Pharmaceutical Research, Vol. 12, N0. 7, 1995
`
`Review
`
`Pharmaceutical Solids: A Strategic Approach to
`Regulatory Considerations
`
`Stephen Byrn,1’4 Ralph Pfeiffer,1 Michael Ganey,2’3 Charles Hoiberg,2 and
`Guirag Poochikian2
`
`
`
`Purpose. This review describes a conceptual approach to the characterization of pharmaceutical solids.
`Methods. Four flow charts are presented: (1) polymorphs, (2) hydrates, (3) desolvated solvates, and (4)
`amorphous forms. Results. These flow charts (decision trees) are suggested as tools to develop infor-
`mation on pharmaceutical solids for both scientific and regulatory purposes. Conclusions. It is hoped
`that this review will lead to a more direct approach to the characterization of pharmaceutical solids and
`ultimately to faster approval of regulatory documents containing information on pharmaceutical solids.
`KEY WORDS: polymorph; hydrate; amorphous form; desolvated solvate.
`
`Interest in the subject of pharmaceutical solids stems in
`part from the Food and Drug Administration’s (FDA’s) drug
`substance guideline that states “appropriate” analytical pro-
`cedures should be used to detect polymorphic, hydrated, or
`amorphous forms of the drug substance. These guidelines
`suggest the importance of controlling the crystal form of the
`drug substance. The guideline also states that it is the appli—
`cant’s responsibility to control the crystal form of the drug
`substance and, if bioavailability is affected, to demonstrate
`the suitability of the control methods.
`Thus, while it is clear that the New Drug Application
`(NDA) should contain information on solid state properties,
`particularly when bioavailability is an issue, the applicant
`may be unsure about how to scientifically approach the gath-
`ering of information and perhaps what kind of information is
`needed. This review is intended to provide a strategic ap-
`proach to remove much of this uncertainty by presenting
`concepts and ideas in the form of flow charts rather than a
`set of guidelines or regulations. This is especially important
`because each individual compound has its own peculiarities
`which require flexibility in approach. The studies proposed
`herein are part of the Investigational New Drug (IND) pro-
`cess.
`
`Solid drug substances display a wide and largely unpre—
`dictable variety of solid state properties. Nevertheless, ap-
`plication of basic physicochemical principles combined with
`appropriate analytical methodology can provide a strategy
`
`
`
`1 Department of Medicinal Chemistry and Pharmacognosy, Purdue
`University, West Lafayette, Indiana 47907.
`2 Division of Oncology and Pulmonary, Food and Drug Administra-
`tion, 5600 Fishers Lane, Rockville, Maryland 20857.
`3 Current Address: Pfizer Central Research, Groton, Connecticut.
`4 To whom correspondence should be addressed.
`
`for scientific and regulatory decisions related to solid state
`behavior in the majority of cases. By addressing fundamen-
`tal questions about solid state behavior at an early stage of
`drug development, both the applicant and the FDA are in a
`better position to assess the possible effects of any variations
`in the solid state properties of the drug substance. The re-
`sulting early interaction of the parties with regard to this area
`would not only tend to ensure uniformity of the materials
`used throughout the clinical trials but also fully resolve solid
`state issues before the critical stages of drug development. A
`further benefit of these scientific studies is the development
`of a meaningful set of solid state specifications which criti-
`cally describe the solid form of the drug substance. These
`specifications would thus also facilitate the approval of a
`change in supplier or chemical process.
`Our approach in this review is to suggest a sequence for
`collecting data on a drug substance that will efficiently an-
`swer specific questions about solid state behavior in a logical
`order. In “difficult” cases, perhaps where mixtures of forms
`must be dealt with, or other unusual properties are encoun-
`tered, the suggested sequences would still have to be fol-
`lowed as a first stage in this investigation.
`We have chosen to present this approach in the form of
`a series of decision trees, or flow charts (algorithms), one for
`each of the most common solid state forms. The charts are
`
`accompanied by examples from the literature representing
`the kind of data that would be useful in supporting the var-
`ious decisions.
