POLYMORPHIC CRYSTAL FORMS
`AND COCRYSTALS IN DRUG
`DELIVERY (CRYSTAL ENGINEERING)
`
`2
`
`NLNG SHAK1
`MICHAEL J . ZAWDROTKO
`1 Thar Pharmaceuticals, Inc.,
`Tampa, FL
`2 Department of Chemistry, University of
`South Florida, Tampa FL
`
`Active pharmaceutical ingredients, APIs, are
`most conveniently developed and delivered
`orally as solid dosage forms that contain a
`defined crystalline form of an API. This means
`that the pharmacokinetic profile of a dosage
`form is at the very least linked to the physi-
`cochernical properties of the crystal form that
`is selected for development. Furthermore,
`that crystal forms of new chemical entities are
`novel,
`lack obviousness, and have utility
`makes them patentable. Therefore, selection
`of a specific crystal form for a given API is a
`profoundly important step in drug develop-
`ment from clinical, legal, and regulatory per-
`spectives. In this context, scientific develop-
`ments that afford greater understanding of
`and diversity in the number of crystalline
`forms available for a given API, which have
`traditionally been limited to salts, poly-
`morphs, and hydrates/solvates [1], are ob-
`viously of relevance to the pharmaceutical
`industry. The science of crystal engineering
`[2] focuses upon self-assembly of existing mo-
`lecules or ions and it has evolved in such a
`
`manner that a Wide range of new crystal forms
`can be generated without the need to invoke
`Covalent-bond breakage or formation. This
`contribution will address the impact of crystal
`engineering upon our fundamental under-
`standing of crystal form diversity and how
`physical properties of crystals can be custo-
`mized via the emerging class of crystal
`forms that have been termed pharmaceutical
`cocrystals [3].
`
`1. INTRODUCTION
`
`The importance of crystallization and crystal
`forms to pharmaceutical science is the result
`of multiple practical considerations. In terms
`
`‘I87
`
`of processing, crystallizations tend to afford
`highly pure products, they are typically repro-
`ducible and scalable, and they are generally
`stable when compared to amorphous solids or
`solutions. They are therefore preferred by
`developers and regulatory bodies. Further-
`more,
`although crystallization has been
`widely studied scientifically since at least the
`early nineteenth century, this does not mean
`that crystallization is predictable [4] or even
`controllable [5]. New crystal forms are there-
`fore likely to be patentable in their own right
`since they meet the primary criteria for pa-
`tentability: novelty, lack of obviousness and
`utility. Finally, it has been known for over 100
`years that rate of dissolution of a solid is at
`least partly determined by thermodynamic
`solubility of a compound [6] and it is well
`recognized that solubility can significantly
`influence the bioavailability and pharmacoki-
`netics of an API. Given that the majority of
`APIS Currently under development fall into
`Biopharmaceutical Classification Scheme [7]
`(BCS) classification ll (low solubility, high
`permeability), the importance of API crystal
`form screening and selection is, if anything,
`increasing in scope and importance. In short,
`the existence of multiple crystal forms of an
`API affords both challenges and opportunities
`to the pharmaceutical industry. In this con-
`text, the emergence of the concept of crystal
`engineering is timely and relevant.
`Crystal engineering [2] was coined by R.
`Pepinsky [2c] in 1955 and brought to practice
`by G.M.J. Schmidt in the context of topochem-
`ical reactions [2d]. Crystal engineering has
`more recently matured into a paradigm for
`the understanding of existing crystalline so-
`lids and the design of new compounds with
`customized composition and physical proper-
`ties.
`Indeed, crystal engineered materials
`have been studied in the context of host—guest
`compounds, nonlinear optical materials, or-
`ganic conductors, and coordination poly-
`mers [&11]. However, given that APIS are
`perhaps the most valuable crystalline sub-
`stances known and their very nature (i.e., the
`presence of hydrogen -bonding functionality at
`their periphery) makes them predisposed to-
`ward crystal engineering, it is perhaps unsur-
`prising that crystal engineering concepts are
`increasingly being applied to pharmaceutical
`
`Bttrgeris Medicirial Chemistry, Drug Discovery, and Development, Seventh Edition,
`edited by Donald J. Abraham and David P. Rotella
`Copyright (C) 2010 John Wiley & Sons, Inc.
