Dehydration of Risedronate Hemi-Pentahydrate:
`Analytical and Physical Characterization
`
`CATHY LESTER, GWEN LUBEY, MICHAEL DICKS, GREGORY ANDOL, DANA VAUGHN,
`R. THOMAS CAMBRON, KATHERINE POIESZ, NANCY REDMAN-FUREY
`
`Analytical Department, Procter & Gamble Pharmaceuticals, Inc., Mason, Ohio
`
`Received 25 January 2006; revised 22 March 2006; accepted 12 April 2006
`
`Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20662
`
`ABSTRACT: Dehydration of hydrates of pharmaceutical active ingredients (pharma-
`ceutical hydrates) may easily occur during storage or manufacturing. Loss of water may
`have little effect on the crystal lattice, produce less hydrated forms or possibly amorphous
`forms. Characterizing the effects of water loss on crystal hydrate forms is important for
`understanding the behavior of pharmaceutical hydrates throughout the manufacturing
`and storage processes. This study shows that exposure of the hemi-pentahydrate form of
`risedronate monosodium to gentle heating (608C) or conditions of low relative humidity
`(<10% RH) results in the loss of 1 mole of channel-type water. Upon removal of the
`channel-type water, the crystal lattice adjusts producing a distinct phase characterized
`by X-ray, thermal, IR, Raman, and NMR data. Adjustment of the crystal lattice appears
`to compromise crystal integrity and can result in reduced crystallite and particle sizes.
`ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 95:2631–2644, 2006
`Keywords: hydrate; dehydration; solid state; X-ray diffractometry; Risedronate;
`hemi-pentahydrate; FTIR; Raman spectroscopy; solid-state NMR; thermal analysis
`
`INTRODUCTION
`
`Environmental variables encountered during the
`drug manufacturing process may effect the for-
`mation of different crystalline states of hydration
`for the active pharmaceutical ingredient of a drug
`product.1 These different solid forms or hydrates
`can possess different physical properties such as
`differences in solubility, stability, bioavailability,
`dissolution rate, and particle habit. Water in
`pharmaceutical hydrates can be described by
`three different structural classes that include
`those residing in isolated lattice sites,
`lattice
`channel sites, or ion-coordinated sites.2,3 In iso-
`lated lattice sites, water molecules are isolated
`from other water molecules due to contact with
`drug molecules. Water molecules forming lattice
`channel sites are in contact with other water
`molecules of adjoining unit cells along an axis of a
`
`Correspondence to: Nancy Redman-Furey (Telephone:
`607-335-2601; Fax: 607-335-2300;
`E-mail: redmanfurey.nl@pg.com)
`
`Journal of Pharmaceutical Sciences, Vol. 95, 2631–2644 (2006)
`ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association
`
`unit cell. It has been shown that some channel
`water containing hydrates may undergo dehydra-
`tion under conditions of low relative humidity
`(RH) or pick up water under conditions of high
`RH.3 Ion-coordinated water participates in an ion-
`water bond which usually is much stronger than
`any hydrogen bonds present. In addition to the
`formation of different hydrates, a single hydrate
`form of an API may contain more than one struc-
`tural class of water.2
`Characterization of different hydrate forms
`the bisphosphonate compound risedronate
`of
`[1-hydroxy-2-(3-pyridinyl)ethylidene] bis [phos-
`phonic acid] monosodium salt, which is prescribed
`for the treatment of osteoporosis, was recently
`reported.4 Three different hydrate forms were
`characterized including a monohydrate, a hemi-
`pentahydrate, and a variable hydrate containing
`between 4 and 6 moles of water. The hemi-
`pentahydrate form, which is the commercial form,
`was found to contain two classes of water mole-
`cules including at least 1 mole of lattice-type
`(possibly ion-coordinated) and 1 mole of channel
`water. Channel water is characteristically mobile
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`2631
`
`Merck Exhibit 2240, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2632
`
`LESTER ET AL.