`
`Decision trees provide conceptual frameworks for un—
`derstanding how the justification for different crystal forms
`might be presented in the drug application. Industry may
`wish to use these decision trees as a strategic tool to organize
`the gathering of information early in the drug development
`process. Put another way, these decision trees provide a
`thought process that will lead to development of the most
`
`IPR2018-00126
`
`945
`
`Page 1 of 10
`
`0724»8741/95/0700»0945$07.50/0 © 1995 Plenum Publishing Corporation
`I-MAK 1013
`
`IPR2018-00126
`
`Page 1 of 10
`
`I-MAK 1013
`
`

`

`946
`
`appropriate analytical controls. One should also note that it
`is the responsibility of the industry to select the appropriate
`test or tests to identify the phase of the solid and determine
`its relevant pharmaceutical properties. This approach is su—
`perior to simply performing a broad range of tests without
`regard to their relevance.
`We should point out that, from a regulatory standpoint,
`if a company can establish a specification/test to ensure pro-
`duction of a well defined solid form of the drug substance,
`then it is not necessary to do all of the physical/chemical
`testing outlined in the decision trees. From a scientific stand—
`point, however, such an approach is risky since new forms
`may appear unpredictably during various stages of the de-
`velopment process. The appearance of these new forms usu-
`ally slows the drug approval process and makes planning
`difficult.
`Four decision trees are described in the sections that
`
`follow: Polymorphs; Hydrates (Solvates); Desolvated
`Solvates; and Amorphous Forms. Polymorphs exist when
`the drug substance crystallizes in different crystal packing
`arrangements all of which have the same elemental compo-
`sition (Note that hydrates can exist in polymorphs). Hy-
`drates exist when the drug substance incorporates water
`in the crystal lattice in either stoichiometric or non—
`stoichiometric amounts. Desolvated solvates are produced
`when a solvate is desolvated (either knowingly or unknow—
`ingly) and the crystal retains the structure of the solvate.
`Amorphous forms exist when a solid with no long range
`order and thus no crystallinity is produced. It is apparent
`that the appropriate flow chart can only be determined after
`the solid has been characterized using some of the tests de—
`scribed in the first decision point of the decision trees/flow
`charts (i.e. X-ray powder diffraction, elemental analysis,
`etc.). If there is no interest in marketing or producing an
`amorphous form or desolvated solvate at any stage in the
`process, then the corresponding flow charts do not need to
`be addressed. As already mentioned, it is advisable to inves—
`tigate the drug substance for the existence of polymorphs
`and hydrates since these may be encountered at any stage of
`the drug manufacturing process or upon storage of the drug
`substance or dosage form.
`
`Byrn, Pfeiffer, Ganey, Hoiberg, and Poochikian
`
`All of the flow charts end (see for example Figure 1)
`with an indication of the types of controls which will be
`required based on whether a single morphic form or a mix—
`ture will be produced as the drug substance. Although this
`ending provides a simplistic View of a very complicated pro-
`cess of selecting appropriate controls, it is included to illus-
`trate the consequence of the decisions made with regard to
`the drug substance. The reader should realize that the actual
`selection of the appropriate control could be the subject of
`another review which might contain another set of flow
`charts or decision trees.
`
`POLYMORPHS
`
`The flow chart/decision tree for polymorphs is shown in
`Figure 1. It outlines investigations of the formation of poly-
`morphs, the analytical tests available for identifying poly-
`morphs, studies of the physical properties of polymorphs
`and the controls needed to ensure the integrity of drug sub-
`stance containing either a single morphic form or a mixture.
`
`A. Formation of Polymorphs—Have Polymorphs
`Been Discovered?
