European Journal of Pharmaceutical Sciences 8 (1999) 301–308
`
`Physical state of L-histidine after freeze-drying and long-term storage
`¨
`a
`Thomas Osterberg , Tommy Wadsten
`¨
`¨
`Pharmaceutical and Analytical R&D, Astra Pain Control AB, S-15185 Sodertalje, Sweden
`bDev & Res Wadsten AB, S-11327 Stockholm, Sweden
`
`b
`
`a
`
`Received 27 November 1998; received in revised form 2 March 1999; accepted 18 March 1999
`
`Abstract
`
`Liquid samples of L-histidine of varying pH values and mixed with salt, metal ions, polysorbate 80 and sucrose have been analysed by
`differential scanning calorimetry to evaluate the influence of these additives on the glass transition temperature and crystallisation of
`L-histidine during freezing and thawing. L-Histidine solutions of varying pH were freeze-dried with and without a thermal cycle and the
`physical state of the freeze-dried cakes, following long-term storage, were studied by powder X-ray diffraction. Amorphous L-histidine
`crystallised when it was exposed to moisture, and the identity of the crystalline materials is reported. The crystallisation of L-histidine
`during freezing and thawing is dependent on the pH of the solution and is shown to be at a minimum at pH 6, which coincides with the
`pK of the imidazoline function. Sucrose inhibited the crystallisation of L-histidine during thawing, while sodium chloride or polysorbate
`a
`21
`21
`80 did not. The addition of metal ions (Ca
`and Mg
`) up to 10% (w/w) did not depress the glass transition temperature significantly,
`21
`while the addition of Zn
`increased it. The physical state of L-histidine after freeze-drying is shown to be dependent on both the pH of
`the solution and the freezing cycle. The risk of crystallisation of amorphous L-histidine is low if the freeze-dried material is protected from
`moisture.
`1999 Published by Elsevier Science B.V. All rights reserved.
`
`Keywords: Amorphous; Crystallisation; Freeze-drying; Glass transition; L-Histidine; Metal ions
`
`1. Introduction
`
`Protein drugs are generally chemically and physically
`unstable in solution and freeze-drying is frequently used to
`obtain an acceptable shelf life (MacKenzie, 1966, 1977;
`Pikal, 1990). Sugars and/or amino acids are often included
`in the formulation to prevent inactivation during freeze-
`drying and to stabilise the protein during long-term
`storage. Sugars and amino acids protect the protein by
`preferential exclusion during freezing and by glass forma-
`tion and /or by functioning as a water substitute in the
`dried state (Carpenter and Crowe, 1989; Franks et al.,
`1991; Arakawa et al., 1992). In contrast to sugars, amino
`acids, in addition to their stabilising properties, may also
`function as buffers. L-Histidine has recently been shown to
`function as both buffer and stabiliser
`in freeze-dried
`¨
`formulations of recombinant factor VIII (Osterberg et al.,
`1997) and recombinant factor IX (Bush et al., 1998). It is a
`basic amino acid often found at the active site in enzymes
`and in the coordination of metal ions in metalloproteins.
`The specific properties of L-histidine reside in the im-
`idazoline function, which possesses both basicity and p-
`
`¨
`E-mail address: thomas.osterberg@eu.pnu.com (T. Osterberg)
`
`electron acceptor capability (Sundberg and Martin, 1974).
`The imidazoline function confers good buffer capacity in
`the pH range 5–7 (Fig. 1), which is often a suitable pH
`range for many protein drugs. L-Histidine forms strong
`21
`21
`complexes with certain metal ions such as Cu , Zn
`and
`21
`1
`1
`Fe
`, but simple salts with the alkali metals (Na , K ,
`21
`21
`Ca
`and Ba
`) (Greenstein and Winitz, 1961). The latter
`fact is especially important in the formulation of protein
`drugs that require free calcium ions in the buffer for
`stability reasons (e.g. factor VIII). Trace levels of metal
`ions such as copper and iron are often present in buffer
`salts. These ions can often facilitate the oxidation of
`proteins (Lamfrom and Nielsen, 1970; Shihong et al.,
`1993). Since L-histidine forms strong complexes with these
`ions, it may also function as an antioxidant. The solubility
`of L-histidine in water is 41.9 mg/ml at 258C and is
`sufficient for proper buffering and to allow it to function as
`a non-crystallising (amorphous) stabiliser for many protein
`drugs. In this study the freezing/thawing behaviour of
`L-histidine of varying pH in the presence of sodium
`chloride, metal ions and sucrose was studied by means of
`differential scanning calorimetry (DSC). L-Histidine solu-
`tions of varying pH were freeze-dried with and without a
`thermal cycle. The freeze-dried cakes were examined with
`
`0928-0987/99/$ – see front matter
`PII: S0928-0987( 99 )00028-7
`
`1999 Published by Elsevier Science B.V. All rights reserved.
