o
`H •w
`fa
`s
`m
`Vl