`Implications
`
`LIAN Yu,* NATHANIEL MILTON, EDWARD G. GROLEAU, DINESH S. MISHRA, AND RONALD E. VANSICKLE
`
`Contribution from Eli Lilly and Company, Lilly Research Laboratories, Indianapolis, Indiana 46285.
`
`Received August 7, 1998. Accepted for publication November 10, 1998.
`
`Abstract q We report thermal and crystallographic evidence for a
`previously unknown mannitol hydrate that is formed in the process of
`freeze-drying. The mannitol hydrate was produced by freeze-drying
`pure mannitol solutions (1-4% w/v) using the following cycle: (1)
`equilibration at —5 °C for 1 h; (2) freezing at —40 °C; (3) primary
`drying at —10 °C for 15 h; and (4) secondary drying at 10 °C for 2
`h and then 25 °C for 5 h. This crystal form was also observed upon
`freeze-drying in the presence of sorbitol (1% w/v). The mannitol hydrate
`showed a distinct X-ray powder diffraction pattern, low melting point,
`and steplike desolvation behavior that is characteristic of crystalline
`hydrates. The mannitol hydrate was found to be metastable, converting
`to anhydrous polymorphs of mannitol upon heating and exposure to
`moisture. The amount of the mannitol hydrate varied significantly from
`vial to vial, even within the same batch. The formation of mannitol
`hydrate has several potential consequences: (1) reduced drying rate;
`(2) redistribution of the residual hydrate water during accelerated
`storage to the amorphous drug; and (3) vial-to-vial variation of the
`moisture level.
`
`Introduction
`The crystalline)-11 and amorphous12 forms of n-mannitol,
`a commonly used pharmaceutical excipient, have been
`extensively studied. The crystallization and polymorphic
`behaviors of mannitol have also been investigated in frozen
`aqueous solutions,13-17 with an aim to understand and
`control the freeze-drying process. Unlike many excipients
`(e.g., sorbitol and disaccharides), mannitol has a strong
`tendency to crystallize from a frozen aqueous solution, both
`during cooling and reheating. The vial breakage phenom-
`enon13,14 is a striking illustration of this tendency. Mannitol
`has been observed to continue to crystallize after freeze-
`drying, especially as a result of heat and moisture,18,19
`which indicates that the freeze-drying process can produce
`a partially amorphous and partially crystalline material.
`The crystallization of mannitol during freeze-drying can
`lead to different anhydrous polymorphs (cc, )3, and 6) and
`their mixtures as a result of different formulation and
`processing conditions,15 which leaves room for polymorphic
`transformations during storage. To the best of our knowl-
`edge, hydrated crystal forms of mannitol have not been
`reported.2°
`The formation of a crystalline hydrate by an excipient
`during freeze-drying may have several practical conse-
`quences. The difficulty of removing bound water from a
`crystal lattice can significantly limit the drying rate.
`Certain hydrates lose water without significantly altering
`the initial lattice structures ("isomorphic dehydration")21
`and lead to materials that are highly hygroscopic and
`
`Corresponding author. Tel. (317) 276 1448. Fax (317) 277 5519.
`E-mail yu_lian@lilly.com.
`
`undergo structural relaxation during storage. The residual
`water that is not removed by freeze-drying may be a
`potential threat to product stability if it is released during
`storage, especially under "accelerated" conditions.
`We present here thermal and crystallographic evidence
`for a mannitol hydrate that is formed during freeze-drying
`and discuss the practical implications in terms of process
`design and product stability.
`
`Experimental Section
`Freeze-Drying—A laboratory freeze-drier (FTS Systems Inc.)
`was used. The following conditions produced the "mannitol
`hydrate" as characterized below: (1) equilibration at —5 °C for 1
`h; (2) freezing at —40 °C; (3) primary drying at —10 °C for 15 h;
`and (4) secondary drying at 10 °C for 2 h and then 25 °C for 5 h.
`The cooling rate not controlled during freezing, and it took
`approximately 3 h to reach —40 °C. The chamber pressure was
`set to 100 ,um of Hg throughout the drying process. Mannitol
`solutions were prepared at several concentrations (1, 2, and 4 w/v
`%) by dissolving mannitol (99+%, ACS reagent, Sigma; USP
`quality) in deionized water. The freeze-drying vials (tubing type
`manufactured by Wheaton) were 5 mL with 18.4 mm i.d. The fill
`volume was 2.0 mL/vial. The samples were stored at —20 °C before
`analysis.