`
`and may migrate into and out of the crystal lattice
`as a function of ambient humidity. This mobility
`was illustrated previously with vapor sorption
`data presented for the hemi-pentahydrate form
`of risedronate.4 Over a RH range of 20–90%,
`the water content remains stable. As the RH is
`decreased from 9% to 5%, a weight loss of 5% occurs
`corresponding to the loss of 1 mole of water. The
`channel water then reenters the lattice as the
`humidity is increased from 13% to 20% resulting in
`complete rehydration.
`In contrast to the removal of lattice water, the
`removal of channel water often leaves the crystal
`lattice relatively intact.3 However, upon the
`removal of channel water from risedronate hemi-
`pentahydrate, thermal and spectroscopic data
`indicate the lattice undergoes an adjustment and
`the crystal integrity appears to be compromised.
`This study describes the formation of dehydrated
`risedronate by removing 1 mole of channel water
`with heat or desiccation and discusses the asso-
`ciated physical and chemical changes that occur.
`These changes are shown to be reversible, and the
`crystal lattice returns to its original state as the
`sample rehydrates.
`
`EXPERIMENTAL
`
`Materials
`
`The hemi-pentahydrate was sourced from P&G
`Pharmaceuticals, Inc. commercial supply and
`used as received. The dehydrated material was
`prepared by desiccating hemi-pentahydrate over
`anhydrous calcium sulfate for at least 48 h.
`Rehydration was accomplished by exposing the
`dehydrated material to RH of >20% until rehy-
`dration was complete, typically within hours.
`
`Analytical Methodology
`
`Thermal Analysis
`
`Simultaneous thermogravimetry and differential
`thermal analysis curves (TGA/DTA) were gener-
`ated using a Seiko SSC/5200 custom equipped
`with a quartz glass window.5,6 Samples (approxi-
`mately 10 mg) were scanned under a dry nitrogen
`purge from 25 to 2508C at 58C/min. Photomicro-
`graphs were obtained by mounting a microscope
`with a video feed above the quartz glass window.
`Acceptable depth of field and focus was achieved
`using top illumination, a 0.5 objective on the
`
`microscopy, and 10 magnification in the eye-
`pieces and camera feed.
`
`Infrared and Raman Spectroscopy
`
`Fourier transform infrared spectra (FTIR) were
`obtained using a BioRad FTS-3000 spectrometer
`with 4/cm resolution. Sample desiccation was
`minimized by dispersing materials in both Fluor-
`olube (4000–1350/cm) and Nujol (1350–400/cm)
`mulling agents. This sample preparation techni-
`que enables the collection of infrared spectra from
`the hemi-pentahydrate and dehydrated forms of
`risedronate in their native states. Fourier trans-
`form Raman spectra (FT-Raman) were obtained
`using a Nicolet FT Raman 960 spectrometer at 8/
`cm resolution. Localized heating due to absorp-
`tion of the laser light is known to impact Raman
`spectra of the hemi-pentahydrate and dehydrat-
`ed forms of risedronate. The laser intensity
`employed for spectral acquisition (0.5 W) and
`cumulative sample exposure (16 scans) were
`below the threshold known to cause spectral
`changes due to localized heating.4 Samples were
`prepared by placing material to be analyzed into
`small quartz tubes and illuminating the sample
`with the laser only during spectral acquisition.
`
`Solid-State NMR Spectroscopy
`
`Cross-polarization/magic-angle spinning (CP/MAS)
`solid-state NMR (SSNMR) spectra were obtained
`using a Varian UnityINOVA 300 NMR spectro-
`meter equipped with a Varian 7 mm CP/MAS
`probe.7 Each sample was
`characterized by
`121.4 MHz 31P, 75.4 MHz 13C, and 79.4 MHz
`23Na SSNMR spectroscopy. Chemical shifts were
`referenced externally for phosphorus to phospho-
`ric acid, 85 weight % (neat) at 0.0 ppm; for carbon
`to hexamethylbenzene at 17.3 ppm;7 and for
`sodium to 0.1 M NaCl (aq) at 0.0 ppm. The
`risedronate samples were not ground, and were
`packed into 7 mm silicon nitride rotors fitted with
`Torlon caps and spun at the rate of 5 kHz.