`
`The first step in the polymorphs decision tree is to crys—
`tallize the substance from a number of different solvents in
`
`order to attempt to answer the question: Are polymorphs
`possible? Solvents should include those used in the final
`crystallization steps and those used during formulation and
`processing and may also include water, methanol, ethanol,
`propanol, isopropanol, acetone, acetonitrile, ethyl acetate,
`hexane and mixtures if appropriate. New crystal forms can
`often be obtained by cooling hot saturated solutions or partly
`evaporating clear saturated solutions. The solids produced
`are analyzed using X-ray diffraction and at least one of the
`other methods. In these analyses, care must be taken to
`show that the method of sample preparation (i.e. drying,
`grinding) has not affected the solid form. If the analyses
`show that the solids obtained are identical (e.g. have the
`same X—ray diffraction patterns and IR spectra) then the an-
`swer to the question “Are polymorphs possible?” is “No”,
`
`POLYMORPHS
`
`Drug Substance
`
`Polymorphs
`Discovered?
`
`Different Recrystallizing
`Solvents (different
`polarity) - vary
`temperature, concentration,
`ag‘m‘mm PH
`
`Tests for Polymorphs
`— X—Ray Powder Diffraction
`- DSC (Thennoanalyfical
`Methods)
`. Microscopy
`_
`IR
`, Solid State NMR
`
`No
`
`Yes
`
`m No
`
`Yes
`
`Drug
`Substance
`Composition?
`
`Single Morphic Form
`Qualitative Control
`(e.g., DSC or XRD)
`Mixture of Forms
`
`Quantative Control
`(e.g., XRD)
`
`Monitor Polymorph
`in Stability Studies
`
`_
`Different
`Physical
`Properties?
`Different
`Stability (Chemical
`& Physical)
`- Solubility Profile
`- Morphology of Xtals
`- Calorimetric Behav.
`’ % R” Proms
`
`Figure 1. Flow chart/decision tree for polymorphs.
`
`IPR2018-00126
`
`Page 2 of 10
`
`I-MAK 1013
`
`IPR2018-00126
`
`Page 2 of 10
`
`I-MAK 1013
`
`

`

`Pharmaceutical Solids: A Strategic Approach to Regulatory Considerations
`
`947
`
`
`
`if
`
`31‘
`
`ff
`
`.,
`
`; g K’Kgfi‘fi;
`
`33 mg
`
`aa
`
`£35
`.9:
`
`Figure 3. Stereoscopic drawings of the crystal packing of both poly-
`morphs of 8-(2-methoxycarbony1amino-6-methylbenzyloxy)-2-
`methyl-3-(2-propynyl)-imidazo{1,2-a}pyridine viewed along the
`shortest axis (Form A, b-axis; Form B, a-axis) (1).
`
`B. Do the Polymorphs Have Different Physical Properties?
`
`is necessary to examine
`If polymorphs exist then it
`the physical properties of the different polymorphs that
`can affect dosage form performance (bioavailability and sta-
`bility) or manufacturing reproducibility. The properties of
`interest are solubility profile (intrinsic dissolution rate, equi-
`librium solubility), stability (chemical and physical), and
`crystal morphology (including both shape and particle size),
`calorimetric behavior, and %RH profile. If there are no dis-
`cernible differences between these physico-chemical prop-
`erties, then the answer to the second question in the decision
`tree, “Different physical properties?” is “No.”
`The variable physical properties of several drugs with
`different polymorphs are reported in the literature. For ex-
`ample, the dissolution profiles of the polymorphs of chlor-
`amphenicol are significantly different (4). In addition, van’t
`Hoff solubility analysis has been used to elucidate the dif-
`
`Form A
`
`Form B
`
`u
`E%—I}.C
`‘5
`‘UCL2
`
`and further research is not needed. The work of Miyamae et
`al. serves as a good example of solid state studies of a drug
`substance which exists as polymorphs (1). Powder diffrac-
`tion showed that there were two crystal forms (see Figure 2).
`These workers also carried out single crystal analysis of
`the two crystal forms of the compound. The structures are
`shown in Figure 3. While such studies are not required, and
`indeed sometimes not possible, they provide an unequivocal
`confirmation of the existence of polymorphs. Moreover,
`’ once the single crystal structure of a phase has been deter-
`mined, it is possible to calculate the corresponding X—ray
`powder pattern. This provides an irrefutable standard for
`identifying the phase by that method.
`The DSC thermal curves of the two forms are slightly
`different, as shown in Figure 4 and thus may not be the
`preferred way of differentiating these polymorphs.
