`
`Catalog record is available from thé Library of Congress
`
`'n-ademark Notice: Producr _o_: -jc::_n_-p9;aie-
`identification and explmmtion3-Wit1:pi.it-gijfileijt-‘ig--tnfiihgéo.
`
`be-
`
`'C{'§O0l CRC Press
`
`'
`
`-'
`
`U.S. Government works
`. "Book Number |—5Tr'49[—I20-l
`.c.If.Amen'ca
`5 6 7 8 9 0
`_ on acid-fmee paper
`
`Exh. 1014
`
`
`
`Preformulation as an
`Aid to Product Design in
`Early Drug Development
`
`Gerry Steele
`
`Astrazeneca R&D Charnwoocl
`
`Loughborough. United Kingdom
`
`ulation is usually defined as the science of the physicochemical characterization of
`e drugs. However. any studies carried out to define the conditions under which the
`e drug should be formulated can also be termed preformulation. This is a broader
`than was used in Chapter 3, and, as such, it can include studies on preliminary
`ns under a variety of conditions. These studies may influence the Product Design
`_d be conducted at the earliest opportunity at the start of development. In the in-
`..75' ter drug development and reduced drug usage, preformulation studies should not
`-3.4.-.a taken on a “check-list” basis. Rather, they should be conducted on a need—to-know
`
`there are many traditional approaches to dosage form design, newer approaches
`-ert systems are now becoming available. Expert systems are discussed further in
`Product Optimisation.
`
`Exh. 1014
`
`
`
`175
`
`Pharmaceutical Preformularron and Formulation
`
`P’B'f°’m“i3i’b” 33 5'” Aid -
`
`-
`
`v
`
`ll’? 53""? DWQ D9V5"9P”'i‘5'”‘
`
`1 77
`
`
`
`Lessthanaverage
`
`Favourable
`
`Favourable
`
`Favourable
`
`Favourable
`
`Favourable
`
`
`
`Lessthanaverage
`
`
`
`Veryfavourable
`
`
`
`Veryfavourable
`
`
`
`veryfavourable
`
`Favourable
`
`Favourable
`
`Favourable
`
`the same chemical compound can have different crystal structures (poly-
`for example.
`niurphs), external shapes (habits) and hence different flow and compression properties.
`Cartensen et al. (1993) have usefully, although briefly, reviewed the physicochemical
`properties of particulate matter, dealing with the topics of cohesion, powder flow, mi-
`cromeretics. crystallization, yield strengths and effects of moisture and hygroscopicity. Buck-
`ton (1995) has reviewed the surface characterization of pliarmaceuticals with regard to
`understanding sources ofvariability. A general overview of the methods available for the phys-
`ical characterization of pharmaceutical solids has been presented by Brittain er al. (1991). York
`( 1994) has also dealt with these issues and produced a hierarchy of testing techniques for pow-
`dered raw materials. Finally, there is a book dealing with the physical characterization of
`pharmaceutical solids. edited by Brittain (1995).
`A number of other studies can be performed on a candidate drug to determine other im-
`portant solid-state properties, for example, particle size, powder flow and compression and
`polymorphism. Therefore, when a sample undergoes initial preformulation testing the fol-
`lowing parameters should be noted: particle size, true. bulk and tapped density, surface area,
`compression properties and, powder flow properties. Some of these factors will be discussed
`in this chapter; others, however, are dealt with in more detail in Chapter 11 on Solid Oral"
`Dosage Forms.
`
`Particle Size Reduction
`
`The particle size of pharmaceuticals is important since it can afiect the formulation charac-
`teristics and bioavailability of a com pound (Chaumefl 1998). For example, sedimentation .
`flocculation rates in suspensions are, in part, governed by particle size, and inhalation ther
`of pulmonary diseases demands that asmall particle size (2-5 turn} is delivered to the lung
`the best therapeutic effect. Particle size is also important in the tableting field, since it can
`very important for good homogeneity in the final tablet. In this respect, Zhang and Iolui
`(1997) showed that a blended jet—milled compound exhibited a smaller range of poten
`when compared to those blends where the compound had a larger particle size. It is theref
`important that the particle size be consistent throughout the development studies of a ',a:'-'
`uct to satisfy formulation and regulatory demands (Turner 1987).
`Thus. to reduce the risk of dissolution rate-limited hioavailability, and if there is su
`compound, grinding in a mortar and pestle should be done to reduce the particle size
`compound. lflarger quantities are available, then ball milling or micronization can be u
`reduce the particle size. The main methods of particle size reduction have been revi-
`Spencer and Dalder (1997), who devised the mill selection matrix shown in Table 6.1.