`XRD—A Siemens D5000 X-ray diffractometer was used, which
`was equipped with a Cu Ka source(.. = 1.54056 A) operating at a
`tube load of 50 kV and 40 mA. The divergence slit size was 0.6
`mm, the receiving slit 1 mm, and the detector slit 0.1 mm. Data
`were collected by a Kevex solid-state (SiLi) detector. The freeze-
`dried cake was broken, spread over the sample holder, and gently
`pressed before analysis. Each sample was scanned between 4 and
`35° (29) with a step size of 0.03° and a maximum scan rate of 2
`s/step.
`DSC—Differential scanning calorimetry (DSC) was conducted
`in sealed Al pans at 10 °C/min using a Seiko DSC 210 under 50
`mL/min nitrogen purge. Samples (5-10 mg) were either loosely
`packed into sample pans or first pressed into pellets using a
`stainless steel pellet-maker of local design. Sample preparation
`was carried out in a dry glovebag maintained at <5% RH.
`TG/DTA—Simultaneous thermal gravimetric analysis (TGA)
`and differential temperature analysis (DTA) were conducted at
`10 °C/min in open Al pans using a Seiko TG/DTA 220 under 150
`mL/min nitrogen purge. Three to five milligrams were used for
`each analysis.
`
`Results and Discussions
`Figure 1 (curve 1) shows the XRD pattern of a freeze-
`dried mannitol sample from a 4% w/v solution. In addition
`to peaks that belong to the 6 (major) and )3 (minor)
`polymorphs of mannito1,22 additional peaks (marked by
`asterisks and listed in Table 1) were observed that could
`not be attributed to any known mannitol polymorphs.
`Heating this sample to 70 °C for 30 min eliminated the
`additional peaks (Figure 1, curve 2), with the remaining
`peaks attributable to the 6 and )3 mannitol. These observa-
`tions indicate the existence of a metastable crystal form of
`mannitol that was produced during freeze-drying and
`
`196 / Journal of Pharmaceutical Sciences
`Vol. 88, No. 2, February 1999
`
`10.1021/js980323h CCC: $18.00 (cid:9)
`Published on Web 12/19/1998 (cid:9)
`
`c 1999, American Chemical Society and
`American Pharmaceutical Association
`
`Mylan Ex 1049, Page 1
`
`(cid:9)
`(cid:9)
`
`
`the loss of solvent (water). The steplike TGA loss and the
`well-defined DTA endotherm suggest the removal of struc-
`tural water from a crystalline hydrate, rather than loosely
`bound ("free") water. This interpretation is supported by
`the sealed-pan DSC data. The use of sealed sample pans
`prevented the simultaneous evaporation of water during
`melting, making it possible to observe a sharp, homoge-
`neous melting of the crystalline hydrate. The broad endot-
`hermic event following the sharp hydrate melting in the
`DSC trace can be attributed to the temperature-depressed
`melting of anhydrous mannitol crystals in the presence of
`water. We conclude therefore that a mannitol hydrate was
`formed in the process of freeze-drying and survived what
`appeared a "typical" drying cycle. This crystal form was
`metastable, converting to anhydrous polymorph(s) of man-
`nitol upon heating.
`Conditions of Formation—Using the same drying
`cycle, we have observed the mannitol hydrate under
`different formulation conditions: mannitol concentrations
`ranging from 1 to 4% w/v, with or without sorbitol (1% w/v),
`and in a drug formulation.23 The same X-ray pattern
`assigned to the mannitol hydrate (Table 1) has also been
`observed by Cavatur and Suryanarayanan using in situ
`powder X-ray diffractometry.24
`Anhydrous mannitol crystals are nonhygroscopic at room
`temperature, gaining less than 1% moisture at 90% RH.25
`This indicates that hydrate formation from anhydrous
`crystals is unlikely upon moisture exposure at the room
`temperature. The DSC data of Martini et al.'s indicate that
`upon cooling a 10% w/v mannitol solution, ice forms (with
`substantial supercooling) before mannitol crystallizes.
`These observations, along with the previous failure to
`crystallize the mannitol hydrate at the room temperature
`or above,20 suggest that the formation of the mannitol
`hydrate is likely a low-temperature phenomenon, relevant
`in particular to freeze-drying.