`The IR spectra of the two polymorphs are quite simi-
`lar(1), and IR does not appear to be a powerful method for
`differentiating the crystal forms in this case. Thus, for 8-(2-
`methoxycarbonylamino-6-methylbenzyloxy)—2-methyl-3-(2-
`propynyl)-imidazo{1,2-a}pyridine, powder diffraction ap-
`pears to be the best method for differentiating the two forms.
`Solid-state NMR is another powerful technique for an-
`alyzing different crystal forms (2,3). Figure 5 shows the
`solid-state C-13 NMR spectra of Forms I and II of prednis-
`olone. Differences in the positions of the two resonances in
`the 120 ppm range clearly differentiate the two forms. In
`principle, solid state NMR is an absolute technique in which
`the signal intensity is proportional to the number of nuclei
`provided appropriate conditions are met. In addition, solid
`state NMR is a bulk technique which is not very sensitive to
`surface changes. This method appears to be very sensitive
`and will undoubtedly be used more often in the future as a
`tool to detect different crystal forms. However, with present
`technology, errors in solid-state quantitative studies may be
`rather large.
`
`1 J
`
`im);
`
`Ditty-Clio!» not.
`
`20.
`
`down
`
`Figure 2. Powder X—ray diffraction patterns of the polymorphs of
`8-(2-methoxycarbonylamin0-6—methylbenzyloxy)-2-methyl-3-(2-
`propynyl)-imidazo{l ,2-a}pyridine (l).
`
`Figure 4. DSC thermal curves of the polymorphs of 8-(2-
`methoxycarbonylamino-o-methylbenzyloxy)-2-methyl-3-(2-
`propynyl)-imidazo{l,2-a}pyridine (1). These curves show that Form
`A melts whereas Form B undergoes a small endothermic transition
`and then melts at the same temperature as Form A.
`
`120
`
`1/40
`
`160
`
`180
`
`200
`
`Tanperature,
`
`°C
`
`IPR2018-00126
`
`Page 3 of 10
`
`I-MAK 1013
`
`IPR2018-00126
`
`Page 3 of 10
`
`I-MAK 1013
`
`

`

`948
`
`200
`
`‘50
`
`100
`
`$0
`
`0
`
`PP'I
`
`WWW
`zoo
`ISO
`100
`so
`0 m
`
`Figure 5. Solid state NMR of the two Crystal Forms of Predniso-
`lone (2).
`
`ferent solubilities of two polymorphs of methyl predniso-
`lone(5). This method involves determining the equilibrium
`solubility of each polymorph at various temperatures. The
`log of the equilibrium solubility is then plotted vs l/T. This
`should give straight lines for each polymorph and the tem-
`perature at which the curves intersect is the transition tem-
`perature. This technique does not work if the polymorphs
`interconvert.
`
`For balance, it is important to point out that there are
`also cases where polymorphs exist but they have virtually
`identical dissolution properties(6).
`
`C. Drug Substance Control
`
`The important question lies in the properties that differ
`among polymorphs and whether those properties affect the
`dosage form performance (i.e., quality or bioavailability). If
`they do then from a regulatory standpoint it is appropriate to
`establish a specification/test (e.g. powder X-ray diffraction
`or IR) to ensure the proper form is produced. From a pro—
`duction standpoint, it is important to develop a process that
`reproducibly produces the desired polymorph.
`If mixtures of forms cannot be avoided, then quantita-
`tive control is needed to ensure that a fixed proportion of
`forms is obtained. Furthermore, the method of analyzing for
`the proportion of forms would have to be validated. Also, the
`proportion of forms would have to remain within stated lim-
`its through the retest date of the drug substance and poten-
`tially throughout the shelf life of the product; a difficult re-
`quirement if the forms interconvert. Thus, the way to avoid
`a substantial amount of work in this area is to select a single
`
`Bym, Pfeiffer, Ganey, Hoiberg, and Poochikian
`
`solid form for production. Usually, this would be the most
`physically stable form when their bioavailabilities are not
`significantly different. Selection of the most stable from
`would, of course, insure that it there would be no conversion
`into other forms.