`
`Ball Milling
`
`In a review ofmilling, it was stated that ball milling was “the most commonly used if '
`bling mill in pharmacy" (Parrot 1974). Indeed, it is probably used most often at the p ,
`'
`lation stage to reduce the particle size of small arnounts of a compound. esDeciall3'
`'--"'-
`
`
`
`
`
`Table6.1Millselectionmatrix.
`
`FluidEnergy
`
`
`
`
`
`VeryfavourableVeryfavourable
`
`
`
`
`
`VeryfavourableVeryfavourable
`
`
`
`unfavourableunfavourable
`
`
`
`LessthanaverageLessthanaverage
`
`
`
`
`
`
`
`Verylavourable
`
`
`
`FavourableLessthanaverage
`
`
`
`Lessthanaverage
`
`
`
`Lessthanaverage
`
`Average
`
`
`
`Verylevourable
`
`
`
`Verylavourable
`
`Favourable
`
`
`
`
`
`
`
`'3..Sizingupgrindingmills.G".'8l'T?fCaiEl1gll'.l&E.l'li|'lQ,Vol.ll]-ll.No.4.pp.8ll—B?099?].Withpannission.
`
`
`
`
`
`Exh. 1014
`
`
`
`1 78
`
`Pharrnaceurieai Prefottnutat.-on and Form uiation
`
`Preformuiation as an Aid. .
`
`. in Easy Drug Development
`
`1 79
`
`milling process and these include rotation speed, mill size, wet or dry milling and amount o
`material to be milled.
`_
`Although ball milling can effectively reduce the particle size of compounds. prolonged
`milling may be detrimental in terms of compound crystaliinity and stability. This has been
`.
`-
`.
`'
`‘
`'
`'
`"h cl
`illustrated in a study that examined the effect of ball mill grmding on cefixime in y rate
`(Kitamura et al 1989) Using a variety of techniques, It was shown that the crystalline solid
`was converted to an amorphous solid after 4 h in a ball mill. The stabilitl' 0f the ‘“_'“°"F'h°“5
`solid was found to be less than that of the crystalline solid, and the samples were d1SC0101-11‘ECl
`due to grinding.
`_
`g
`,
`d
`It is important to check this aspect of the milling process, since amorphous compoun s
`can Show increased bioavailability and possible pharmacological activity compared to the cor-
`.
`.
`-
`-
`-
`‘ -
`f
`—
`responding crystalline form. Ball milling may also change the polymorphic. form o a com
`pound. as shown by the work of Leung ‘it 3]‘ U999) V'“tl_'l 35P3_”am“3-
`f
`Figure 5_1, for example, shows the X-ray powder diffraction (){R£_’D) patterns 0 a sam-
`ple of 3, com pound “as received” and after ball milling. After ball milling for l h. the sample.-
`was rendered amorphous, and hence a shorter milling period was used.
`
`Micronizatfon
`.
`.
`-
`‘
`'
`'
`t‘
`to
`cl‘ '
`I
`. 9
`.
`I
`-
`'
`If instrumeiitatton and sufficient compound are available, {hen m1f<fr:It1_l::glint: 1:: I
`taken. This technique is routinely used to red!-IE6‘ the Panic f-:51“ 0 3 y ‘
`_g
`_
`the maximum surface area is exposed to enhance the solubility and dissolution properties
`Poorl)’ soluble compounds. Because of the enhanced surface area, the bioavailability
`
`compounds is often improved, e.g., rnicroiiization enhanced the bioavailability of felodipine
`when administered as an extended release tablet (lo-hansson and Abrahamsson 1997).
`The process involves feeding the drug substance into a confined circular chamber where
`the powder is suspended in a high velocity stream of air. Interparticulate collisions result in a
`size reduction. Smaller particles are removed from the chamber by the escaping air stream to-
`wards the centre ofthe mill where they are discharged and collected. Larger particles recircu-
`iate until their particle size is reduced. Micronized particles are typically less than 10 pm in
`diameter [Midoux et al. 1999).
`
`Effect of Mftifng and Micronfzation
`
`Although micronization of the drug offers the advantage of a small particle size and a larger
`surface area, it can result in processing problems due to high dust, low density and poor flow
`properties. Indeed, micronization may be counterproductive, since the micronized particles
`may aggregate, which may decrease the sttrface area. in addition, changes in crystallinity of the
`drug can also occur, which can be detected by techniques such as microcalorimetry (Briggner
`et al. i994), dynamic vapour sorption (Ward and Schultz 1995} and inverse gas chromatogra-
`phy (Feeley et al. 1998).