`
`Case No. 2:10—cv-05954
`Janssen Products, LP. at al.
`v. Lupin Limited, at al.
`
`PTX10
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`lPR2015-01030
`
`(Page 1 of 32)
`
`

`
`188
`
`POLYMORPHIC CRYSTAL FORMS AND COCRYSTALS IN DRUG DELIVERY
`
`science by both industrial and academic re-
`searchers [12-—23]. It can be asserted that
`crystal
`engineering is
`finally realizing
`Desiraju”s vision that crystal engineering is
`“the understanding of intermolecular interac-
`tions in the context of crystal packing and
`utilization of such understanding in the de-
`sign of new solids with desired physical and
`chemical properties” [2e].
`The range of crystal forms that are typi-
`cally exhibited by APIs represents a micro-
`cosm of organic compounds although it would
`be fair to assert that APIs are more promis-
`cuous than “typical” organic compounds be-
`cause they contain multiple hydrogen-bond
`ing sites and/or torsional flexibility. It is hy-
`drogen bonding sites or, more specifically, the
`detailed understanding of the supramolecular
`chemistry of these hydrogen-bonding sites
`that is the key to understanding the struc-
`ture—property relationships in crystal forms.
`The existence of multiple crystal forms for an
`API is therefore to be expected and they are
`typically categorized as follows: polymorphs,
`
`solvates, hydrates,
`salts,
`(Fig. 1).
`
`and cocrystals
`
`the exis-
`— Polymorphs: Polymorphism,
`tence of more than one crystal form for
`a compound, has been described as “the
`nemesis of crystal design” by one of the
`pioneers of crystal engineering, GR. De-
`siraju. Indeed, there are probably many
`researchers in the pharmaceutical indus-
`try who would regard polymorphism as
`the nemesis of crystal form selection
`since the unpredictability of polymorph-
`ism complicates all aspects of crystalliza-
`tion from laboratory scale discovery
`through to industrial scale processing.
`
`— Salts: Salts have long been an integral
`part of crystal form selection because
`they offer diversity of composition and
`can therefore exhibit a Wide range of
`physi.cochemical properties. However,
`salts, especially chloride salts, tend to be
`prone to exist as hydrates, there are a
`
`Neutral API
`
`Waterlsolvent
`
`
`
`Charged API
`Counter ion
`
`Neutral cocrystal
`former
`
`Figure 1. Crystal forms typically exhibited by molecular organics.
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 2 of 32)
`
`

`
`limited number of pharmaceutically ac-
`ceptable counterions, and not all APIs are
`acidic or basic enough to form salts
`[24].
`
`Hydroztes and Solvates: Solvates are crys-
`talline compounds in which solute and
`solvent molecules coexist, normally but
`not always through interaction of nonco-
`valent bonds such as hydrogen bonds.
`Likewise, hydrates are compounds that
`contain water bound Within the crystal
`lattice. One might think that hydrates
`are typically prepared using water as a
`solvent but the ubiquitous presence of
`water means that they are most typically
`isolated through the presence of adven-
`titious water molecules. Indeed, they re-
`present more than 10% of the >500,D00
`crystalline organic compounds that have
`been archived in the Cambridge Struc-
`tural Database, CSD. However, just as
`polymorphs are unpredictable, so are sol-
`vates and hydrates. Furthermore, sol-
`vates and hydrates are less likely to be
`selected as dosage forms because they
`tend to be prone to desolvation or dehy-
`dration in dry conditions.
`
`Cocrystals: Cocrystals represent a class of
`compounds that could reasonably be de-
`scribed as long known but little stu-
`died [25]. Indeed, to our knowledge the
`term cocrystal was not coined until
`1967 [26] and it was not popularized in
`the context of small molecules until M.C.
`
`Etter used the term extensively in the
`19805 [2a]. Furthermore, even today the
`term cocrystal is poorly defined and re-
`presents ambiguity or
`even contro-
`versy [27]. We define a cocrystal as fol-
`lowing: a multiple component crystalline
`solid formed in a stoichiometric ratio be-
`
`tween two compounds that are crystal-
`line solids under ambient conditions. At
`
`least one of these compounds is molecular
`(the cocrystal former) and forms supra-
`molecular synthons(s) with the remain-
`ing component(s) [3a——3e]. If one uses this
`definition then the first cocrystals were
`reported in the 1800s [25] and they have
`had various terms applied to them: addi-
`tion compounds, organic molecular com-
`
`INTRODUCTION
`
`189
`
`pounds, complexes, and heteromolecular
`crystals [28—35]. Cocrystals are also dis-
`tinct from solvates, salts and inclusion
`compounds if one employs this definition.