`Powder diffraction is often a useful method to determine
`
`the percentages of polymorphs in a mixture; however, the de-
`tection limit is variable from case to case and can be as high as
`15%. Matsuda (7) carried out a mixture analysis of phenylbu—
`tazone polymorphs. Diffraction lines disappear and appear as
`the ratio of the crystal forms change. Some of these calibration
`curves developed from this analysis are almost horizontal,
`meaning that any given mixture gives the same line intensity in
`this mixture range. However, other calibration curves are
`sloped and would appear to allow a reasonable analysis. It is
`fair (although Matsuda did not carry out an estimate) to esti—
`mate the errors in this analysis as :15%.
`Tanninen and Ylirussi (8) used computer curve fitting to
`carry out a mixture analysis of prazosin. In this particular
`case, they reported a highly accurate analysis, and, in fact,
`showed a calibration curve that could detect 0.5% of one
`
`form in another. This is obviously a highly accurate mixture
`analysis by powder diffraction and shows the power of this
`method for some applications. However, this analysis re-
`quired extreme care in sample preparation and may be more
`difficult to carry out in a production setting where particle
`size may not be controlled. Similar comments apply to the
`analysis of mixtures by IR, where the accuracy and precision
`may also vary considerably from case to case. Given the
`analytical problems in dealing with mixtures of forms, it may
`generally be simpler to develop a method to prepare only one
`crystal form.
`In summary, it is important to determine whether poly-
`morphs are present and to solve any problems before pivotal
`clinical studies are initiated.
`
`D. Determination of the Polymorph Present in the
`Drug Product
`
`In cases where stability or bioavailability issues exist,
`the solid form present in the drug product should be inves-
`tigated, if possible.
`For bulk drug substances, X-ray powder diffraction and
`other techniques can identify the polymorph; however, solid
`state NMR appears to be the best method for the study of the
`drug substance in the dosage form (2, 3). Solid-state NMR
`study of three commercial products containing prednisolone
`showed that the products A and B contain Form I, whereas
`product C contains Form 11.. This analysis was possible even
`though these tablets contain approximately 95 mg of excip-
`ients and 5 mg of drug. There are numerous cases, often
`involving complex mixtures or low dose products, where
`solid state NMR (and,
`in fact, any technique) will not be
`sensitive enough to identify the polymorph present in the
`drug product. However, the safety and efficacy is, of course,
`controlled by the potency assays and by the physical tests
`(e.g., dissolution).
`
`HYDRATES (SOLVATES)
`
`The flow chart/decision tree for hydrates (solvates) is
`shown in Figure 6. It outlines investigations of the formation
`
`IPR2018-00126
`
`Page 4 of 10
`
`I-MAK 1013
`
`IPR2018-00126
`
`Page 4 of 10
`
`I-MAK 1013
`
`

`

`Pharmaceutical Solids: A Strategic Approach to Regulatory Considerations
`
`949
`
`HYDRATES (SOLVATES)
`Drug Substance and Solvent
`
`No
`
`Yes
`
`—— No
`
`Yes
`
`Drug
`Substance
`9
`.
`.
`COmPOSItIOIl.
`
`Single Hydrate
`ualitative Control
`Q
`(e.g., DSC OI' XRD)
`
`Mixture of Forms
`Quantative Control
`(e. g, XRD)
`.
`Mommr Hydrate
`-
`.
`-
`~
`in Stability Studies
`
`.
`Different
`Physical
`Properties?
`D.“
`‘
`l eren
`_ Stability (Chemical
`.
`& Phys’c‘“)
`— Solubility Profile
`‘ Morphology 0f Xtals
`- Calonmetrlc Behav.
`- % RH profile
`
`Hydrates
`Discovered?
`
`.
`~
`4
`Different Recrystalllzmg
`Solvents (different
`polarity) - vary
`temperature concentration.
`ag‘m‘m’ 9“! wa‘er wmem
`
`.