`Dehydration of hemi-pentahydrate was moni-
`tored first in situ by 31P SSNMR spectroscopy
`using a rotor with a vented cap. Spectra were
`collected during probe heating from 20 to 608C
`(108C increments) and then during a 608C hold,
`until no further changes were detected in the
`spectra. Dehydrated risedronate samples created
`by heating at 608C in an oven or by desiccation
`were packed using a glove box under a dry
`nitrogen gas purge. CP/MAS 31P and 13C spectra
`were obtained for each sample. 13C spectra were
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`DOI 10.1002/jps
`
`Merck Exhibit 2240, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`DEHYDRATION OF RISEDRONATE HEMI-PENTAHYDRATE: ANALYTICAL AND PHYSICAL CHARACTERIZATION
`
`2633
`
`recorded with and without sideband suppression
`(TOSS8); also with and without proton dephas-
`ing.9 For each material, the same Hartman-Hahn
`match was used, and the contact time was optimi-
`zed for maximum signal intensity.10 23Na spectra
`were collected using a Bloch decay with a 308
`excitation pulse and a 10 s relaxation delay.
`31P sideband intensities were analyzed by the
`Herzfeld–Berger method11 using the HBA program12
`to obtain chemical shift tensor information.
`
`X-Ray Diffraction
`
`X-ray powder diffraction was performed on the
`samples using the Bruker D5000 X-ray diffract-
`ometer. The D5000 was equipped with a 2.2 kW
`Cu anode X-ray tube, an Anton Parr TTK-1 low
`temperature stage, and high-speed position sensi-
`tive detector (PSD). Cu Ka radiation (l¼ 1.5418 A˚ )
`was used to obtain all powder patterns. A dual
`foil, nickel filter was placed in the receiving path
`the X-rays to remove the Kb radiation.
`of
`Risedronate sodium, hemi-pentahydrate material
`was mounted and analyzed on a front loading
`sample holder, without any special sample pre-
`paration. Environmental conditions for the ana-
`lysis were manipulated to facilitate drying and
`rehydration of the sample without removing it
`from the instrument. The sample was dehydrated
`by heating the material to 608C and holding for
`duration of the analysis. After dehydration, it was
`re-hydrated by allowing the material to cool and
`stabilize at room RH for 20 min. All scans were
`performed over the range of 3.5–408 2 theta, at a
`0.028 step size for 0.2 s/step.
`
`Light Microscopy
`
`Light micrographs were obtained using a Nikon
`Eclipse e600 Polarizing Light Microscope (PLM)
`with an Optronics 3-Chip color camera. Slides
`were prepared by mixing the powdered sample
`with low viscosity immersion oil, and the result-
`ing dispersion placed between a clean glass slide
`and cover slip. Each prepared slide was examined
`using brightfield and slightly uncrossed polarized
`light using a Nikon PlanFluor 10/0.30 objective.
`
`Particle Size Analysis
`
`Particle size of the hemi-pentahydrate and dehy-
`drated forms of risedronate was determined using
`laser diffraction. Approximately 50 mg of each
`sample was introduced as a dry powder into
`
`Isopar V, a synthetic isoparaffinic oil. The result-
`ing dispersions were analyzed using a Horiba
`LA-920.
`
`RESULTS AND DISCUSSION
`
`Thermal Analysis
`
`Comparing fully hydrated risedronate to desic-
`cated risedronate via thermal analysis cle-
`arly illustrates that desiccation results in loss of
`1 mole of channel water from the molecule while
`leaving the remaining 1.5 moles of water intact.
`Representative thermal curves are provided
`below in Figure 1. The only difference in the
`TGA curves (Fig. 1A) is the initial mass loss of
`5.2% observed from room temperature to approxi-
`mately 708C in the fully hydrated sample that is
`missing completely from the desiccated sample.
`This value is in agreement with theory for loss of
`1 mole of water from the hemi-pentahydrate
`(5.16%). The higher temperature mass losses of
`the two samples match in both temperature and
`magnitude. A detailed interpretation of the TGA
`losses was provided in an earlier publication.13,14
`In the earlier study, the dehydration was assign-
`ed to loss of channel water based upon the
`temperature of the water loss (below the boiling
`point of water) and the observed variation in the
`loss profile as a function of scan rate and venting.