`
`CFAD Exhibit 1030
`
`1
`
`(cid:211)
`(cid:211)
`

`

`302
`
`¨
`T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`The solutions were sterile-filtered and 1.0 ml was
`dispensed into glass vials. About 10 vials of each formula-
`2
`tion were prepared. A pilot freeze-dryer with a 1.2-m
`shelf area (Edward Kniese, Marburg, Germany) was used.
`The samples were frozen on the shelves (from 0 to 2608C
`in 1 h). The thermal cycle was 260 to 2358C in 1 h,
`maintaining 2358C for 4 h and cooling to 2608 in 1 h.
`Primary drying was carried out at a shelf temperature of
`about 2208C and the pressure was adjusted to give a
`product temperature of about 2408C. On completion of the
`primary drying, the shelf temperature was raised to 408C
`over 10 h. The completion of the primary and secondary
`drying was determined by a pressure increase test. The
`vials were stoppered under a vacuum in the freeze-drier.
`The freeze-drying process is illustrated schematically in
`Fig. 2.
`
`2.2.2. Powder X-ray diffraction
`Powder X-ray diffraction data were obtained with a
`´
`powder diffractometer (XPert, Philips, The Netherlands).
`Data were collected at room temperature from 3 to 358, 2u,
`the step size was 0.028 and the count time was 0.5 s.
`Powder samples were prepared as thin layers on glass or
`aluminium specimen holders. The samples were first
`analysed at ambient
`temperature and then exposed to
`moisture from 20 to 80% RH at room temperature in the
`diffractometer by using a humidity control unit
`(also
`Philips-made).
`
`2.2.3. Water sorption/desorption
`The water sorption/desorption profile was determined
`with an isothermal dynamic water uptake instrument (MB-
`300W, VTI Corporation, USA). The sample was dried
`under nitrogen at 608C. The temperature was thereafter set
`to 358C and the relative humidity was increased from 0 to
`80% for sorption and decreased from 80 to 5% for
`desorption in steps of 5%. The initial sample mass was
`about 7 mg and the equilibrium criterion was less than
`0.005 mg mass gain per 3 min. Two vials of each
`formulation (pH 5, 6 and 8) were analysed.
`
`Fig. 2. Plot of process variables during freeze-drying.
`
`Fig. 1. Simulated titration curve of L-histidine showing the major
`electrical forms at varying pH.
`
`powder X-ray diffraction after long term-storage and after
`moisture exposure. The exposure to moisture induced
`crystallisation and the identity of the crystalline materials
`is reported.
`
`2. Experimental
`
`2.1. Materials
`
`L-Histidine used for the freeze-drying experiments con-
`formed to the requirements laid down in DAB and USP.
`L-Histidine (SigmaUltra), L-histidine monohydrochloride
`monohydrate, sucrose (SigmaUltra), and sodium chloride
`(SigmaUltra), used for DSC studies and as references for
`the powder X-ray experiments, were
`from Sigma
`(Sweden). Magnesium chloride (ACS reagent grade, ICN
`Biomedicals), calcium chloride dihydrate (ACS reagent
`grade, Acros) and Tween 80 (Polysorbate 80) were
`obtained from Chemicon (Sweden). Zinc chloride (Ph.