`Ward and Schultz (1995) reported subtle differences in the crystallinity of salbutamol sul-
`phate after micronization by air jet milling. They found that amorphous to crystalline con-
`"versions occurred that were dependent on temperature and relative humidity (RH). It was
`suggested that particle size reduction of the powder produced defects on the surface that, if
`-- nough energy was imparted, led to amorphous regions on the surface. In turn, these regions
`are found to have a greater propensity to sorb water. On exposure to moisture, these regions
`..u stallized and expelled excess moisture. This is illustrated in Figure 6.2, which shows the
`
`Figure 6.1 XRPD patterns showing the effect of ball milling on a compound.
`t
`
`Before ball milling (cryslallinei
`
`/
`
`After ball milling laiiiorphollfil
`
`§
`
`Lin{Counts}
`
`.liMllllllll.i.
`
`cl
`
`it
`i" -
`
`-" lire 6.2 DVS isotherm showing crystallization effects due to moisture.
`Ls
`I
`
`|
`
`"""I.'uur{(inI|
`-*.I\i.lm|nd|m
`
`l
`
`Exh. 1014
`
`
`
` ‘
`
`‘I80
`Pharmaceutical Preformufatfon and Formutatfon
`oisture, as measured by dynamic vapour sorption (DVS), of a rnicronized
`development compound. Note howthepercentmasschange increasesandthen decreasesas
`uptake of m
`the RH is increased between 40 and 60 percent during the sorption phase. This corresponds
`to crystallization of the compound and subsequent ejection of excess moisture. The com-
`pound also exhibits some hysteresis.
`This effect can be important in some formulations, such as dry powder inhaler devices,
`since it can cause agglomeration of the powders and variable flow properties. In many cases,
`this low level of amorphous character cannot be detected by techniques such as XRPD. Since
`microcalorimetry can detect < 10 percent amorphous content (the limit of detection is 1 per-
`cent or less}. it has the advantage over other techniques such as XRPD or DSC. Using the am‘-
`poule technique with an internal hygrostat. as described by Briggner et al. 1994. and shown
`in Figure 6.3, the amorphous content of a micronizecl drug can be determined by measur-
`ing the heat output caused by the water vapour inducing crystallization of the amorphous.
`regions.Figure 6.4 shows the calibration curve of heat output versus amorphous content of a de-
`velopment compound. In this case, the technique is used to crystallize, or condition. th _ -3
`amorphous regions by exposure to elevated R1-Is. Thus. if authentic 100 percent amorpho .-;-
`and crystalline phases exist, it is possible to construct a calibration graph of heat output v r‘-
`sus percentage crystallinity, so that the amount of amorphous character introduced by .12;
`milling process can be quantified.
`
`Figure 6.3 Internal hygrostat and heat output due to crystallization of an amorphous I
`--
`phase measured by isothermal mierocalorimetry.
`w
`P to )
`
`. in Earfy Dmg Development.Preformufarion as an A.-‘o’. 1811-j
`
`
`
`Figure 5_4 Crystallization e k
`p 3 energyVersus3m9"Ph0U'-3CflmentU5i"Q "'|lCl'0C3l0Tl""9T‘Y-
`
`
`
`CrystallizationPeal:Energy(Hg)
`
`
`
`
`
`40-0"
`
`50.00
`
`min
`
`,
`
`_
`
`mm
`
`Amulphous Content {‘.‘ii)
`
`Verse Gas Chromatography
`‘
`-
`'
`4
`an microca orirnet ‘
`-
`Wddition to the DVS
`d
`_
`I
`_
`n S of pawders a recently introduced ttéicchtechnlqlzlesforcharacterizingthesurfacePTUP»
`n also be used This techni
`'
`r
`-
`g ' C mmamgmphy
`ha
`'
`nique
`nown as inverse
`as
`]-1
`_
`que differs from traditional
`-
`I
`‘
`h
`tationary Phase is the
`.
`_
`_
`335 C Fflmdtograpliy insofar as
`P°WElEr under investigation In thi
`f
`and polar adsorbates (
`.
`5 type 0 study, a range of non-
`Probes) are used c
`alkane
`f
`h
`lethal. Dr ah
`_
`-
`-8-»
`S.
`roni
`exane to decane acetone
`yl acetate. The retenti
`.
`’
`_
`’
`‘
`qu-“ed to 1
`0" V0l|-lme, 1.8., the net volume of carrier gas (mum-
`‘
`lbetwfien E ute the prob:, is then measured. The surface partition coefficient (K ) ofthe
`arrier gas an surfaces of test
`‘
`-
`s
`_
`powder particles can then be cal
`l
`cl
`' 5 energy can be calculated which ca
`C“ are ' From
`when C°mP31‘€d to another im 1 in n 5:19; that ope batch may faV°u’ab]Y *1d501‘b the
`£._expel_imenta1 parameter I:neaEuY dear I(13érence in the surface energetics.