`In this study, the assignment was confirmed by
`the disappearance of this first dehydration step in
`the desiccated sample. Collectively, these data
`show how the 1 mole of channel water may be
`quantitatively drawn out of the crystal lattice
`either by gentle heating (as demonstrated by
`the TGA curve of the fully hydrated sample) or by
`desiccation at room temperature.
`Unlike the TGA comparison, differences are
`observed in the DSC curves in addition to those
`anticipated for simple loss of the mole of channel-
`type water of hydration (Fig. 1B). The desiccated
`sample yielded a flat baseline through 908C, over
`the same temperature range that a broad endo-
`therm due to dehydration is observed for the fully
`hydrated sample,
`just as expected. At higher
`temperatures, additional changes are observed.
`The loss of the final mole of lattice-type water of
`hydration from the hemi-pentahydrate occurs just
`above 1408C.13,14 In the desiccated sample, this
`dehydration occurs as a much sharper and larger
`endotherm than that observed for the fully
`hydrated sample. Subsequent endotherms due to
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`Merck Exhibit 2240, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2634
`
`LESTER ET AL.
`
`Figure 1. A: Comparison of TGA curves for untreated and desiccated samples of
`risedronate. Loss of 1 mole of channel water observed by 708C in untreated sample but not
`desiccated sample. B: Comparison of DSC curves. Note lack of initial dehydration
`endotherm and difference in dehydration endotherm near 1508C for the desiccated
`sample.
`
`dehydration of monohydrate formed in situ and
`degradation match between the untreated and
`desiccated samples.13,14 The change in dehydra-
`tion profile for loss of water from the isolated
`lattice site suggests that channel water loss upon
`desiccation results in a lattice adjustment in
`response to the missing water.
`Use of a TGA/DTA equipped with a quartz glass
`window within the furnace wall and a microscope,
`enabled visualization of the impact the loss of
`channel water from the crystal. The series of
`photomicrographs provided in Figure 2 visually
`illustrate the macroscopic impact of drying upon
`the crystal and correspond to the first mass loss
`step observed in Figure 1A. Initially (258C), the
`
`hemi-pentahydrate crystals are essentially clear
`with some noticeable internal fracturing. As the
`channel water is driven out of the crystal, it
`becomes more and more opaque, indicating some
`type of lattice adjustment to the loss of water and
`result fracturing. This is clearly beginning to occur
`with the initial onset of mass loss at 368C. As the
`last of the channel water exits the crystal, notice-
`able expansion can be seen as the result of
`continued fracturing (56–758C).
`
`Infrared and Raman Spectroscopy
`
`Infrared spectroscopy is particularly sensitive
`to the polar functional groups observed in
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`DOI 10.1002/jps
`
`Merck Exhibit 2240, Page 4
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`DEHYDRATION OF RISEDRONATE HEMI-PENTAHYDRATE: ANALYTICAL AND PHYSICAL CHARACTERIZATION
`
`2635
`
`Figure 2. TGA/DTA/Microscopy photos of the impact of drying upon the hemi-
`pentahydrate. As the channel water driven off the crystal becomes more opaque
`(beginning at 368C) and as the last of the channel water is driven off, the crystal expands
`in volume (56–758C).
`
`risedronate as well as water of hydration and its
`effect on hydrogen bonding. The infrared spec-
`trum of the dehydrated form of risedronate
`compared to spectra from the hemi-pentahydrate
`and monohydrate forms of the material exhibit
`significant differences.4 A comparison of infrared
`spectra from the hemi-pentahydrate and dehy-
`drated forms of risedronate is illustrated in
`Figure 3A. There are significant changes in the
`OH stretch region (3000 to 3600/cm) as well
`as throughout the spectra indicating significant
`differences in hydrogen bonding characteristics
`within the crystal lattice between the two dif-
`ferent hydration states. The hemi-pentahydrate
`form exhibits two sharp, distinct, OH stretch
`peaks at 3566/cm and 3618/cm. The shape and
`location of these peaks are indicative of con-
`strained water in the crystal
`lattice in two
`separate environments. Dehydration of the hemi-
`pentahydrate form results in the disappearance of
`the peak at 3618/cm producing spectral evidence
`indicating water was removed from discrete
`locations in the crystal lattice. These spectral
`changes associated with dehydration of the hemi-
`pentahydrate form are easily observed by repeat-
`ed spectral acquisition during sample dehydra-
`
`tion (Fig. 3B) and are reversible upon rehydration
`of the material.