`Nevertheless, the term pharmaceutical
`cocrystal, that is, a cocrystal between an
`API and a molecular cocrystal former,
`was not widely used until recent years.
`Pharmaceutical cocrystals were reported
`as far back in the 1930s [36], yet only in
`recent years has their diversity in terms
`of crystal form and physical properties
`been fully recognized in the context APIs.
`
`Salt screening and selection is covered in a
`different chapter and solvates and hydrates
`tend to exhibit
`lower stability than poly-
`morphs or pharmaceutical cocrystals. This
`chapter will therefore focus upon polymorphs
`and cocrystals with emphasis upon how they
`can be subjected to rationalization through
`crystal engineering. The key to crystal engi-
`neering in the context of APIS lies with under-
`standing the hydrogen-bonding groups pre-
`sent in the API. Two approaches have been
`developed to analyze existing crystal struc-
`tures with the View to utilize the structural
`
`knowledge thereby gained to rationalize and
`even control the composition or even structure
`of new crystal forms. These related and coin-
`patible approaches, graph sets and supramo-
`lecular synthons, were developed by Etter [2a]
`and Desi.raju [2i], respectively. In both in-
`stances, there is reliance upon utilizing the
`Cambridge Structural Database [37], to gath-
`er statistical information about crystal pack-
`ing and intermolecular interactions. We shall
`focus herein upon supramolecular synthons,
`which are defined as “a structural unit within
`
`the supermolecule that can be formed and/or
`assembled by known or conceivable intermo-
`lecular interactions.” Supramolecular syn-
`thons focus upon functional groups rather
`than molecules and exist in two distinct cate-
`
`gories: supramolecular homosynthons that are
`composed of identical complementary func-
`tional groups, for example, carboxylic acid di-
`mers [38], amide dimers [39] (Fig. 2a and b);
`supramolecular heterosynthons composed of
`differentbut complementary functional groups
`such as acid-amide [40] and acid-aromatic
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 3 of 32)
`
`

`
`190
`
`POLYMORPHIC CRYSTAL FORMS AND COCRYSTALS IN DRUG DELIVERY
`
`(6)"l-<:>- ‘flit
`1% _£ 0
`
`(cl
`
`Figure 2. Prototypal supramolecular homosynthons (a) and (b) and supramolecular heterosynthons (c) and
`(d).
`
`nitrogen [41] (Fig. 2c and d). The aforemen-
`tioned supramolecular synthons are particu-
`larly salient because carboxylic acids are pre-
`sentin25ofthe top 100 mostprescrihed drugs in
`the United States. Furthermore, they are fre-
`quently encountered in pharmaceutical excipi-
`ents, salt formers and cocrystal formers.
`
`2. CRYSTAL FORM TYPES
`
`2.1. Polymorphs
`
`The first observation of polymorphism can be
`attributed to Wohler and Von Liebig, who in
`1832 reported that upon cooling a boiling solu-
`tion of benzamide, needle—shaped crystals
`would initially form [42]. However, upon
`standing the needle-shaped crystals would
`slowly be replaced by rhombic crystals. This
`observation is a manifestation of Ostwald’s
`
`step rule, that is, that the crystal form first
`obtained upon crystallization of a substance
`from a solution or a melt will be a metastable
`
`polymorph, a long recognized [43] and quali-
`tative generalization about crystallization.
`However, despite a long history, it would be
`fair to say that, until recently, polymorphism
`has been more of a scientific curiosity than an
`urgent challenge of commercial relevance.
`Pharmaceutical science has been largely re-
`sponsible for a change in this situation since
`
`most orally delivered APIs receive regulatory
`approval for a single crystal form or poly-
`morph and novel crystal forms are patentable.