`Eur) came from KEBO Lab (Sweden). The sterile filter
`used was a 0.22-mm, Millex GV (Millipore, Sweden). The
`containers used were of type 1 (Ph. Eur). Bottles were
`closed with a bromobutyl rubber stopper and sealed with
`an aluminium seal. Water for injections, hydrochloric acid
`and sodium hydroxide were of pharmacopoeial grade.
`
`2.2. Methods
`
`2.2.1. Preparation of freeze-dried samples
`L-Histidine was dissolved in water for injections (15
`mg/ml) and the pH was adjusted to pH 4, 5, 6, and 7 with
`hydrochloric acid and to pH 8 with sodium hydroxide.
`
`2
`
`

`

`¨T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`303
`
`2.2.4. Thermal analysis
`
`2.2.4.1. Preparation of samples for thermal analysis
`for
`L-Histidine, 30 mg/ml, was dissolved in water
`injections and adjusted with hydrochloric acid or sodium
`hydroxide to the different pH values. Mixtures of L-
`histidine and sucrose were prepared by mixing L-histidine
`(30 mg/ml) and sucrose (30 mg/ml) in different volume
`ratios. The solutions were dispensed (about 30 ml) in a
`50-ml aluminium pan without a lid. An empty pan was
`used as reference.
`
`2.2.4.2. DSC
`A differential scanning calorimeter, DSC6200, with an
`EXTAR6000 workstation (Seiko Instruments, Japan), was
`used. Cyclohexane (SigmaUltra) and indium (Laboratory
`of the Government Chemist, UK) were used to calibrate
`the instrument. The oven was cooled with liquid nitrogen.
`The sample was cooled to 21108C at a rate of 108C/ min.
`After an equilibration period of 2 min the sample was
`heated to 2208C at 108C/min. The sample was cooled
`again to 21108C, equilibrated for 2 min and subsequently
`reheated to 08C at 108C/min. The sample was inspected
`under a magnifying glass or stereomicroscope directly after
`analysis when some ice was still melting to observe
`possible crystallisation. Runs were also conducted with a
`glass lid on the oven so that crystallisation could be
`observed visually. Reported glass transition temperatures
`are midpoint values (mean value from two runs) from the
`second scan, and were determined with help of
`the
`software and the derivative of the DSC signal.
`
`3. Results and discussion
`
`The selection of buffer for a protein formulation is very
`important and several factors have to be considered. The
`buffer must have low local and systemic toxicity and be
`compatible with the active protein and other essential
`ingredients (e.g. metal ions). It must also be chemically
`stable and the pK should preferably be close to the
`a
`formulation pH in order to give good buffer capacity. The
`latter requirement also implies that the extent of ionisation
`of the buffer varies considerably around the formulation
`pH. This also applies to drug compounds with a pK close
`a
`to the pH of the formulation. For example, a compound
`with a pK of 7.0 is 50% in the ionised state at pH 7.0, and
`a
`a decrease by one pH unit increases the ionisation to 90%.
`Thus, for compounds with a pK near the formulation pH,
`a
`we are dealing with different
`ratios of charged and
`uncharged forms depending on relatively small changes in
`the formulation pH. This is in contrast to non-ionisable
`compounds (such as sucrose and mannitol) or compounds
`with a pK a long way from the formulation pH. Histidine
`a
`has three ionisable functions: the carboxyl group (pK 5
`1
`1.9),
`the imidazole nitrogen (pK 56.1) and the amino
`2
`
`nitrogen (pK 59.1) at 258C and 0.15 ionic strength
`3
`(Sundberg and Martin, 1974). A theoretical titration curve
`with the major electrical forms of L-histidine is shown in
`Fig. 1. Since the pK is close to the pH range often used in
`2
`the formulation of protein drugs, the net electrical charge
`of the imidazoline moiety as a function of pH is of
`fundamental importance for its physicochemical properties
`and its behaviour during freezing and in the dried state.
`
`3.1. DSC
`
`The thermal behaviour of L-histidine during freezing and
`thawing has recently been studied by (Chang and Randall,
`1992), who reported that L-histidine (10 mg/ml) remains
`amorphous during freezing, with a T value of 2328C.