`re
`‘
`-
`.
`Parameter is related to the Surfai: P [F:fPerime1i:Fts_is the net retention volume,
`.the concentration
`ar 1 ion coe icient, K, which is the ratio
`of the
`robe
`'
`-
`5
`.
`molecule in the stationary and mobile phases
`
`P
`
`V
`Ks =-jxA$P
`
`Exh. 1014
`
`
`
`
`
`EXh. 1014EXh. 1014
`
`Exh. 1014
`
`
`
`‘I81’!
`
`Pharmaceutical Preformufarion and Formufatfon
`
`Have a suitable absorbancy
`
`Not swell the particles
`
`Disperse a wide range of particles
`
`Slow sedimentation of particles
`
`Allow homogeneous dispersion of the particles
`
`Be safe and easy to use
`
`mdmink
`
`
`
`
`
`.:o_Sw.E__.._Emu.m:_m:_...a.=..mmmE..m_§____o....=uu_:Eo_EmE_.a_=£_=m__.mn_mm_....._...m.n_
`
`
`
`
`
`Preforrnufarion as an Aid. .
`
`. in Early Drug Deveiopmenr
`
`185
`
`asEflunsoE:§.ua,sa
`
`E1..__%.w.nas.3QE...cannB...5Q
`REmm—m_H«mm«E033.:HEnema
`.
`
`S-mm_m.oH__5.=._.._._o.::2
`
`
`£...o_u......nE.__.:_ou_w£...o_u..=..
`uE:_n:aE..__o>
`
`2.8..one3..
`
`8.2:..mummm.“
`
`8.9:.N3mam
`
`8.2:.5.»cam
`
`8.8._Em3...;
`
`8.2:.5.2mm.5
`
`
`
`3%and.mmmmm
`
`3.9N3“
`
`
`
`5.22._._m:_u,_$mmm._8m_E_o..._a_§_m:¢
`
`£.mm=.30ES.EnemaEE3.“_E8m
`
` aman$3
`
`Exh. 1014
`
`
`
`‘I87
`
`(6)
`
`where N is the Arogad ro's number, ACS is the cross—sectional area of the adsorbate and M is the
`moleular weight of the adsorbate. it follows that the specific surface area is given by Slim.
`where m is the mass of the sample. According to the U. S. Pharmacopeia (USP), the data are
`considered to be acceptable if. on linear regression, the correlation coefficient is not less than
`0.9975, i.e., r3 is not less than 0.995.
`Figure 6.6 shows the full adsorption-desorption isotherm of two batches of the mi-
`cronized powder shown earlier in Figure 6.4.
`
`I 86
`
`Pharrnaeeuticaf Preformularfon and Formulation
`
`In terms of sample preparation, it is necessary to deaggregate the samples so that the pri-
`mary particles are measured. To achieve this, the sample may be sonicated, although there is a
`potential problem of the sample being disrupted by the ultrasonic vibration. To check for this,
`it is recommended that the particle dispersion be examined by optical microscopy.
`Although laser light diffraction is :1 rapid and highly repeatable method in determining
`the particle size distributions of pharmaceutical powders, the results obtained can be affected.
`by particle shape. In this respect, Kanerva et al. (1993) examined narrow sieve fractions of
`spherical pellets. cubic sodium chloride and acicular anhydrous theophylline. Size distribu-
`tions were made using laser light diffraction and compared to results using image analysis.
`The results showed that all determinations using the laser light scattering resulted in a broad"-
`ened size distribution compared to image analysis. In addition, it has been pointed out that‘
`the refractive index of the particles can introduce an error of 10 percent under most circum--
`stances if it is not taken into account (Zhang and Xu 1992).
`Another laser-based instrument. relying on light scattering, is the Aerosizer. This inst
`ment is for a particle sizing and is based on a time-of-flight principle as described by Niv_j-1
`(1993). The Aerosizer with aero-disperser is specifically designed to carry deaggregated pa
`cles in an air stream for particle sizing. This instrumentation has been evaluated using a sail)"
`tamol base. terbutaline sulphate and lactose (Hindle and Byron 1995).
`For submicron materials, particularly colloidal particles, quasi-elastic light scattering‘ "it
`the preferred technique. This has been usefully reviewed by Phillies (1990). The particle
`distribution of ofloxacinfprednisolone acetate for ophthalmic use has been investiga
`image analysis photon correlation spectroscopy (PCS) and single particle optical
`(SPOS) (Hacche et al. 1992). Using these techniques, it was shown that ball milling yield
`particle size of -~ 1 um and that increasing the ball-milling time increased the reprodu
`ity of diameter of particles. PCS was then used to show that extended ball milling reduc-' n}""_-
`particle size to a constant value.