`Awareness of the matter heightened following
`a now classic patent litigation between Glaxo
`and Novopharm in which Glaxo defended its
`patent for the form Il polymorph of ranitidine
`hydrochloride, the API in Zantaca’. The Glaxo
`patent on form I of ranitidine hydrochloride
`(US patent 4,128,658) expired on December 5
`1995, but
`the form II patent
`(US patent
`4,52 1,431) did not expire until 2002. Although
`Novopharm ultimately prevailed, Glaxo re-
`tained exclusivity beyond the patent expira-
`tion of form I for several years on what was at
`the time the bop-selling drug in the world. In
`addition to legal, regulatory and commercial
`considerations, polymorphism in drug sub-
`stances can also have direct clinical implica-
`tions since dissolution rates are sometimes
`
`polymorphism. However,
`by
`impacted
`although polymorphism might be long recog-
`nized and topically relevant to pharmaceuti-
`cal science [44], this does not mean that poly-
`morphs are predictable or that their discovery
`is routine despite McCrone’s statement on the
`subject in 1965 [-45]: “Every compound has
`different polymorphic forms and the number
`of forms known for a given compound is pro-
`portional
`to the time and energy spent in
`research on that compound.” This provocative
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 4 of 32)
`
`

`
`statement has often been debated and many
`solid-state scientists would be inclined to sup-
`port such an assertion. However, McCrone’s
`statement cannot
`realistically be proved
`through experiment and even today the nu1n-
`ber of publicly disclosed cases of polymorph-
`ism in organic compounds remains quite low
`based upon CSD statistics: only 8525 out of
`195,222 organic compounds archived are poly-
`morphic and since there must be at least two
`entries for each compound this represents
`32.2% of organic compounds; only 667 out of
`11,501 compounds with biological or pharma-
`cological activity are polymorphic (selected
`using
`keywords
`“activity,"
`“agent,"
`“biological,”
`“drug,”
`“pharmaceutical,"
`“phar1nacological”), representing just 52.9%
`of this subset. One must also bear in mind that
`
`the CSD is unlikely to be representative in the
`context of polymorphs since entries are biased
`by the compounds that have been of interest to
`crystallographers at particular points in time.
`The focus until recently has been upon mole-
`cular structure rather than crystal structure
`and many polymorphs probably remain
`unpublished.
`Although polymorphism might remain lar-
`gely unpredictable it can be rationalized and
`categorized through understanding the mole-
`cular and supramolecular structure of the
`compound in question, allowing us to define
`at least two classes of polymorphism [45]: con-
`
`CRYSTAI. FORM TYPES
`
`T91
`
`formations! polymorphism is the consequence
`of more than one conformer in the solid state
`
`(i.e., the shape of the molecule is different);
`packing polymorphism. occurs when rigid mo-
`lecules exhibit more than one packing ar-
`rangement. Packing polymorphs might be
`caused by different supramolecular synthons
`(i.e., the intermolecular connectivity is differ-
`ent) or they might retain their supramolecular
`synthons but exhibit different crystal packing.
`Such a situation might be termed supramole
`cular synthon polymorphism. Conformational
`polymorphism is exemplified by what is thus
`far the most promiscuous molecule in terms of
`the number of structurally characterized poly-
`morphs,
`5-1nethyl—2—[(2—nitrophenyl)amino]—
`3-thiophenecarbonitrile, a pharmaceutical in-
`termediate that has been called ROY because
`
`its eight crystallographically characterized
`polymorphs are red, orange, or yellow in col-
`or [46,47]: ROY is illustrated in Fig. 3, which
`highlights the portion of ROY that is respon-
`sible for its conformational flexibility. Six
`room temperature polymorphs of ROY were
`reported by Yu et al. in 2000 [46] and two
`additional polymorphs, Y04 and YT04, were
`reported in 2005 [47]. Y04 was prepared from
`a melt at room temperature, and YT04 was
`obtained via solid-state transformation of
`
`Y04. Y04 and YT04 exemplify polymorphs
`that would likely be missed by solvent-based
`screening, highlighting the experimental
`
`
`
`Figure 3. The molecular structure of the ON polymorph of 5—methyl-2-[(2—nitrophenyl)amino]-3-thiophe-
`necarbonitrile, ROY (CSD refcode :
`indicating the region of torsional flexibility.