`9
`g
`The T of L-histidine (30 mg/ml) measured in the present
`9
`g
`study was 231.58C, although L-histidine crystallised dur-
`ing thawing at about 2108C. Since this was in contrast to
`the findings reported by Chang and Randall (1992), a
`sample of 10 mg/ml and rapid freezing was also investi-
`gated. The T was 2328C, although crystallisation took
`9
`g
`place at about the same temperature as for the sample
`consisting of 30 mg/ml. However, since the crystallisation
`takes place during the softening/melting of the ice, the
`exotherm is obscured. An exotherm was observed at about
`2108C in the sample of 30 mg/ml but not in the sample of
`10 mg/ml. The crystallisation could also be observed
`visually if a glass lid was used on the calorimeter oven.
`The crystallisation takes place mainly on the surface of the
`sample solution and the crystals have a flat, round appear-
`ance. This type of crystals was also observed on the
`surface of some of the freeze-dried samples. The thermo-
`gram from the pH 4 sample showed crystallisation between
`230 and 2208C. Crystallisation was also observed visual-
`ly in the sample cup in this temperature range and also in
`the sample cup directly after thawing. No crystallisation of
`L-histidine during thawing was observed in samples in the
`pH range 5.5–6.0. The pH 6.5 samples showed one small
`crystal. Thus, the findings from this study indicate that the
`tendency for crystallisation is at a minimum between 5.5
`and 6.5. Crystallisation of L-histidine was also observed in
`the samples containing sodium chloride. Thus, the reduced
`crystallisation of L-histidine between pH 5.5 and 6.5 is
`most
`likely due to the ionisation of
`the imidazoline
`function, since the addition of sodium chloride (up to 0.6
`M and unadjusted pH) did not inhibit the crystallisation.
`The thermal analysis of L-histidine solutions of varying pH
`showed a sigmoidal relationship between pH and T (Fig.
`9
`g
`3). The T at pH 5 was about 108C lower than at pH 7.
`9
`g
`The effect of pH on T can be explained by the plasticis-
`9
`g
`ing effect of ions (from pH adjustment with HCl or NaOH)
`in the freeze-concentrated L-histidine glass and/or a lower
`T of L-histidine with an ionised imidazole moiety. It is
`9
`g
`well known that salts decrease T
`and the collapse
`9
`g
`temperature of freeze-concentrated amorphous materials.
`The T and the collapse temperature of amorphous materi-
`9
`g
`
`3
`
`

`

`304
`
`¨
`T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`Fig. 3. Dependence of T during the second heating of frozen aqueous
`9
`g
`L-histidine solutions on the pH and on the addition of sodium chloride.
`
`als are generally closely related (MacKenzie, 1977). The
`of
`the sucrose/ sodium chloride system has been
`9T
`g
`described by (MacKenzie, 1985). The addition of sodium
`chloride to sucrose in 1:9 and 2:8 (w/w) ratios depressed
`the T by about 10 and 218C, respectively. If one assumes
`9
`g
`that sodium and chloride ions depress the collapse tem-
`perature equally, the chloride ions alone would depress the
`9T by about 5 and 108C, respectively,
`in the sucrose /
`g
`sodium chloride system above. L-Histidine (30 mg/ ml)
`adjusted to pH 6 with hydrochloric acid contains about 3.4
`mg of chloride ions. The L-histidine/chloride ion ratio is
`thus approximately 10:1 (w/w). If one further assumes that
`sodium chloride depresses the T of sucrose and L-his-
`9
`g
`tidine equally, the chloride ions would depress the T of
`9
`g
`L-histidine (pH 6) by about 108C. The measured T was
`9
`g
`about 88C lower at pH 6 compared to the unadjusted state,
`which is quite close to the value estimated from the
`sucrose/sodium chloride system. However, it is not pos-
`sible to establish the T of L-histidine at pH 6 per se due to
`9
`g
`the presence of the pH adjuster, although the reasoning
`above indicates that
`the ionisation of the imidazoline
`function does not change the T dramatically. Since L-
`9
`g
`histidine was shown to crystallise during thawing in the
`important pH range 6.5–8 it was important to investigate
`whether the addition of a non-crystallising excipient could
`function as an inhibitor. Sucrose was selected since it is
`often used as a non-crystallising excipient in freeze-dried
`formulations of protein drugs. The T of L-histidine/ suc-
`9
`g
`rose mixtures of varying pH is shown in Fig. 4. The T of
`9
`g
`mixtures of two non-crystallising compounds can generally
`be approximated using the Fox equation (Fox, 1950),
`
`Fig. 4. Dependence of T during the second heating of frozen aqueous
`9
`g
`L-histidine–sucrose solutions on the weight to weight ratio of L-histidine–
`sucrose and pH.