`We have observed significant vial-to-vial variations in
`the amount and stability of the mannitol hydrate, even
`within the same batch. The relative intensity of the hydrate
`pattern ranged from comparable to that shown in curve 1
`(Figure 1) to barely detectable or rapidly diminishing
`during measurement; the TGA losses ranged from 1 to 6%,
`and the onset of TGA loss ranged from 40 to 60 °C. The
`mannitol hydrate produced in certain vials could withstand
`mild heat (40 °C), humidity (60% RH) and compression (in
`DSC sample preparation), while in others it was easily
`destroyed, even by gentle compression. Although the drying
`cycle used in this study seemed a reasonable one, more
`aggressive secondary drying (same temperature but longer
`drying time) did eliminate traces of the mannitol hydrate
`and resulted in anhydrous mannitol crystals.
`Several factors may contribute to the vial-to-vial varia-
`tion in the amount and stability of the mannitol hydrate.
`First, the hydrate is formed in a low-temperature, concen-
`trated solution, conditions that are not favorable for
`producing well-developed crystals in a reasonable process-
`ing time. Second, the ice crystallization temperature can
`vary as a result of the cooling ratel° and thus the vial
`location in a drier. Other more subtle factors can also affect
`the onset of ice crystallization (e.g., defects on the vial
`surface). Such variability can influence the degree of freeze
`concentration and the ice structure and, in turn, the
`subsequent crystallization of mannitol. Finally, the hydrate
`formation is in competition with the crystallization of the
`anhydrous polymorphs (a, )3, or 6), directly or through
`subsequent solid—solid conversion. Depending on the rate
`of anhydrous crystallization, the hydrate amount will vary.
`We suspect that these effects acting in concert contributed
`to the vial-to-vial variation and the previous failure to
`detect the mannitol hydrate.
`
`Journal of Pharmaceutical Sciences /197
`Vol. 88, No. 2, February 1999
`
`1, ;)initial
`ki41,, (cid:9)
`L (cid:9)
`4,
`
`ti A
`V V (cid:9)
`
`1.1 .111 (cid:9)
`
`I
`
`I (cid:9)
`
`I.
`
`after heating
`,o-.0.,,i4-(n+-4
`I .
`
`I (cid:9)
`
`• (cid:9)
`
`I (cid:9)
`
`I (cid:9)
`
`il
`
`. (cid:9)
`
`15 (cid:9)
`
`25 (cid:9)
`
`I (cid:9)
`30 (cid:9)
`
`I (cid:9) I
`
`35
`
`2
`
`13
`5
`a
`
`5 (cid:9)
`
`•
`10 (cid:9)
`
`20 (cid:9)
`20(°)
`Figure 1—X-ray powder diffraction patterns of a freeze-dried mannitol sample
`(curve 1) and the same sample after heating at 70 °C for 30 min (curve 2).
`Curve 1 contains peaks that do not belong to any known mannitol polymorphs,
`whereas all peaks in curve 2 are attributable to the 6 and /3 polymorphs of
`mannitol. The a-polymorph pattern is also shown for comparison.
`
`Table 1—X-ray Powder Pattern for the Mannitol Hydrate
`
`27.0
`25.7
`17.9
`16.5
`9.6
`20, deg
`16
`0.2
`0.3
`0.2
`0.83
`Illo
`3 Overlapping the (020) reflection of 6-mannitol (20 = 9.7°). b Best i.d.
`peak.
`
`4, 6% wt. loss
`
`TGA
`
`7 (cid:9)
`
`DTA (open pan)
`
`DSC (sealed pan)
`
`endothermic
`
`160
`
`200
`
`40 (cid:9)
`
`80 (cid:9)
`
`120
`T (°C)
`Figure 2—TGA (top), DTA (middle), and DSC (bottom) data of a freeze-dried
`mannitol sample. The TGA and DTA data were recorded simultaneously in
`an open sample pan, which show the melt/desolvation behavior of a crystalline
`hydrate. The DSC data was recorded in a hermetically sealed sample pan,
`which shows the homogeneous melting of the hydrate.
`
`capable of converting to anhydrous mannitol polymorphs
`upon heating.
`Figure 2 shows the thermal characteristics of the same
`sample described above. TGA (top curve) showed a steplike
`weight loss near 50 °C. The simultaneously conducted DTA
`(middle curve) showed a well-defined endotherm coinciding
`with the weight loss, which was followed by the melting of
`the anhydrous mannitol (mp 169 °C). DSC conducted in
`hermetically sealed pan (bottom curve) showed a sharp
`endotherm slightly below the desolvation onset.