`The O––POH group produces broad IR bands
`in the spectral region between 2000 and 2725/cm
`that are complex with multiple overlapping peaks
`due to multiple O––POH groups in the molecule.
`Changes observed upon dehydration of hemi-
`pentahydrate show shifting and changes in rela-
`tive peak intensities in this spectral region.
`The largest O––POH peak at 2105/cm is more
`intense in the hemi-pentahydrate spectrum than
`the dehydrated form possibly due to delocalization
`of hydrogen bonding associated with the hydrat-
`ion state of the material. The fingerprint region of
`the spectrum between 900 and 1215/cm is domi-
`nated by peaks characteristic of the phosphonic
`acid group overlapped with pyridine ring deforma-
`tion bands. Dehydration of the hemi-pentahydrate
`form produces a very large decrease in intensity at
`approximately 1150/cm associated with the asym-
`metric P––O stretch. The symmetric P––O stretch
`observed as an intense shoulder band at approxi-
`mately 1103/cm in the hemi-pentahydrate broad-
`ens and shifts in frequency due to dehydration.
`Differences observed in spectral features asso-
`ciated with the O––POH and P––O groups
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`Merck Exhibit 2240, Page 5
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2636
`
`LESTER ET AL.
`
`Figure 3. A: Infrared spectra of the hemi-pentahydrate and dehydrated forms of
`risedronate. B: Infrared spectra collected during the dehydration of risedronate hemi-
`pentahydrate.
`
`between the hemi-pentahydrate and dehydrated
`forms result from changes in hydrogen bonding
`due to loss of water.
`Raman spectroscopy is particularly sensitive
`to the underlying carbon-based skeleton of the
`risedronate molecule in addition to low frequency
`vibrational modes delocalized across multiple unit
`cells in the crystal lattice. The Raman spectrum
`from the dehydrated form of risedronate compared
`to spectra from the hemi-pentahydrate and mono-
`hydrate forms of risedronate exhibit significant
`differences.4 A comparison of Raman spectra from
`the hemi-pentahydrate and dehydrated forms is
`illustrated in Figure 4A. Raman spectra from
`risedronate are dominated by peaks associated
`with the substituted pyridine ring in the spectrum
`between approximately 1000 and 1075/cm. Spec-
`tral differences in peaks associated with the
`substituted pyridine ring exhibit numerous dif-
`ferences between the hemi-pentahydrate and
`dehydrated forms. Both spectral broadening and
`shifting in the pyridine degenerate ring mode
`between approximately 1610 and 1650/cm in
`addition to changes in relative peak intensity
`associated with the ring stretching/bending modes
`
`at 1024/cm and 1055/cm are observed due to
`dehydration. These spectral differences are con-
`sistent with a change in environment of the
`substituted pyridine ring induced by dehydration.
`Additional spectral evidence indicating different
`pyridine ring environments between these forms is
`also observed as splitting in the aromatic CH
`stretch region at approximately 3095/cm combined
`with a slight shift to lower frequency for a CH
`stretch from 3001 to 2995/cm (Fig. 4).
`Significant changes occur in the aliphatic CH
`stretch region of the Raman spectra as the hemi-
`pentahydrate form is dehydrated. The disappear-
`ance of the CH stretch at 2965/cm is inversely
`correlated to the appearance of a new peak at 2937/
`cm as illustrated in Figure 4B. Similar inverse
`correlations between the disappearance and
`growth of spectral peaks are observed throughout
`the Raman spectrum providing evidence for the
`presence of two discrete solid-state forms (hemi-
`pentahydrate and dehydrated) during the dehy-
`dration of hemi-pentahydrate. Spectral changes
`observed during the dehydration of hemi-penta-
`hydrate are reversible under conditions where
`hemi-pentahydrate is the equilibrium form.