`
`are the
`is the new T and W , T , W and T
`where T
`g
`1
`g 1
`2
`g 12
`g 2
`weight fractions and the T of the individual compounds.
`g
`The T values measured in this study were 231.58C for
`9
`g
`L-histidine (30 mg /ml) and 232.38C for sucrose (30
`mg/ml). Thus, the estimated T of any L-histidine–sucrose
`9
`g
`mixture will not deviate much from 2328C.
`It was,
`therefore, interesting to note that the T of all the mixtures
`9
`g
`(unadjusted pH) was about 3–48C higher compared to the
`T values estimated from the Fox equation. The T at pH
`9
`9
`g
`g
`7 was essentially unchanged for the samples with 33 and
`50% (w /w) L-histidine. The samples with 33 and 50%
`(w/w) L-histidine of pH 6 showed a T above 2358C
`9
`g
`which is a reasonable T from a practical point of view.
`9
`g
`The T of pure L-histidine at pH 6 was 2408C. A T at or
`9
`9
`g
`g
`below 2408C is generally considered to be too low since
`the sublimation of ice is very slow at this temperature. As
`a rule of thumb, the time for the primary drying is halved
`if the product temperature is increased by 6–78C. Thus, the
`increased T
`of the L-histidine/sucrose mixtures is of
`9
`g
`economic and practical
`importance. Another important
`observation was that the addition of sucrose abolished the
`crystallisation of L-histidine. The reduced tendency for
`crystallisation of L-histidine is very important
`in the
`formulation design, since even if the product temperature
`never reaches the devitrification temperature during normal
`operating conditions, accidental over-heating might occur
`in the freeze-dryer due to technical failures.
`
`3.2. Addition of metal ions and polysorbate 80
`
`1/T 5 W /T 1 W /T
`g 12
`1
`g 1
`2
`
`g 2
`
`Some proteins like factor VIII require divalent metal
`
`4
`
`

`

`¨T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`305
`
`21
`
`) in the formulation buffer for stability
`ions (e.g. Ca
`reasons. Proteins that are therapeutically active at very low
`concentrations usually require the addition of non-ionic
`surfactants (e.g. polysorbate 80) in order to retard the
`surface adsorption of the protein during manufacture and
`in the final packaging. It was,
`therefore, of interest
`to
`21
`21
`investigate whether the divalent metal ions, Ca
`, Mg
`21
`and Zn
`(added as chloride salts) or polysorbate 80, could
`retard the crystallisation and how these ingredients in-
`fluenced the T of L-histidine. Sucrose was included in
`9
`g
`order to study how an uncharged compound interacts with
`these metal
`ions. The crystallisation of L-histidine (30
`mg/ml unadjusted pH) was largely unaffected by the
`presence of calcium chloride (up to 10 mg/ml), but was
`completely retarded at 20 mg/ml. The addition of poly-
`sorbate 80, 0.2 mg/ ml, did not inhibit the crystallisation.
`21
`21
`21
`The effect of Ca
`, Mg
`and Zn
`on the T of L-
`9
`g
`histidine and sucrose is shown in Fig. 5. The addition of
`metal ions to sucrose showed an expected concentration-
`21
`dependent depression of the T . Mg
`depressed the T of
`9
`9
`g
`g
`21
`21
`sucrose most, followed by Ca
`and Zn . However, in
`21
`21
`contrast to sucrose, the addition of Ca
`and Mg
`to
`L-histidine showed a surprisingly small depression of T of
`9
`g
`up to 10% (w/w) ratio. Another unexpected finding was
`21
`that the addition of Zn
`increased the T considerably.