`
`Surface Area Measurements
`
`The surface areas of drug particles are important because dissolution is a function of u
`rameter (as predicted by the Noyes-Whitney equation). Surface area can also be quot
`particle size is difficult to measure (Curzons et al. 1993).
`_
`Surface areas are usually determined by gas adsorption (nitrogen or krypton) -=‘
`though there are a number of theories describing this phenomenon, the most W
`.
`method is the Brunauer, Emmet and Teller, or BET, method. Adsorption methods.
`area determination have been reviewed in detail by Sing (1992). Two methods are ‘L’:
`multipoint and single— point.
`._
`Without going into too much theoretical detail, the BET isotherm for Type 1I- -'-'5'-‘ii '-
`processes (typical for pharmaceutical powders) is given by:
`P
`_
`1
`J c-1 H
`
`lrfh
`
`n\_¢-Ir’
`
`'
`
`+ p-U"
`
`Exh. 1014
`
`
`
`
`
`EXh. 1014EXh. 1014
`
`Exh. 1014
`
`
`
`
`
`EXh. 1014EXh. 1014
`
`Exh. 1014
`
`
`
`
`
`EXh. 1014EXh. 1014
`
`Exh. 1014
`
`
`
`194
`
`Pharmaceutics! Preformofation and Formulariori
`
`Preforrnuiation as an Aid. .
`
`. in Eariy Drug Development
`
`‘I 95
`
`Bonding resulting from mechanical tangling
`
`Bonding resulting from steric effects
`
`Bonds via static electricity
`
`Bonds due to free liquid
`
`Bonds due to solid bridges
`
`The caking tendency of a development compound was investigated when it was discov-
`ered to be lumpy after storage. An experiment was performed on the compound whereby it
`was stored at different RHs [from saturated salt solutions) for 4 weeks in a desiccator. Results
`revealed that caking was evident at 75 percent R1-l with the compound forming loosely massed
`porous cakes (Table 6.6). TGA of the samples showed that caked samples lost only a small
`amount of weight on heating (0.62 percent wiw), which indicated that only low levels of mois-
`ture were required to produce caking for this compound.
`It is known that micronization of compounds can lead to the formation of regions with
`a degree of disorder which, because of their amorphous character, are more reactive compared
`to the pure crystalline substance. This is particularly true on exposure to moisture and can
`lead to problems with caking, which is detrimental to the performance of the product. It has
`been argued that these amorphous regions transform during moisture sorption, due to sur-
`face sintering and recrystallization at R1-Is well below the critical RH.
`
`Polymorphism Issues
`
`Because polymorphism can have an effect on so many aspects of drug development, it is im-
`portant to fix the polymorph (usually the stable form) as early as possible in the development
`cycle.
`A US. Food and Drug Administration (FDA) reviewer’s perspective of regulatory con-
`siderations in crystallization processes for bulk pharmaceutical chemicals has been presented
`by DeCamp (1996). In this paper, he stated that
`
`process validation should include at least one, if not more, checks to verify
`that the process yields the desired polyrnorph. At the time of a New Drug
`Application (NDA) submission, it would be expected that occurrences of
`polymorphism would be established and whether these affect the dissolu-
`tion rate or the bioavailibility.
`
`He continued that
`
`is not necessary to create additional solid state forms by techniques or
`It
`conditions unrelated to the synthetic process for the purpose of clinical tri-
`als. l-lowever, submission of a thorough study ofthe effects of solvent, tem-
`perature and possibly pressure on the stability of the solid state forms
`should be considered.A conclusion that polymorphism does not occur with
`a compound must be substantiated by crystallization experiments from a
`range of solvents. This should also include solvents that may be involved in
`the manufacture of the drug product, e.g.. during granulation.
`
`Whilst it is hoped that the issue of polymorphism is resolved during prenomination and
`early development, it can remain a concern when the synthesis of the drug is sca|ed—up into a
`larger reactor or transferred to another production site. In extreme cases, and despite inten-
`sive research, work may have only produced a metastable form, and the first production batch
`produces the stable form. Dunitz and Bernstein (1995) have reviewed the appearance of, and
`subsequent disappearance of, polymorphs. Essentially, this describes the scenario whereby,
`after nucleation of a more stable form, the previously prepared metastable form could no
`longer be made.