`The thermal data indicate that the crystal form trans-
`formation during heating (Figure 1) was accompanied by
`
`Mylan Ex 1049, Page 2
`
`(cid:9)
`(cid:9)
`
`
`Practical Implications—The formation of the mannitol
`hydrate during freeze-drying has several practical implica-
`tions. First, the rate of water removal may be reduced due
`to strongly bound water that resides in the crystal lattice.
`In certain hydrates, such as the mannitol hydrate, the
`structural water cannot be fully removed without destroy-
`ing the crystal lattice, a process that may require more
`aggressive or atypical drying conditions (e.g., longer drying
`time and higher temperature). Depending on the detailed
`mechanism of formation, it may be necessary, for example,
`to anneal the frozen solution to promote crystallization of
`the anhydrous form, thereby reducing or eliminating the
`mannitol hydrate.
`Second, the residual hydrate water that is not fully
`removed by freeze-drying may pose a long-term threat to
`product stability, as the hydrate water can be released and
`redistributed, especially in "accelerated" storage, to the
`often amorphous drug and thus increases the potential for
`chemical and physical changes. Herman et al. have shown
`that methylprednisolone sodium succinate (MPSS) stored
`at accelerated conditions and formulated with mannitol is
`chemically less stable than materials formulated with
`lactose.26 The difference in chemical stability was at-
`tributed to mannitol crystallization and the redistribution
`of water within the freeze-dried cake which promoted the
`hydrolysis of MPSS. Roos and Karel have shown that as
`amorphous lactose crystallizes and releases water, the glass
`transition temperature of the remaining amorphous mate-
`rial is depressed and the crystallization process acceler-
`ated.27 It is conceivable, therefore, that the release of
`hydrate water during accelerated storage may also com-
`plicate the interpretation of stability data and cause a
`premature termination of an otherwise promising formula-
`tion.
`Finally, our observations indicate that certain vial-to-
`vial variation in the water level may be associated with
`the formation of the mannitol hydrate.
`
`Conclusions
`
`We have reported thermal and crystallographic evidence
`for the formation of a previously unknown mannitol
`hydrate in the freeze-drying process. The mannitol hydrate
`can survive a "typical" drying cycle, but can be converted
`to anhydrous polymorph(s) of mannitol on gentle heating
`and more aggressive secondary drying. Potential conse-
`quences of the mannitol hydration formation include
`reduced drying rate, moisture release during accelerated
`storage, and vial-to-vial variation in the water level.
`
`References and Notes
`1. Marwick, T. C. An X-ray Study of Mannitol, Dulcitol and
`Mannose. Nature 1931, 127, 11-12.
`2. McCrea, G. W. An X-ray Study of Mannitol. Nature 1931,
`127, 162-163.
`3. Rye, A.; Sorum, H. Crystalline Modifications of D-Mannitol.
`Acta Chem. Scand. 1952, 6, 1128-1129.
`4. Berman, H. M.; Jeffrey, G. A.; Rosentein, R. D. The Crystal
`Structures of the a' and /3 Forms of D-Mannitol. Acta
`Crystallogr. 1968, B24, 442-449.
`5. Kim, H. S.; Jeffrey, G. A.; Rosenstein, R. D. The Crystal
`Structure of the K Form of D-Mannitol. Acta Crystallogr.
`1968, B24, 1449-1455.
`6. Walter-Levy, L. Sur les varietes crystallines du D-mannitol.
`C. R. Acad. Sc. Paris 1968, 267, 1779-1782.
`7. Debord, B.; Lefebvre, C.; Guyot-Hermann, A. M.; Hubert, J.;
`Bouche, R.; Guyot, J. C. Study of Different Crystalline Forms
`of Mannitol: Comparative Behaviour under Compression.
`Drug Dev. Ind. Pharm. 1987, 13, 1533-1546.
`8. Jones, F. T.; Lee, K. S. The Optical and Crystallographic
`Properties of Three Phases of Mannitol. Microscope 1970,
`18, 279-285.
`
`198 / Journal of Pharmaceutical Sciences
`Vol. 88, No. 2, February 1999
`
`9. Grindley, T. B.; McKinnon, M. S.; Wasylishen, R. E. Towards
`Understanding 13C-N.M.R. chemical shifts of Carbohydrates
`in the Solid State. The Spectra of D-Mannitol Polymorphs
`and of DL-Mannitol. Carbohydr. Res. 1990, 197, 41-52.