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 5 of 32)
`
`

`
`192
`
`POLYMORPHIC CRYSTAL FORMS AND COCRYSTALS IN DRUG DELIVERY
`
`challenge of polymorph discovery. Packing
`polymorphism is exhibited by numerous APIs
`and exemplified herein by Piracetam and As-
`pirin. Piracetam, 2-oxo-pyrrolidineacetamide,
`is a nootropic drug that improves cognitive
`ability and it exhibits five structurally char-
`acterized polymorphs [48]. Although one of
`these polymorphs, the high-pressure form IV,
`is a conformational polymorph, forms I and II
`are examples of packing polymorphism caused
`by different supramolecular heterosynthons.
`Form I exists as a cyclic tetramer whereas
`form II forms infinite tapes in which a1nide—a-
`mide dimers are hydrogen bonded to adjacent
`dimers through amide-carboxamide N—I-I .
`.
`.
`0 hydrogen bonds. Aspirin had long been
`considered to represent an example of a com-
`pound that does not exhibit polymorphism.
`However, in 2005, metastable form II of as-
`pirin Was discovered during an attempted
`cocrystallization reaction [49] Forms I [50]
`and II are illustrated in Fig. 4, which reveal
`that both crystal forms of aspirin contain
`dimers that are sustained by the carboxylic
`acid supramolecular homosynthon. How-
`ever, C—H .. . O interactions between
`adjacent dimers are different and in turn
`cause different crystal packing. Subsequent
`Work has suggested that forms I and II
`might coexist within the same crystal
`(Fig. 5) [51].
`In conclusion, polymorphs can generally be
`rationalized through supramolecular con-
`cepts such as crystal engineering but this
`does not mean that they can yet be predicted
`
`from first principles. However, although
`one should not confuse crystal engineering
`with crystal structure prediction, crystal
`structure prediction using computer modeling
`has advanced considerably Within the past
`decade [52].
`
`2.2. Cocrystals
`
`2.2.1. What is a Cocrystal? That there is not
`yet a recognized definition of
`the term
`“cocrystal" has engendered debate on the sub-
`ject [27]. We have been using the following
`operating definition: a multiple component
`crystalline solid formed in a stoichiometric
`ratio between two compounds that are crystal-
`line solids under ambient conditions. At least
`
`one of these compounds is molecular (the co-
`crystal former) and forms supramolecular
`synthons(s) with the remaining component
`(s)
`[3a—3e]. That all components are solids
`under ambient conditions has important prac-
`tical considerations since synthesis of cocrys-
`tals can be achieved via solid-state methods
`
`(e.g., mechanochemistry) and chemists can
`execute a certain degree of control over
`the composition of a cocrystal since they can
`invoke molecular
`recognition,
`especially
`hydrogen bonding, during the selection of co-
`crystal formers. These features distinguish
`cocrystals from solvates and despite restric-
`tions they still
`represent a broad range
`of compounds since most molecular coin-
`pounds
`exist
`as
`solids under
`ambient
`conditions [53].
`
`
`
`Form I
`
`BISMEVD3
`
`Form II
`
`BISMEV
`
`Figure 4. Forms I and II of piracetam exhibit packing polymorphism because they exhibit different
`supramolecular synthons. Form I exists as a cyclic tetramer whereas form II forms infinite tapes.
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 6 of 32)
`
`

`
`CRYSTAL FORM TYPES
`
`T93
`
` Dimer
`
`Catemer
`
`Figure 5. The two polymorphs of aspirin are both based upon carboxylic dimer supraniolecular homosyn-
`thons. However, they differ in the manner in which adjacent dimers interact. In form I C—H .
`.
`. O dimers are
`formed whereas in form II the structure is sustained by (PH .
`.
`. 0 catemers.
`
`2.2.2. Why are Cocrystals of Interest to the
`Pharmaceutical Industry? Pharmaceutical co-
`crystals, that is, cocrystals in which the target
`molecule or ion is an active pharmaceutical
`ingredient, API, and the cocrystal former is a
`pharmaceutically acceptable molecule or ion,
`are emerging rapidly because of a number of
`factors including the following:
`
`o Design: Our scientific understanding of
`the noncovalent forces that sustain ino-
`
`lecular organic crystals has advanced to
`the extent that control over the stoichio—
`
`metry and composition of cocrystals can
`be asserted. This is not ordinarily the
`case for polymorphs and solvates for
`which high—throughput screening, which
`to a certain extent practices serendipity,
`tends to be relied upon rather than de-
`sign, or for salts, which require an ioniz-
`able functional group.