`The role of related substances in the case of the disappearing polymorphs of sulphathia—
`role has been explored {Blagden et al. I998}. These studies showed that a reaction by—product
`from the final hydrolysis stage could stabilize different polymorphic forms of the compound,
`depending on the concentration of the by-product. Using molecular modelling techniques,
`they were able to show that ethamidosulphthiazole, the by-product, influenced the hydrogen
`bond network, and hence form and crystal morphology.
`in the development of a reliable commercial recrystallizatioti process for dirithromycin,
`Wirth and Stephenson (1997) proposed that the following scheme should be followed in the
`production of candidate drugs:
`
`j_j_.
`Table 6.6
`
`Selection of solvent system
`
`Effect of moisture on the caking of a development compound.
`
`Characterization of the polymorphic forms
`
`Moisture Content
`
`Appearance and Flow Properties
`
`[)3]
`0.2-it
`
`I12?
`
`0.32
`
`0.34
`0.62
`
`Free—flovving powder. passed easily through sieve
`Ditto
`
`Less free-flowing powder
`
`Base of powder bed adhered to petri dish;
`however. material above this flowed
`
`Less Free flowing
`Material cakerl
`
`Base of powder adhered to petri dish
`0.25
`
`
`Optimisation of process times. temperature, solvent compositions, etc.
`
`Examination of the chemical stability of the drug during processing
`
`Manipulation of the polymorphic form, if necessary
`
`Whilst examples of disappearing polymorphs exist, perhaps more common is the
`C1‘ystalli'r.ation of mixtures of polymorphs. Many analytical techniques have been used to
`quantitate mixtures of polymorphs, e.g., XRPD has been used to quantitate the amount of
`ccfepime ‘ 2HCl dihydrate in cefepime - 2l"ICl rnonohydrate (Bugay et al. 1996). As noted
`by these workers, a crucial factor in developing an assay based on a solid—state technique is
`the production of pure calibration and validation samples. Moreover, whilst the production
`
`Exh. 1014
`
`
`
`
`
`EXh. 1014EXh. 1014
`
`Exh. 1014
`
`
`
`1 98
`
`Pharmaceutical Preforriiulariori and Formulation
`
`Preformiiiaiion as an Aid. _
`
`. in Early Drug Development
`
`1 99
`
` .
`
`-
`.
`Figure 6.11 Solubility as a function of co-solvent volume for a development CDWDOUN
`
`:1.
`
`
`
`Solubility(rngl’mL]
`
`Propylene glycol LPG}
`Ethanol (HIGH)
`PEG4-00
`40% PUG + I095 ETOH
`
`|'
`40
`
`I
`60
`
`‘Bl’: Co-solvent (vfv)
`j1m—
`
`Figure 6.12 Effect of flow rate on the precipitation oi a PEG4UU solution of a drug
`compound.
`
`%Transmittauce
`
`—I— U_03inL.l'rniii
`' --o-- ans mumin
`—1‘— [ll rnL.i"iniii
`U.5 nil,-liiiin
`-—I— l.UinUinin
`
`Ti me l'rom Start of Injection (min)
`
`increased as the pKa was passed, to reach a maximum between pH 2 and 4 and then decreased
`due to the common ion effect. As the second pKa was passed in the alkaline region, the solu-
`bility again increased.
`When the solubility experiments were performed in 0.2 M citrate-phosphate buffer, the
`compound solubility decreased, and this illustrates the effect that ionic strength may have on
`drug solubility. Clearly, the region between pH 2 and 5 represents the area to achieve the best
`solubility. However, caution should be exercised if the solution needs to be buffered, since this
`can decrease the solubilty. Myrdal et al. (1995) found that a buffered forrriulation ofa coin-
`pound did not precipitate on dilution and did not cause phlebitis. In contrast, the unbuffered
`drug formulation showed the opposite effects. These results reinforce the importance of
`buffering parenteral formulations instead of simply adjusting the pH.
`
`C0-solvents
`
`The use of co—solvenls has been utilized quite effectively for some poorly soluble drug sub-
`stances. It is probable that the mechanism of enhanced solubility is the result of the polarity
`of the co-solvent mixture being closer to the drug than it is in water. This was illustrated in a
`series of papers by Rubino and Yalkowsky (1984, 1985a, b, 198721, b) who found that the solu~
`bilities of plienytoin. beiizocaiiie and diazepam in co-solvent and water mixtures were ap-
`proximated by the log—linear equation
`
`logSm:flogSL_+(1—fllogSw
`(8)
`where Sm = the solubility of the compound in the solvent mix, 5“, = solubility in water, SC is
`the solubility of the compound in pure cosolvent,f= the volume fraction of co—solvent and
`o 2 the slope of the plot of log (Sm/SW} versus)‘. Furthermore, they related s to indexes of co-
`solvent polarity such as the dielectric constant, solubility parameter, surface tension. interfa-
`cial tension and octanol-water partition coefficient.