`
`10. Pitkanen, I.; Perkkalainen, P.; Rautiainen, H. Thermoana-
`lytical Studies of Phases of D-Mannitol. Thermochim. Acta
`1993, 214, 157-162.
`
`11. Burger, A.; Hetz, S.; Weissnicht, A. On the Polymorphism
`of Mannitol. Eur. J. Pharm. Biopharm. 1994, 40(Suppl.),
`21S.
`
`12. Yu, L.; Mishra, D. S.; Rigsbee, D. R. Determination of the
`Glass Properties of D-Mannitol Using Sorbitol as an Impurity.
`J. Pharm. Sci. 1998, 87, 774-777.
`
`13. Willaims, N. A.; Lee, Y.; Polli, G. P.; Jennings, T. A. The
`Effect of Cooling Rate on Solid-Phase Transitions and
`Associated Vial Breakage Occurring in Frozen Mannitol
`Solutions. J. Parenteral Sci. Technol. 1986, 40, 135-141
`
`14. Williams, N. A.; Dean, T. Vial Breakage by Frozen Mannitol
`Solutions: Correlation with Thermal Characteristics and
`Effect of Stereoisomerism, Additives and Vial Configuration.
`J. Parenteral Sci. Technol. 1991, 45, 94-100.
`
`15. Kim, A. I.; Akers, M. J.; Nail, S. L. The Physical State of
`Mannitol after Freeze-Drying: Effects of Mannitol Concen-
`tration, Freezing Rate, and a Noncrystallizing Cosolute. J.
`Pharm. Sci. 1998, 87, 931 —935.
`
`16. Martini, A.; Kume, S.; Crivellente, M.; Artico, R. Use of
`Subambient Differential Scanning Calorimetry to Monitor
`the Frozen-State Behavior of Blends of Excipients for Freeze-
`Drying. PDA J. Pharm. Sci. Technol. 1997, 51, 62-67.
`
`17. Cavatur, R. K.; Suryanarayanan, R. Characterization of
`Frozen Aqueous Solutions by Low-Temperature X-ray Pow-
`der Diffractometry. Pharm. Res. 1998, 15, 194-199.
`
`18. Kovalcik, T.; Guillory, J. K. The Stability of Cyclophospha-
`mide in Lyophilized Cakes. Part I. Mannitol, Lactose, and
`Sodium Bicarbonate as Excipients. J. Parenteral Sci. Tech-
`nol. 1988, 42, 29-39.
`
`19. Isutsu, K.; Yoshioka, S.; Terao, T. Effect of Mannitol Crystal-
`linity on the Stabilization of Enzymes during Freeze-Drying.
`Chem. Pharm. Bull. 1994, 42, 5-8.
`
`20. Pitkanen et al. wrote in 1993, "We took efforts to crystallize
`hydrous forms, but did not succeed" (ref 10).
`
`21. Stephenson, G. A.; Groleau, E. G.; Kleemann, R. L.; Xu, W.;
`Rigsbee, D. R. Formation of Isomorphic Desolvates: Creating
`a Molecular Vacuum. J. Pharm. Sci. 1998, 87, 536-542.
`
`22. The Powder Diffraction File. International Centre for Dif-
`fraction Data, 12 Campus Blvd., Newtown Square, PA 19073-
`3273.
`
`23. Yu, L.; Milton, N. Unpublished result.
`
`24. Cavatur, R. K.; Suryanarayanan, R. Characterization of
`Phase Transitions During Freeze-Drying by in Situ Powder
`X-ray Powder Diffractometry, presentation abstract submit-
`ted to the AAPS Annual Meeting and Exposition, Nov 15-
`19, 1998, San Francisco, CA.
`
`25. Wade, A.; Weller, P. J., Eds. Handbook of Pharmaceutical
`Excipients, 2nd ed.; American Pharmaceutical Association:
`Washington, D.C., 1994.
`
`26. Herman, B. D.; Sinclair, B. D.; Milton, N.; Nail, S. L. The
`Effect of bulking Agent on the Solid-State Stability of Freeze-
`Dried Methylprednisolone Sodium Succinate. Pharm. Res.
`1994, 11, 1467-1473.
`
`27. Roos, Y.; Karel, M. Crystallization of Amorphous Lactose.
`J. Food Sci. 1992, 57, 775-777.
`
`Acknowledgments
`
`We thank Russell A. Miller for assistance in freeze-drying and
`Dr. Raj Suryanarayanan for helpful discussions.
`
`JS980323H
`
`Mylan Ex 1049, Page 3
`
`