`9
`g
`21
`21
`and Mg
`The small depression of T by Ca
`up to 10%
`9
`g
`(w/w) ratio indicates that L-histidine interacts with these
`metal
`ions in the freeze-concentrated phase. It
`is well
`21
`known that L-histidine forms a strong complex with Zn ,
`and this probably explains the increase in the T . The
`9
`g
`
`21
`
`21
`
`of L-
`of the T
`and Mg
`small depression by Ca
`9
`g
`histidine is of considerable importance in the formulation
`of proteins that require metal
`ions for stability. If L-
`histidine is used as buffer/stabiliser in formulations con-
`21
`21
`taining Ca
`and Mg , the T is essentially unchanged,
`9
`g
`the T
`whereas if sucrose is used,
`can be depressed
`9
`g
`considerably.
`
`3.3. Freeze-drying
`
`The freezing step is very important since the internal
`structure of the product
`is determined by this step. A
`thermal cycle (Fig. 2) is sometimes necessary to maximise
`the crystallisation of water and/or to promote the crys-
`tallisation of bulking agents such as sodium chloride or
`mannitol. Since a thermal cycle might promote crystallisa-
`tion of ingredients (e.g. L-histidine) that are to remain
`amorphous, the L-histidine samples were freeze-dried with
`and without a thermal cycle. Freeze-drying of L-histidine
`from solutions having a pH in the range 4–8 showed that
`L-histidine has a rather low tendency to crystallise during
`freeze-drying (Table 1). The samples freeze-dried at pH 8
`crystallised readily, but only when they were freeze-dried
`with a thermal cycle. The samples of pH 7 and 8, freeze-
`dried without a thermal cycle, and the sample with pH 7,
`freeze-dried with a thermal cycle, showed very thin crystal
`clusters (1–2 mm in diameter) on the surface of the cakes.
`All samples showed good cake structures, with the excep-
`tion of the pH 4 sample, which collapsed to a transparent
`skin during freeze-drying. Examination of the samples
`with a polarising microscope showed that the pH 8 sample
`freeze-dried with a thermal cycle had a crystalline struc-
`ture, while the other samples had an amorphous appear-
`ance. The crystallisation of L-histidine at pH 8 during
`freeze-drying might be explained by the fact
`that
`the
`imidazoline moiety is almost completely un-ionised at this
`pH and that crystallisation is generally facilitated if only
`one molecular species is present. At pH 6 there is a 1:1
`molar ratio of the charged and the uncharged imidazole
`moiety and it is suggested that this ratio is optimum for the
`inhibition of crystallisation during freeze-drying. However,
`a complicating factor in the interpretation of the results is
`that both the ion product of water (pK ) and the dissocia-
`w
`tion constant pK of the imidazole moiety change with
`a
`temperature. In addition, when the water is frozen out, the
`concentration effect has also to be considered. The mean-
`
`Table 1
`Physical state of L-histidine after freeze-drying with and without a thermal
`a
`cycle from solutions of varying pH
`
`Freezing process
`
`pH 4
`
`pH 5
`
`pH 6
`
`pH 7
`
`pH 8
`
`No Thermal cycle
`Thermal cycle
`
`b
`A
`nt
`
`A
`nt
`
`A
`A
`
`c
`A
`c
`A
`
`c
`A
`C
`
`Fig. 5. Dependence of T during the second heating of frozen aqueous
`9
`g
`L-histidine and sucrose solutions on the weight to weight ratio of L-
`histidine or sucrose/salt.
`
`a
`
`b
`
`c
`
`A, X-ray amorphous; C, Crystalline; nt, not tested.
`The cake collapsed.
`Thin crystal clusters on the surface of the cakes.