`It was found that the aprotic co—solvents gave a much higher degree of solubility than the
`amphiprotic co-solvents. This means that if a co—solvent can donate a hydrogen bond, it may
`be an important factor in determining whether it is a good co—so|vent. Deviations from log-
`linear solubility were dealt with in a subsequent paper (Rubino and Yalkowsky 1987). Figure
`6.11 shows how the solubility of a development drug increases in a number of water-solvent
`systems. Care must be taken when attempting to increase the solubility of a compound; a
`polar drug might actually show a decrease in solubility with increasing co—solvent composi-
`tion (GouId et al. 1984).
`it is often necessary to administer a drug parenterally at a concentration that exceeds its
`aqueous solubility. Co—solvents offer one way of increasing drug solubility, but the amount of
`co-solvent that can be used in a parenteral IV formulation is often constrained by toxicity cori-
`siderations. The formulation may cause haemolysis (Fu et al. 1987), or the drug may precipi-
`tate when diliited or injected, causing plilebitis (e.g., Ward and Yalkowsky 1993). Yalkowslcy
`and co-workers (1983) have developed a useful in wire technique, based on ultraviolet (UV)
`spectrophotometry, for predicting the precipitation of parenteral formulations in viva follow-
`ing injection. Figure 6.1;’ shows the effect of injection rate on the transmittance at 600 nm of
`a PEG 400 formulation of a compound being introduced into flowing saline. As shown, the
`faster the injection rate. the more precipitation was detected by the spectrophotometer. This
`simple technique can be used to assess whether preciptation of a compound might occur on
`dilution or injection.
`
`Exh. 1014
`
`
`
`200
`
`Pharmaceutical Prefoirnutarion and Formulation
`
`Preformuiaiion as an Aid. .
`
`_ in Early DW9 D"3V5'l5'Pm9”l
`
`201
`
`Whilst co—solvents can increase the solubili
`ty ofcompounds, on occasion they can have a
`detrimental effect on their stabilit
`P
`)’
`P
`. For exam le, a arenteral formulation of the novel anti-
`tumour agent carzelsin (_U80,244)
`, using a polyethylene glygol 400 (PEG 400)i'absolute
`etlianolfpolysorbate 80 (PET) formulation (ratio 6:3: I, vi'vi'v).
`has been reported (loi1kinan-
`De Vries et al. 1995}. \/Vliilst this formulation effect
`ively increased the solubility of the con1~
`pound, this work showed that
`interbatch variation of PEG 400 could affect the stability ofthe
`drug through pH effects.
`One point that is often overlooked when considering co-
`solvents is their influence on
`buffers or salts. Since these are conjugate acid-base systems, it
`is not surprising that by intro-
`ducing solvents into the solution, a shift in the pKa of the buffer or salt can result. These ef-
`fects are important in formulation terms, since many inject-able formulations that contain
`co-solvents also contain a buffer to control the pH (Rubino 1987}.
`
`Ernulsfon Formulations
`
`Oil-in—water (olw) emulsions have been successfully employed to deliver drugs with poor
`water solubility, e.g. diazepam (Collins—Gold et al. 1990). In preforrnulation terms, the solu-
`bility of the compound in the oil phase (often soybean oil) is the main consideration in using
`this approach. However, the particle size ofthe emulsion and its stability (physical and chem-
`ical} also need to be assessed. Ideally. the particle size of the emulsion droplets should be in the
`colloidal range to avoid problems with phlebitis. To achieve this size, a microfluidizer should
`be used. since other techniques may produce droplets of a larger size, as shown in Table 6.7
`
`The particle size of emulsions can be measured using PCS (e.g.. Whateley et al. 1984),
`whilst the surface charge or zeta potential can be measured using electrophoretic mobility
`measurements (Levy and Benita 1989). Physical instability ofernulsions can take a number of
`forms. e.g.. creaming, flocculation. coalescence or breaking, whilst chemical instability can be
`due to hydrolysis of the stabilizing moieties. In order to assess the stability of the emulsion,
`heating and freezing cycles can be employed, as well as centrifugation (Yalabilc-Kas 1985}.
`Chansiri et al. (1999) have investigated the efifect ofstearn sterilization (l2]°C for 15 min) on
`
`
`Table 6.7
`Size of emulsion droplets produced by various methods.
`Method of Manutacture
`Particle Size (urn)
`Vortex
`t}.i}3—2-‘i
`Blade rniiter
`-" M "
`
`,
`,
`.