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`DOI 10.1002/jps
`
`Merck Exhibit 2240, Page 6
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`DEHYDRATION OF RISEDRONATE HEMI-PENTAHYDRATE: ANALYTICAL AND PHYSICAL CHARACTERIZATION
`
`2637
`
`Figure 4. A: Raman spectra of the hemi-pentahydrate and dehydrated forms of
`risedronate. B: Raman spectra collected during the dehydration of risedronate hemi-
`pentahydrate.
`
`Solid-State NMR Spectroscopy
`
`Chemical shift in NMR spectroscopy depends
`upon the electronic screening or local magnetic
`environment about the nucleus. Rapid molecular
`tumbling in solution state NMR spectroscopy
`creates a single average electronic or chemical
`shift environment for the nucleus. In contrast, the
`isotropic chemical shift in solid-state NMR spec-
`troscopy is sensitive to the asymmetry of the 3D
`electronic screening.7 Risedronate monohydrate
`and anhydrate forms were used as model com-
`pounds. The monohydrate contains 1 mole of
`lattice water and the anhydrate has no water of
`crystallization. Both risedronate hemi-pentahy-
`drate and dehydrated risedronate hemi-pentahy-
`drate display a distinct 31P, 13C, and 23Na
`spectrum providing a unique identity, as shown
`in Figures 5, 6, and 7, respectively. Spectra of
`dehydrated material are also unique from rise-
`dronate monohydrate and anhydrate forms. Both
`31P and 13C spectra are consistent with the
`dehydrated material containing two nonequiva-
`lent risedronate molecular environments within
`the unit cell. The 23Na spectra show that the
`sodium environment also has been altered.
`
`31P NMR Spectra
`The solid-state 31P NMR spectra of dehydrated
`risedronate are compared with spectra of risedro-
`nate hemi-pentahydrate, rehydrated desiccated
`material, monohydrate and anhydrate forms in
`Figure 5. Only the isotropic peaks in the central
`spectral region are shown. The substantial differ-
`ences seen between spectra of dehydrated mate-
`rial and hemi-pentahydrate arise from changes in
`the local electronic environments of the phos-
`phono groups created by removal of channel
`water. Changes in the hydrogen bonding network
`among the phosphono groups and neighboring
`water affect the local electronic environments of
`the 31P nuclei, resulting in isotropic chemical shift
`values that are sensitive to the type of water
`present in the risedronate hydrate form.4 An
`initial 608C in situ dehydration study (not shown)
`was undertaken to determine how well dehy-
`drated risedronate could be distinguished from
`hemi-pentahydrate. The temperature profile was
`optimized to remove 1 mole of channel water
`without removing the remaining 1.5 moles.
`The hemi-pentahydrate peaks did not undergo
`changes in chemical shift, but decreased in
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`Merck Exhibit 2240, Page 7
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2638
`
`LESTER ET AL.
`
`Figure 5. Solid-state 121.4 MHz 31P CP/MAS NMR spectra (top to bottom): hemi-
`pentahydrate (HPH); dehydrated by heat (DH-h); dehydrated by desiccation (DH-d);
`rehydrated desiccated material (RH); monohydrate (MH); and risedronate anhydrate
`(AH). The spectral region shown excludes spinning sidebands.
`
`intensity and disappeared over time. Concurrent
`with their disappearance, was the appearance
`and growth of new peaks. The final spectra
`showed only the new peaks from the dehydrated
`
`material. Spectra of hemi-pentahydrate dehy-
`drated in a 608C oven match those from the
`in situ study. Dehydrating risedronate hemi-
`pentahydrate either by heating or desiccation
`
`Figure 6. Solid-state 75.4 MHz 13C CP/MAS NMR spectra (top to bottom): hemi-
`pentahydrate (HPH); dehydrated by heat (DH-h); dehydrated by desiccation (DH-d);
`rehydrated desiccated material (RH); monohydrate (MH); and risedronate anhydrate
`(AH). Sideband suppression was used to aid in identifying isotropic peaks.