`
`- Discovery: That mechanochemistry can
`be utilized to synthesize cocrystals has
`been known since the first cocrystals
`were discovered in the 1840s by dry
`grinding [25a], but it has only recently
`been realized and accepted that “solvent-
`drop” or “liquid assisted" grinding are
`preferred methodologies [54]. Indeed, it
`is fair to assert that cocrystals are most
`readily accessible through solvent—free or
`solvent—reduced methods although other
`techniques such as slurrying [55] and
`solution [56] are complementary.
`
`Diversity: It has become apparent that
`pharmaceutical cocrystals always exhibit
`different
`physicochemical
`properties
`compared to the pure crystal form(s) of
`APls, that a given API might form cocrys-
`tale with dozens of cocrystal formers and
`that some of these cocrystals might ex-
`hibit enhanced solubility or stability to
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 7 of 32)
`
`

`
`194
`
`POLYMORPHIC CRYSTAL FORMS AND COCRYSTALS IN DRUG DELIVERY
`
`hydration. Therefore, pharmaceutical co-
`crystals represent an opportunity to di-
`versify the number of crystal forms of a
`given API and in turn fine-tune or even
`customize its physicochemical properties
`without the need for chemical (covalent)
`modification.
`
`- Development: Whereas pharmaceutical
`cocrystals can be designed using crystal
`engineering strategies this does not
`mean that details of their crystal struc-
`tures or physical properties can be pre-
`dicted before they have been measured.
`Therefore, one might assume that it will
`be possible for pharmaceutical cocrystals
`of existing APls to be patented as new
`crystal forms and, if they exhibit clinical
`advantages, developed as new drugs.
`This has implications for drug develop-
`ment because it abbreviates some aspects
`of drug development timelines and initi-
`gates costs and risks related to discovery
`and toxicology of new APls.
`
`- Delivery: As mentioned earlier, being
`able to fine-tune solubility can be a cri-
`tical factor that influences the clinical
`
`performance of an API if its bioaVailabil-
`ity is affected by rate of dissolution. This
`is generally considered to be important
`for BCS Class Ii APR: [57], perhaps the
`most common classification for the cur-
`
`rent generation of APIs.
`
`The August 2008 release of the CSD contains
`structural information on 456,628 organic,
`metalwrganic, and organometallic crystal
`structures, but there is not a great deal of
`structural information on cocrystals. There
`are only two cocrystal entries prior to 1960
`and even today there are only ca. 2083 (0.46%
`of the CSD) hydrogen-bonded cocrystals ver-
`sus 50,0]9 hydrates 00.95% of the CSD).
`Therefore, it would be fair to summarize co-
`crystals as being a long known but little stu-
`died class of compounds. Nevertheless, the
`realization that there will be multiple cocrys-
`tal formers for a given API makes pharma-
`ceutical cocrystals somewhat diverse in terms
`of their composition. The scope of available
`cocrystal formers is not yet set but even if it is
`limited to “generally regarded as
`safe”
`
`(GRAS) compounds or compounds that have
`already been approved by the federally man-
`dated Food and Drug Administration (FDA)
`for use in formulation such as a “salt formers,"
`there could be 100 or more possible pharma-
`ceutically acceptable cocrystal formers for
`an API.
`
`In terms of the pharmaceutical industry,
`perhaps the earliest example of a pharmaceu-
`tical cocrystal was reported in a 1934 French
`patent that disclosed cocrystals of barbitu-
`rates with 4-oxy-5-nitropyridine, 2-ethoxy-5-
`acetaminopyridine,
`N-methyl-oz-pyridone,
`and maminopyridine [36]. In 1995, Eli Lilly
`and Co. patented complexes of cephalosporins
`and carbacephalosporins, a class of B—lactam
`antibiotics, with parabens and related coin-
`pounds [57]. In terms of the scientific litera-
`ture, there were few reports ofpharmaceutical
`cocrystals until the past decade. However,
`Caira demonstrated that “old” drugs such as
`sulfonamides can form cocrystals [58] and also
`emphasized
`their
`potential
`in
`drug
`development.