`-
`I
`th a bi h negative zeta potenlld
`the stability or oiwhemulsliorgz. 'l‘hE;:’rEJCl.ll;1E(:li$Il-I:ll1i;lu5tliit:Il‘tlSaVf~;ier autoicgiaving Emulsions with a
`did not show any i: ange in
`eir p
`1
`_
`bases during auto_
`'
`hand were found to separate into two P
`lower negative value, on the other
`.
`‘
`_
`_
`_
`d
`the surface
`,
`-
`-
`l‘10I‘IS IS depen erit on
`'
`‘
`-
`1
`cl
`claving. Because the stability of PhD5Pl'l0l1P‘d stabl ‘:6 em; 5
`charge. these emulsions are normallll autoclaved 3‘ PH 3‘ -
`
`Stability Considerations
`
`_
`.
`'lity. Notari
`‘
`'
`-
`ect to solution formulations is stabi _
`_ _.
`.
`.
`.'
`'
`th
`The gibimd mdlorteldogdliieeriifgiimthts ifeljglitrding the merits of a con‘lPl‘~’le l‘“"‘i“': Stablmy
`.
`1
`‘
`16
`eriments were re-
`“'5 pres“
`“'99
`‘
`liable data and no buffer cata ysls. WP
`_
`*"“‘dl"' He calculated that with re
`-
`-
`'
`t
`‘b te to the hydrolysis,
`-
`'
`lete kinetic stability study. If buffer ions C01‘!
`|'1 U
`quired to provide :1 Con‘-_}3
`ression Thus for 3 Single buffer‘ fig” phos-
`then 9351‘ 5PeCi'35 lwntflbutes to the PH4-are six?
`f the com ound should
`I
`1’
`'
`I
`-
`-
`is was required. A stock so utioii o
`P
`'
`Phdtev 3 ““"‘”“”“ “f 6 °""°'”"""
`-
`s
`1.
`dded to e.g.. a buffer
`-
`’
`d a small aliquot (E.g.. . 0 il
`l 3
`'
`ht‘ P1'eP"ed 1" an appropriate mlvent an
`erature and the
`.
`.
`.
`_
`.
`-
`'
`Id be maintained at .1 constant temp
`.
`‘
`A
`r,‘[)lL1I1(‘}l'i at a set pl-I. This solution shnu
`ft
`h
`u h ml-‘mg
`I
`i
`.
`he addition ofKCl(e.g-.1’: ll-5l-A 3” 0"’ 3 _
`‘
`IOHIC Strength may be Controlled by t
`-
`d of interest.
`.
`.
`.
`'
`‘
`sa ed for the compoun
`t
`d
`The solution 15 then sai'nPle'd'at vdrlousethldgdpillildtstlihl saiihpllts are diluted into a medium that
`lflhe rennin" is Very fiat‘ it [5 recomm
`h
`’
`table in acid
`_
`.
`.
`»
`. for example, a compound t at is uns
`_
`will stop or substantially slow the reaction,
`_
`_ fill S]
`reactmns
`.
`-
`'
`l nma alsobeui-6
`OW
`_
`>
`‘
`h
`l
`-
`may be stable 111 an alkaline I"l’1£‘l’.:ll.]I'l1é Ctiifllnigrzgfali gligvatedltemperature. In this situation.
`on the other han 3 lT'l3Y require ‘"15 7'
`.
`nd
`-
`_
`.
`f
`t re. If sufficient compou
`is available, the ejfectjof. e,B~i buffer ciI:;ce(:1ft:lniaCi.d labile Compound with respect m PH is
`The first—or er
`ecomposi I011 P
`_
`_
`.
`,
`H 7
`me deg
`»
`-
`'
`d is very acid labile, and EVCN 41 P
`I 50
`shown in Figure 6.13. Clearly, this compotm _
`‘
`id therefme be difficult
`to
`composition is observed. A stable solution formulation wou .
`is that by London
`fii
`achieve in this pH range.
`‘
`_
`_
`_
`_
`A detailed paper on th: mllrlhammc inlelllgeéillhgvgfrgfieijdd Elli: Silg‘-ftematic interprets-
`U991). More recently‘. V3"
`9" Uuwen *3 3 '
`- d an take a num.
`_
`.
`-
`-
`files obtained when pH is vane C
`.
`HOT! 01 Pl'1'd"-‘gradation Pr°fil-‘3S- The rate pro
`'
`..
`.
`ll
`ist of linear
`he point that they 113113 Y 5°“
`ber of forms. However. 1-Dildo“ “99” makes t
`h l-
`6 ions en-
`,,
`1
`.
`-
`db short curved segments Jndeedti 5 “lea” 8
`_g
`’eg”’“5 “F mlegral slope mnnecte
`Ya
`arded as a composite of
`I
`b
`er-ally ha