`,
`Tests for Hydrates
`» X-Ray Powder Diffraction
`DSC/TGA/H s
`‘ M,
`0‘
`‘age
`- Eleme‘ggicAogzlysis
`_ % RH profile
`_
`IR
`- Solid State NMR
`- Solution NMR (Solvent
`Content and Amount)
`
`Figure 6. Flow chart for solvates or hydrates.
`
`of solvates. For example, one early study showed that TGA
`could differentiate three different hydrated salts of feno-
`profen(10). Combined with IR or other methods, TGA is an
`unequivocal method for the verification of the existence of
`solvates. In addition, TGA is a good method for looking at
`mixtures of solvated and unsolvated crystal forms, and prob— ’
`ably can be developed into an analytical method for deter—
`mining the ratios of solvated and unsolvated forms.
`DSC is also a good method for detecting solvates since
`there is usually heat change involved in desolvation, espe-
`cially for hydrates(ll)l Specifically, DSC by itself does not
`prove the existence of a solvate, but once other analytical
`
`of hydrates (solvates), the analytical tests available for hy-
`drates (solvates), studies of the physical properties of hy—
`drates (solvates) and the controls needed to ensure the in—
`tegrity of drug substance containing either a single morphic
`form or a mixture.
`
`A. Have Hydrates (Solvates) Been Discovered?
`
`The flow chart for hydrates (solvates) (Figure 7) is ap-
`plied after the preliminary crystallizations have been com-
`pleted. These are essentially the same as in the polymorph
`decision tree but, in addition, should include solvent-water
`mixtures in order to maximize the chance for hydrate for-
`mation. These experiments can be guided by the moisture
`uptake (% RH) studies. Any solids that indicate a significant
`change in water content as indicated by the % RH-moisture
`profile should also be examined. The resulting solid phases
`are preferably characterized by a combination of methods—
`two for phase identity and two to reveal composition and
`stoichiometry.
`With a very few exceptions, the structural solvent con-
`tained in marketed crystalline drug products is water. It is
`nevertheless often desirable to characterize other solvated
`crystalline forms of a drug for several reasons: they may be
`the penultimate form used to crystallize the final product and
`thus require controlled characterization; they may form if
`the final crystallization from solvents, especially mixed sol-
`vents, is not well controlled; they may be the actual crystal-
`lized form of a final product that is desolvated during a final
`drying step; they may be the form used in recovery for sub-
`sequent rework. The relevance of these points will vary from
`case to case, but for the present discussion we shall treat the
`subject of solvates in its broadest form.
`Examples taken from the literature serve to illustrate
`the kind of data that proves useful in characterizing solvated
`crystal forms. For example, a recent report from our labo-
`ratory showed that IR and solid state NMR was useful for
`the identification of the different crystal forms of dirithro-
`mycin(9). TGA is another powerful method for the analysis
`
`PENTAHYDRATE FORM j
`
`
`9‘
`WATER '
`CONTENT
`
`VAPOR PRESSURE ISOTHERM FOR
`SODIUM CEFAZOLIN (T - 25°C)
`
`MOLES
`WATER
`
`SESQUIHYDRATE FORM 1
`
`MONOHYDRATE FORM-J
`
`
`
`'/¢ RELATIVE HUMIDITY
`
`Figure 7. Water uptake vs percent relative humidity for sodium
`cefazolin.
`
`IPR2018-00126
`
`Page 5 of 10
`
`I-MAK 1013
`
`IPR2018-00126
`
`Page 5 of 10
`
`I-MAK 1013
`
`

`

`950
`
`data from TGA, NMR, etc. are available, DSC becomes a
`good method for analyzing solvates and determining a per-
`centage of solvates present.
`The three solvates of ethynylestradiol (0.5 acetonitrile,
`1.0 methanol, 0.5 water) provides another interesting exam-
`ple (12). These solvates have different cell parameters and
`are crystallographically completely distinct materials. The
`hemihydrate was obtained from an organic solvent which is
`not completely miscible with water but was saturated with
`water. In fact, it is known that crystallization from water-
`immiscible solvents containing small but slightly different
`proportions of water can produce different hydrates of a
`substance.