`
`5
`
`

`

`306
`
`¨
`T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`ing of pH at low temperature is complex. The arbitrary
`o
`zero potential is the standard potential EH of the revers-
`ible hydrogen electrode, and this by definition is taken as
`zero at all temperatures. Accordingly it is not possible
`from the measurement from the hydrogen electrode po-
`tentials to compare exactly the true hydrogen activity at
`two different temperatures (Taylor, 1981). However, the
`ratio between the ionised and the un-ionised imidazole
`moiety in the freeze concentrated L-Histidine glass has to
`remain 1:1 since electroneutrality must be maintained. The
`carboxy and amino groups of L-histidine are fully ionised
`in the pH range investigated, so the chloride ions must be
`‘neutralised’ by the charged imidazole nitrogens. The
`dependence of the T of L-histidine on the pH has to be
`9
`g
`considered in the design of the freeze-drying cycle, since
`the T determines the maximum product temperature that
`9
`g
`can be used during primary drying. In practice it is usual to
`allow at
`least a 60.5 pH deviation from the specified
`value. For a formulation with a specified value of pH 7.0,
`these limits might give a 68C decrease in the T .
`9
`g
`
`3.4. Powder X-ray diffraction and water sorption/
`desorption
`
`The samples that were freeze-dried without a thermal
`cycle were subjected to powder X-ray diffraction analysis
`after a 2-year storage at 378C. The diffractograms (Fig. 6)
`
`showed that all the samples were X-ray amorphous. The
`samples freeze-dried with a thermal cycle (stored at 2–88C
`for 2 years) revealed some crystallinity, which decreased
`as the pH decreased (Fig. 7). A sample consisting of a
`mixture of sucrose and L-histidine (33%, w/w, of L-
`histidine and pH 7) is also shown in Fig. 7. This sample
`was X-ray amorphous, which indicates that the addition of
`sucrose inhibits crystallisation during freeze-drying and
`long-term storage. It is noteworthy that all the samples
`freeze-dried without a thermal cycle were X-ray amor-
`phous, which indicates that the amorphous to crystalline
`transformation of L-histidine is slow if the product
`is
`protected from moisture contact. The samples were ana-
`lysed in duplicate and the only difference between the
`diffractograms was the normal variation in amplitude
`which always occurs in powder X-ray diffraction. Fig. 8
`shows the water sorption and desorption profiles of
`samples freeze-dried at pH 5, 6 and 8, respectively. All the
`samples were highly hygroscopic and the sorption/ desorp-
`tion profiles indicate that
`they all start
`to crystallise
`between 20 and 40% RH. The end-point of the desorption
`profiles (weight ratio of water and L-histidine) indicated
`that the pH 5 sample crystallised as a monohydrate, the pH
`8 sample crystallised as an anhydrate and the pH 6 sample
`crystallised as a hemihydrate or a mixture of anhydrate and
`monohydrate. The duplicate samples showed very similar
`sorption/desorption profiles (not illustrated). The powder
`
`Fig. 6. Powder X-ray diffraction pattern of freeze-dried L-histidine (pH 4, 5, 6, 7 and 8) after a 2-year storage at 378C. Note that all samples were X-ray
`amorphous.
`
`6
`
`

`

`¨T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`307
`
`Fig. 7. Powder X-ray diffraction pattern of L-histidine (pH 6, 7 and 8) and L-histidine–sucrose (33%, w/ w), freeze-dried with a thermal cycle and after
`long-term storage under refrigeration. Note that the crystallinity is decreased at lower pH and that the sample with sucrose is completely X-ray amorphous.
`
`X-ray analysis of the freeze-dried samples after moisture
`exposure is shown in Fig. 9. The diffractograms showed
`that the solution freeze-dried at pH 5 crystallised as the
`L-histidine monohydrochloride monohydrate. The pH 8
`
`solution crystallised as L-histidine and the pH 6 sample
`crystallised as a 1:1 mixture of L-histidine monohydro-
`chloride monohydrate and L-histidine. The moisture-in-
`duced crystallisation was also readily observed in a
`polarising microscope by exposing the sample to moisture
`when crystallisation took place almost immediately.
`
`4. Conclusions
`
`L-Histidine may be regarded as a multifunctional protein
`stabiliser since it can function as a buffer and a metal ion
`scavenger and stabilise the protein in an amorphous phase.
`When formulating proteins with L-histidine, it is important
`to consider the pH value with regard to both possible
`crystallisation of L-histidine and its influence on the T .g
`9
`21
`21
`can be added to
`Metal
`ions such as Ca
`and Mg
`L-histidine without significant depression of T . The risk of
`9
`g
`crystallisation of amorphous L-histidine is low as long as
`the formulation is protected from moisture contact. The
`addition of sucrose reduces the risk of crystallisation and
`increases the T of the freeze-concentrated mixture.