`
`2.2.3. Design of Cocrystals A crystal engi-
`neering experiment typically involves CSD
`surveys followed by experimental work to pre-
`pare and characterize new compounds that
`are sustained by supramolecular synthons.
`Supramolecular synthons facilitate under-
`standing of the supramolecular chemistry of
`the functional groups present in a given ino-
`lecule and are prerequisites for designing a
`cocrystal since they facilitate selection of an
`appropriate cocrystal
`former(s). However,
`when multiple functional groups are present
`in a molecule, the CSD rarely contains enough
`information to address the hierarchy of the
`possible supramolecular
`synthons. Fortu-
`nately, the hierarchy of the suprainolecular
`synthons that can occur for common func-
`tional groups such as carboxylic acids, amides,
`and alcohols with emphasis upon supram.ole—
`cular heterosynthons
`is becoming better
`defined [12d,e]. Furthermore, it is becoming
`evident that such interactions are key to im-
`plementing a design strategy for cocrystals in
`which a target molecule forms cocrystals with
`cocrystal formers that are carefully selected
`for their ability to form supramolecular het-
`erosynthons with the target molecule.
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 8 of 32)
`
`

`
`The design aspect of cocrystals is illu-
`strated if one focuses upon carboxylic acids,
`perhaps the most important and widely stu-
`died functional group in the context of phar-
`maceutical cocrystals since carboxylic acids
`represent ca. 25% of marketed drugs and car-
`boxylic acids are commonly used as salt for-
`mers or excipients. The CSD enables statisti-
`cal surveys of intermolecular contacts as well
`as intramolecular connectivity and it is there-
`fore a powerful tool for addressing supremo-
`lecular chemistry in the solid state. A survey of
`the CSD revealed that there Were 8154 organ-
`ic carboxylic acids in the CSD as of August
`2008. However, an analysis of intermolecular
`contacts in this subset revealed that only 1926
`of these carboxylic acids exhibit the carboxylic
`acid dimer
`supramolecular homosynthon
`(Fig. 6) and that only 143 exhibit the car-
`boxylic acid catemer motif. So what about the
`remaining 75% of carboxylic acids that have
`been crystallographically characterized‘? As
`
`CRYSTAL FORM TYPES
`
`1955
`
`revealed by Fig. 7, there is a tendency for
`carhoxylic acids to form supramolecular het-
`erosynthons with, for example, chloride an-
`ions and aromatic nitrogen moieties. Further-
`more,
`the statistics seem to strongly favor
`these supramolecular heterosynthons over
`the corresponding supramolecular homosyn—
`thons. For example,
`there are 277 crystal
`structures that contain both a carboxylic acid
`and a chloride anion and 180 of them exhibit
`
`the carboxylic acid chloride supramolecular
`heterosynthon. In only one of this subset of
`277 crystal structures does the carboxylic
`dimer exist. The statistics are similar for the
`
`carboxylic acid—pyridyl supramolecular het-
`erosynthon. There are 606 crystal structures
`that contain both a carboxylic acid and a pyr-
`idyl moiety and 415 of them exhibit the car-
`boxylic acid—pyridyl supramolecular hetero-
`synthon. In only 25 of this subset of 606 crystal
`structures does the Carboxylic dimer exist. In
`short, although these data are raw and une-
`
`900
`
`300
`
`700
`
`600
`
`500
`
`400
`
`300
`
`200
`
`100
`
`-
`
`-
`
`------ --
`
`2.8
`
`3.0
`
` . . . . . . . . . . . . . . . . . . . . . . . .. .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
`
`3.
`
`3.4
`
`3.6
`
`Figure 6. Distribution of carboxylic dimer contacts between 2.4 and 3.615; in organic only carboxylic acid
`crystal structures in the CSD. The distribution reveals 1926 H-bonded contacts between 2.55 and 2.80A
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 9 of 32)
`
`

`
`196
`
`POLYMORPHIC CRYSTAL FORMS AND COCRYSTALS IN DRUG DELIVERY
`
`50’
`
`45'
`
`4(1-
`
`30‘
`
`25‘
`
`20 “
`
`15'
`
`10' .
`
`0.
`2.?!)
`
`.
`280 2.90
`
`..