`
`The DSC/TGA of the three ethynylestradiol solvates(12)
`are unique and in this case it may be possible to develop
`DSC/TGA into an analytical procedure for determining the
`proportions of each solvate. The DSC in some of these
`traces appears to show a melt and recrystallization corre-
`sponding to the loss of solvent of crystallization. However,
`the exact interpretation of this is not possible without either
`a DSC microscope or interrupting the tracing to analyze the
`sample at various temperatures. The methanolate appears to
`lose solvent in two equal steps, indicating that there may
`also be a hemimethanolate of this compound. Again, confir-
`mation of this would require interrupting the heating and
`analyzing the substance after the first solvent loss has oc-
`curred. In addition, the DSC/TGA traces suggest that all of
`the forms are converted to an anhydrous form which then
`melts at a higher temperature. Thus, interrupting any one of
`these thermal curves just prior to the final melt could reveal a
`new form that gives the powder pattern for the anhydrate. Un-
`fortunately, no data of this type is provided in the case cited.
`DSC analysis of solvates should be carried out using
`either an open pan or a pan with a pin-prick; otherwise,
`unusual and variable results will be obtained because the
`
`solvent is not provided a way of escape from the pan. One
`advantage of using an open pan for DSC is that it reproduces
`the conditions under which the TGA is performed.
`Comparison of the ethinylestradiol powder diffraction
`patterns clearly establishes that these solvates are different
`crystal forms as would be expected from the single crystal
`data(12). In summary, DSC, TGA, and powder diffraction
`are all good methods for analysis of the different crystal
`forms of ethinylestradiol.
`Figure 7 shows a percent relative humidity versus water
`uptake study of the type recommended by the USP commit-
`tee on water(13) In this case, there are two hydrates which
`are relatively well behaved insofar as they are completely
`hydrated at about 10% relative humidity and remain uni-
`formly hydrated throughout a wide humidity range. On the
`other hand, the so-called pentahydrate, which really is only
`a pentahydrate at very high humidity, changes water content
`considerably as the relative humidity is changed. The USP
`committee on moisture specifications recommended that
`moisture uptake vs relative humidity studies should be rou-
`tinely performed on all drug substances and excipients (13).
`
`B. Do the Hydrates (Solvates) Have Different
`Physical Properties?
`‘
`
`The physical properties of hydrates are often quite dif-
`ferent from the anhydrate form. Figure 8 shows the dissolu-
`
`IO
`
`O
`
`5
`
`20 25 30 35 4045 50
`15
`TIME X I0“.s
`Figure 8. The dissolution-time curves for anhydrous and hydrated
`theophylline in water at 25°. The two types of open circles represent
`successive experiments (18).
`
`IPR2018-00126
`
`Page 6 of 10
`
`I-MAK 1013
`
`Bym, Pfeiffer, Ganey, Hoiberg, and Poochikian
`
`tion profile of theophylline hydrate and anhydrate. This fig-
`ure shows that the anhydrate reaches a much higher solubil—
`ity in water, and on extended exposure recrystallizes to the
`less soluble hydrate. Such differences must be further exam-
`ined for possible effects on bioavailability.
`In our laboratory we have described the different crystal
`forms of hydrocortisone-Zl—tertiary butylacetate(l4). Our
`studies showed that the nonstoichiometric ethanolate is ox-
`
`ygen-sensitive and, of course, would show different physical
`properties from the stoichiometric ethanolate and the other
`solvates. Prednisolone tertiary-butylacetate also exists as a
`nonstoichiometric hydrate which is oxygen sensitive(15).
`Thus, these are cases where different crystal forms have
`different chemical stability, although there may be no signif—
`icant differences in solubility.
`
`C. Mixtures of Polymorphs and Hydrates
`
`Other drug substances exist as both polymorphs and
`solvates. For example, furosemide exists in two poly—
`morphs, two solvates, and an amorphous form (16, 17). The
`polymorphs are enantiotropically related, which means that
`at one temperature one polymorph is more stable, but at a
`different temperature the other polymorph is physically
`more stable. That is, plots of solubility versus temperature
`cross for the two polymorphs. In addition, the different crys-
`tal forms have different photostability (chemical stability in
`light) and moreover have different dissolution rates. Thus,
`there are significant differences in both chemical and phys-
`ical properties.