`9
`g
`
`Acknowledgements
`
`Fig. 8. Water sorption/desorption of freeze-dried L-histidine (pH 5, 6 and
`8) as a function of the relative humidity. Note crystallisation between 20
`and 40% relative humidity.
`
`We gratefully acknowledge the assistance of Johan
`Andersson, Astra Arcus, who performed the moisture
`balance studies.
`
`7
`
`

`

`308
`
`¨
`T. Osterberg, T.Wadsten / European Journal of Pharmaceutical Sciences 8(1999)301–308
`
`Fig. 9. Powder X-ray diffraction pattern of freeze-dried L-histidine (pH 5, 6 and 8) after moisture-induced crystallisation. L-Histidine and L-histidine?HCl?
`H O (Sigma) were included as references.
`2
`
`References
`
`Arakawa, T., Prestrelski, S.J., Kenney, W.C., Carpenter, J.F., 1992.
`Factors affecting short-term and long-term stabilities of proteins. Adv.
`Drug. Deliv. Rev. 10, 1–28.
`Bush, L., Webb, C., Bartlett, L., Burnett, B., 1998. The formulation of
`recombinant factor IX: stability, robustness, and convenience. Semin.
`Hematol. 35, 18–21.
`Carpenter, J.F., Crowe, J.H., 1989. An infrared spectroscopic study of the
`interactions of carbohydrates with dried proteins. Biochemistry 28,
`3916–3922.
`Chang, B.S., Randall, C.S., 1992. Use of subambient thermal analysis to
`optimize protein lyophilization. Cryobiology 29, 632–656.
`Fox, T.G., 1950. Second-order transition temperatures and related prop-
`erties of polystyrene. I. Influence of molecular weight. J. Appl. Phys.
`21, 581–591.
`Franks, F., Hatley, R.H.M., Mathias, S.F.M., 1991. Materials science and
`the production of shelf stable biologicals. Pharm. Technol. Int. 3,
`24–34.
`Greenstein, J.P., Winitz, M, 1961. Metal and salt combinations with amino
`acids. In: Chemistry of the Amino Acids, Vol. 1, Wiley, New York, pp.
`560–681.
`Lamfrom, H., Nielsen, O., 1970. The iron catalysis of thioglycolate
`oxidation by oxygen. J. Am. Chem. Soc. 79, 1966–1970.
`
`MacKenzie, A.P., 1966. Basic principles of freeze-drying for pharma-
`ceuticals. Bull. Parenteral Drug Assoc. 20, 101–129.
`MacKenzie, A.P., 1977. The physico-chemical basis for the freeze-drying
`process. Dev. Biol. Stand. 36, 51–67.
`MacKenzie, A.P., 1985. Fundamentals and applications of freeze-drying
`to biological materials, drugs and foodstuffs. In: Proceedings of the
`Meeting of the Commission C1, Tokyo, Japan, Refrigeration Science
`and Technology, pp. 21–34.
`¨Osterberg, T., Fatouros, A., Mikaelsson, M., 1997. Development of
`freeze-dried formulation of recombinant Factor VIII SQ. Pharm. Res.
`14, 892–898.
`Pikal, M.J., 1990. Freeze-drying of proteins. Part II: Formulation selec-
`tion. BioPharm 3, 26–30.
`¨
`Shihong, L., Schoneich, C., Wilson, G.S., Borchardt, R.T., 1993. Chemi-
`cal pathways of peptide degradation. V. Ascorbic acid promotes rather
`than inhibits the oxidation of methionine to methionine sulfoxide in
`small model peptides. Pharm. Res. 10, 1572–1579.
`Sundberg, R.J., Martin, B.R., 1974. Interactions of histidine and other
`imidazole derivatives with transition metal
`ions in chemical and
`biological systems. Chem. Rev. 74, 471–517.
`Taylor, M.J., 1981. The meaning of pH at low temperatures. Cryo-letters
`2, 231–239.
`
`8
`
`