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`DOI 10.1002/jps
`
`Merck Exhibit 2240, Page 8
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`DEHYDRATION OF RISEDRONATE HEMI-PENTAHYDRATE: ANALYTICAL AND PHYSICAL CHARACTERIZATION
`
`2639
`
`Figure 7. Solid-state 79.4 MHz 23Na CP/MAS NMR spectra of (left to right): hemi-
`pentahydrate (HPH); dehydrated by heat (DH-h); dehydrated by desiccation (DH-d);
`rehydrated desiccated material; monohydrate (MH); and risedronate anhydrate (AH).
`
`creates materials with equivalent spectra. Rehy-
`drating desiccated risedronate produces material
`with spectra matching spectra of the original
`hemi-pentahydrate form.
`Risedronate hemi-pentahydrate exhibits one
`pair of isotropic peaks (19.2, 17.2 ppm), consistent
`with all risedronate molecules occupying a single
`electronically or magnetically equivalent environ-
`ment within the unit cell. In contrast, dehydrated
`risedronate shows two pair of isotropic peaks (20.9,
`19.6, 18.2, 11.5 ppm), consistent with nonequiva-
`lent risedronate molecules existing within two
`electronic environments within the unit cell.
`Risedronate can adopt a unit cell containing
`two nonequivalent electronic environments, as
`observed for the variable hydrate form.4 Compar-
`ing only isotropic chemical shifts shows one pair
`of peaks from dehydrated risedronate (19.6,
`18.2 ppm) consistent with monohydrate peaks
`(19.6, 18.2 ppm). Although not as closely aligned,
`the second pair of dehydrated risedronate peaks
`(20.9, 11.5 ppm) appears shifted toward the
`electronic environment of the anhydrate form
`(24.1, 10.7 ppm). Herzfeld–Berger analyses of
`hemi-pentahydrate, rehydrated desiccated mate-
`rial, dehydrated material, monohydrate and anhy-
`drate forms reveal that the dehydrate has unique
`chemical shift tensor components, distinguishing
`this form from previously identified forms. The
`phosphorus spectra reflect the changes in the local
`
`phosphono environments due to the loss of channel
`water and show the dehydrated crystal lattice
`has adopted a unit cell with two nonequivalent
`electronic environments.
`
`13C NMR Spectra
`The solid-state13C NMR spectra for dehydrated
`risedronate are compared with spectra of risedro-
`nate hemi-pentahydrate, rehydrated desiccated
`material, monohydrate and anhydrate forms in
`Figure 6. Only the isotropic peaks are shown;
`spinning sidebands were suppressed. Nonproto-
`nated carbons (C1 and C30) were identified by
`proton-dephasing experiments. Other
`carbon
`identities are based on solution state spectra of
`the free acid form. Like the 31P results, the 13C
`spectra are equivalent for dehydrated material
`created either by heating or by desiccation.
`Spectra of rehydrated desiccated material match
`spectra of the hemi-pentahydrate form.
`The hemi-pentahydrate exhibits one isotropic
`peak for each type of carbon nucleus in the
`pyridine and ethylidene moieties, consistent with
`electronically equivalent risedronate molecules in
`the unit cell. Dehydrated risedronate shows two
`isotropic peaks for the C2 methylene, the non-
`protonated C30 and at least two other pyridyl
`carbon nuclei, confirming that two nonequivalent
`electronic environments exist within the unit cell.
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`Merck Exhibit 2240, Page 9
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2640
`
`LESTER ET AL.
`
`Unlike 31P isotropic chemical shifts, none of the
`carbon peaks of dehydrated material is a match
`with monohydrate; both environments within the
`dehydrated material unit cell are clearly unique
`from other hydrate forms.
`
`23Na NMR Spectra
`The solid-state 23Na NMR spectra of dehydrated
`risedronate are compared with risedronate hemi-
`pentahydrate, rehydrated desiccated material,
`monohydrate and anhydrate forms in Figure 7.