`3.00
`
`.
`
`I-17:21.1
`--.-_}f‘F—‘.l'-'!.
`3.10
`3.20
`330 340
`3.50
`
`IEO
`
`150
`
`140
`
`120
`
`100
`
`30
`
`60
`
`40
`
`EC:
`
`0
`2.2
`
`24
`
`
`
`,
`3.0
`
`.
`
`32
`
`3.4
`
`_
`3.5
`
`2‘
`
`
`(8)
`
`(b)
`
`(a) Distribution of carboxylic acid—c-hloride anion contacts between 2.4- and 3.615;. in organic only
`Figure 7.
`carboxylic acid crystal structures that also contain chloride anions. There are 180 short contacts between 2.7
`and 3.3 A (b) Distribution ofcarboxylic acid—aromatic nitrogen contacts between 2.4 and 3. 5 Am organic only
`carboxylic acid crystal structures that also contain aromatic nitrogen moieties.
`
`dited, it strongly suggests that if the relevant
`functional groups are in different molecules
`then a Cocrystal involving supramolecular
`heterosynthons is likely to occur over the
`corresponding single component structures
`that would be sustained by supramolecular
`homosynthons. This principle is exemplified
`by several of the case studies presented
`herein.
`
`2.2.4. Polymorphs, Solvates, and Hydrates of
`Cocrystals There remains a dearth of sys-
`tematic structure and property information
`on cocrystals. However, at this point there is
`no reason to believe that pharmaceutical co-
`crystals Will be more or less promiscuous than
`single component APIs When it comes to crys-
`tal form diversity. For example, both confor-
`mational and packing polymorphs have been
`observed in cocrystals. Figure 8 reveals that
`rotation around the central C-—C bond in 4,4’-
`biphenol can afford conformational poly-
`morphism in the 2 : 1 cocrystal of 4-cyanopyr-
`idine and 4,4’-biphenol [59]. Figure 9 reveals
`how a model cocrystal based upon the pyri-
`dine-carboxylic
`acid supramolecular
`syn-
`thons [60], a supramolecular synthons that is
`
`particularly relevant to APIs, exhibits pack-
`ing polymorphism. In this case, packing poly-
`morphism manifests itself through networks
`and interpenetration in the polymorphs of the
`3: 2 cocrystal of 4,4’-bipyridyletbane and tri-
`mesic acid [6]].
`
`3. CASE STUDIES THAT DEMONSTRATE
`HOW CRYSTAL FORMS CAN IMPACT
`PHYSICOCHEMICAL PROPERTIES AND/OR
`BIOAVAILABILITY
`
`3.1. Case Studies of Polymorphs
`
`The impact of polymorphism on solubility was
`addressed by Pudipeddi and Serajuddin, who
`collated data on 81 polymorphic pairs [62].
`The majority of these polymorphs (63/81) were
`observed to exhibit a solubility ratio of £2 and
`only one pair of polymorphs exhibits a solubi-
`lity ratio of >10. This outlier, premafloxacin
`(Fig. 10), is a broad—spectrum antibiotic initi-
`ally developed for veterinary use by Pharma-
`cia and Upjohn,
`Inc. and it is chemically
`known as [S-(R*,S’“)]-1-cyclopropyl-6-fluoro-
`1,4-dihydro-8-methoxy-7 -{3-[1-methylamino)
`ethyl]-1-pyrrolidinyl}-4-oxo-3-quinolinecarbo
`
`Janssen Ex. 2027
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 10 of 32)
`
`

`
`CASE STUDIES
`
`197
`
`Form I
`Fonn II
`
`U
`
`Figure 8. The conformational polymorphs exhibited by the 2:1 cocrystal of 4-cyanopyridine and 4,4’-
`biphenol.
`
`
`
`Figure 9. The 3:2 cocrystal of 4,4’ bipyridylethane and trimesic acid exhibits two packing polymorphst a (6,3)
`honeycomb network that with 3-fold parallel interpenetration and a (10,3)-a 3D network with 18-fold
`interpenetration.
`
`F
`
`N
`
`COOH
`
`O
`
`N
`
`|
`
`-.___N
`
`H
`
`OCH3 A
`
`Figure 10. Molecular structure of prernafloxacin.
`
`xylic acid. This fluoroq