`The five different forms, or modifications of furosemide,
`give clearly different powder patterns. Thus, powder diffrac-
`tion is a good method for analysis of these different forms.
`There are similarities between the IR spectra of the five
`different forms but there are also some significant differ-
`ences, and expansion and careful analysis could lead to an
`FT/IR method for analysis of these different forms. IR would
`probably be a useful method for analysis at least for pairs of
`these compounds. However, it is not clear whether IR could
`be used to determine the percentages of several different
`forms in a more complex mixture. The DSC and TGA of the
`different forms are significantly different. As expected, the
`solvates show weight loss in the TGA.
`
`I4
`
`THEOPHYLLINE
`
`5U 2
`
`E z
`
`' 8
`8 .m 6
`
`4
`
`Z d
`
`|2
`
`E
`BIO
`
`O (D anhydrous form
`o h drate
`y
`
`IPR2018-00126
`
`Page 6 of 10
`
`I-MAK 1013
`
`

`

`Pharmaceutical Solids: A Strategic Approach to Regulatory Considerations
`
`951
`
`The interconversion of the different forms of furosemide
`have been analyzed and a diagram constructed. Such a dia-
`gram can be experimentally difficult when so many pairs of
`crystal forms must be studied for possible interconversions
`and under different conditions. It is clear from this diagram
`that many of the forms of furosemide can be converted to
`form I. This study is one of the most complete reports of
`solvates and polymorphs available in the literature and
`serves as a model for studies of such systems for regulatory
`submissions.
`
`D. Determination of the Hydrate Present in the
`Drug Product
`
`Another important area is the analysis of the material
`which is produced after wet granulation of a substance which
`can form hydrates. We are aware of cases where the bulk
`drug substance is manufactured and stored as the anhydrate.
`However, upon wet granulation, there is a conversion (either
`partial or complete) to a hydrate. Subsequent drying is some-
`times not adequate to convert the substance back to the
`anhydrate, and a hydrate or a mixture of hydrate and anhy-
`drate remain. The formation of a hydrate and its subsequent
`drying can result in a change in particle size of the drug
`substance (19). It may also be possible to cause transforma-
`tions during other processing steps. It is thus recommended
`that if wet granulation or processing that subjects the drug to
`even brief changes in temperature or pressure (e. g. milling or
`compression) is contemplated, then extensive studies of the
`ability to convert the drug substance to a new crystal form be
`carried out by mimicking the processing step in the labora—
`tory.
`
`DESOLVATED SOLVATES
`
`The term “desolvated solvates” refers to compounds
`that are crystallized as solvates but undergo desolvation
`prior to analysis. Often these “desolvated solvates” retain
`
`the structure of the solvate with relatively small changes in
`the lattice parameters and atomic coordinates, but no longer
`contain the solvent. In addition, desolvated solvates are apt
`to be less ordered that their crystalline counterparts. These
`forms are particularly difficult to characterize properly since
`analytical studies indicate that they are unsolvated materials
`(anhydrous crystal forms) when, in fact, they have the struc-
`ture of the solvated crystal form from which they were de-
`rived. Several observations may give clues that one is deal-
`ing with a desolvated form: (1) The form can be obtained
`from only one solvent; (2) On heating, the form converts to
`a structure known to be unsolvated; and (3) The form has a
`particularly low density compared to other forms of the same
`substance. Experiments that help to clarify whether an ap-
`parently solvent free modification is a desolvated form or a
`true anhydrate include:
`(1) Single crystal X—ray structure
`determination in the presence of mother liquor from the
`crystallization; (2) comparison of the X-ray powder diffrac-
`tion patterns and solid state NMR spectra of the solvated
`and desolvated crystal forms; and (3) determination of the
`vapor pressure isotherm by varying the vapor pressure of the
`specific solvent involved. A desolvated form will often take
`up stoichiometric amounts of the relevant solvent. In addi—
`tion, crystals of the form