`The entire spectral width is shown. Sodium nuclei
`are dominated by the electrical quadrupolar
`effect, whose magnitude reflects the local electric
`field symmetry about the nucleus. The greater the
`3D symmetry or the more spherical the electronic
`field is at the nucleus, the narrower the sodium
`peak.7 As the field becomes less spherical, the
`sodium lineshape broadens and distorts as seen
`for monohydrate versus anhydrate.
`As seen previously in both 31P and 13C spectra,
`the sodium results also show that heating or
`desiccating hemi-pentahydrate creates dehy-
`
`drated materials with equivalent spectra. Rehy-
`drated desiccated material matches the initial
`hemi-pentahydrate. Spectra from dehydrated
`material are unique from the other risedronate
`forms, agreeing with the Herzfeld–Berger analy-
`sis of the 31P spectra, and with the peak compar-
`ison among the 13C spectra.
`
`X-Ray Diffraction
`
`X-ray powder diffraction clearly discriminates
`between the fully hydrated and dehydrated forms
`of risedronate hemi-pentahydrate as shown in
`Figure 8. Upon the loss of 1 mole of water from the
`crystal system, the lattice undergoes a transition.
`This transition results in the formation of a new
`lattice configuration. By slowly heating a fully
`hydrated sample on a hot stage X-ray diffraction
`stage (Fig. 9)
`it was demonstrated that the
`transition does not proceed via a continuum of
`intermediate states. As the water leaves the
`channels, the lattice does not shift gradually.
`Instead the removal of
`the water causes a
`spontaneous lattice readjustment corresponding
`
`Figure 8. Powder Diffraction patterns for a single sample of risedronate hemi-
`pentahydrate showing: initial sample of hemi-pentahydrate (HPH); conversion by heat to
`dehydrated hemi-pentahydrate (DH-h); rehydration at room relative humidity back
`to the hemi-pentahydrate form (RH); dehydrated with desiccation (dry N2 gas) to
`dehydrated hemi-pentahydrate form (DH-d).
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 12, DECEMBER 2006
`
`DOI 10.1002/jps
`
`Merck Exhibit 2240, Page 10
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`DEHYDRATION OF RISEDRONATE HEMI-PENTAHYDRATE: ANALYTICAL AND PHYSICAL CHARACTERIZATION
`
`2641
`
`Figure 9. Risedronate hemi-pentahydrate (top) undergoing the complete transition to
`the dehydrated form (bottom) while being slowly heated on the hot stage XRD system.
`
`to the X-ray signature of the dehydrated hemi-
`pentahydrate. As a result, when a bulk sample is
`dried, the characteristic peaks for the dehydrated
`form grow as the corresponding peaks for the fully
`hydrated form decrease and disappear. These
`results indicate that the population of dehydrated
`form increases relative to the fully hydrated hemi-
`pentahydrate.
`Consequently, a water content corresponding
`to the theoretical value for a dihydrate would not
`indicate the presence of a dihydrate form but
`rather a mix of dehydrated and fully hydrated
`forms. Changes in unit cell dimensions associated
`with dehydration may be responsible for the
`crystal fracturing reported in the thermal analysis
`section and subsequently discussed in the micro-
`scopy and particle size sections of this report. The
`dehydrated crystal lattice has been confirmed to
`be a unique phase through indexing of the cell
`parameters. After 1 mole of water has been
`removed from the hemi-pentahydrate crystal
`lattice, the resulting lattice has changed from a
`P21nC, monoclinic cell with a volume of 2677 A˚ to a
`P-1, triclinic cell with a volume of 1223 A˚ . These
`results characterizing the dehydrated cell sym-
`metry have been confirmed by 31P and 13C SSNMR
`spectroscopy. Although dehydration of the hemi-
`pentahydrate produces a unique crystal form, the
`
`dehydrated form is metastable. Once dehydrated,
`the material will rapidly rehydrate under ambient
`conditions (>15% RH at 258C) returning to its
`original hemi-pentahydrate lattice configuration
`(Fig. 9). The only effect of the dehydration/
`rehydration process is a reduction in crystal size,
`which is shown by differences in the peak widths
`and preferred orientation observed in the powder
`pattern for the rehydrated material and confirmed
`by particle size distribution data.
`
`Particle Size/Light Microscopy
`
`Comparison of