Organic Process Research & Development 2009, 13, 242–249
`
`Crystallization Process Development for a Stable Polymorph of Treprostinil
`Diethanolamine (UT-15C) by Seeding
`
`Hitesh Batra,†,* Raju Penmasta,† Kenneth Phares,‡ James Staszewski,‡ Sudersan M. Tuladhar,† and David A. Walsh†
`United Therapeutics Corporation, Research and DeVelopment Department, 1040 Spring Street, SilVer Spring,
`Maryland 20910, U.S.A., and United Therapeutics Corporation, Research and DeVelopment Department, 55 T.W. Alexander
`DriVe, Research Triangle Park, North Carolina 27709, U.S.A.
`
`Abstract:
`Process development of treprostinil diethanolamine salt (UT-15C)
`involved the development of crystallization and slurry protocols
`to address the polymorph and morphology control issues. Two
`forms of UT-15C were evaluated by differential scanning calo-
`rimetry (DSC), X-ray powder diffraction (XRPD) and thermo-
`gravimetric analysis (TGA). Two crystallization solvent systems
`were developed to produce the thermodynamically stable form in
`high quality and yield. One solvent system gave dense particles
`while the other gave lighter and fly-away particles. Slurrying the
`lighter particles in heptane converted them to denser particles.
`The protocol was executed successfully on large-scale cGMP
`batches.
`
`Introduction
`Polymorphism1 is defined as the ability of a substance or
`compound to crystallize into different, yet chemically identical,
`crystalline forms. In the pharmaceutical industry, the signifi-
`cance of polymorphism was realized recently through some
`relatively high-profile cases.2 In particular, the unexpected
`appearance in early 1998 of a more thermodynamically stable
`form (Form II) of ritonavir2 (Norvir, Abbott Laboratories,
`protease inhibitor for the treatment of HIV), with different
`dissolution properties compared to those of the earlier com-
`mercial Form I. Form II is <50% as soluble as Form I, resulting
`in the observed poor dissolution behavior and eventual with-
`drawal of the capsule from the market. This incident had serious
`implications for the marketed product and the patients receiving
`the drug.2a,b The project was suspended until a modified
`procedure was found. Renitidin, sertraline, and frentizole are
`some important examples of pharmaceuticals that exhibit
`polymorphism.3 These incidents have led to an increased
`awareness of the importance of early-stage polymorph identi-
`fication and characterization. It is evident from the number of
`publications and patents being granted that polymorphism is a
`
`* To whom correspondence should be addressed. Telephone: 240-821-1902.
`Fax: 301-608-0376. E-mail: hbatra@unither.com.
`† United Therapeutics Corporation, Maryland.
`‡ United Therapeutics Corporation, North Carolina.
`(1) (a) Chen, S.; Guzei, I. A.; Yu, L. J. Am. Chem. Soc. 2005, 127, 9881–
`9885. (b) Price, P. P.; Grzesiak, A. L.; Matzger, A. J. J. Am. Chem.
`Soc. 2005, 127, 5512–5517. (c) Zhou, J.; Kye, Y. S.; Harbison, G. S.
`J. Am. Chem. Soc. 2004, 126, 8392–8393. (d) Kim, S.; Wei, Chenkou.;
`Kiang, S. Org. Process Res. DeV. 2003, 7, 997–1001. (e) O’Sullivan,
`B.; Barrett, P.; Hsiao, G.; Carr, A.; Glennon, B. Org. Process Res.
`DeV. 2003, 7, 977–982. (f) Beckmann, W.; Otto, W.; Budde, U. Org.
`Process Res. DeV. 2001, 5, 387–392. (g) Beckmann, W. Org. Process
`Res. DeV. 2000, 4, 372–383.
`•
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`242
`Published on Web 02/25/2009
`
`Scheme 1
`
`topic of high importance for the pharmaceutical industry. To
`cite a few: a publication on a polymorph study of the L-arginine
`salt of ragalitazar describes evaluation of its 12 polymorphs4
`and a paper about sertraline3 describes eighteen polymorphic
`forms assessed via high-throughput crystallization. There were
`over 3600 crystallizations conducted during the course of this
`study.5 United States patent U.S. 5,700,8206 discloses six
`polymorphs of troglitazone; U.S. 5,248,6997 discloses five
`polymorphic forms of sertraline hydrochloride (Zoloft); Euro-
`pean patent EP 4906488 describes four polymorphic forms of
`frentizole; and EP 0225279 also deals with the subject of
`polymorphism in drugs.
`
`(2) (a) Bauer, J.; Spanton, S.; Henry, R.; Quick, J.; Dziki, W.; Porter,
`W.; Morris, J. Pharm. Res. 2001, 18, 859–866. (b) Morissette, S. L.;
`Soukasene, S.; Levinson, D.; Cima, M. J.; Almarsson, O. Proc. Natl.
`Acad. Sci. U.S.A. 1995, 92, 2484–2488. (c) Chemburkar, S. R.; Baur,
`J.; Deming, K.; Spiwek, H.; Patel, K.; Morris, J.; Henry, R.; Spanton,
`S.; Dziki, W.; Porter, W.; Quick, J.; Bauer, P.; Donaubauer, J.;
`Narayanan, B. A.; Soldani, M.; Riley, D.; McFarland, K. Org. Process
`Res. DeV. 2002, 4, 413.
`(3) (a) Agatonovic-Kustrin, S.; Wu, V.; Rades, T.; Saville, D.; Tucker,
`I. G. Int. J. Pharm. 1999, 184, 107–114. (b) Agatonovic-Kustrin, S.;
`Rades, T.; Wu, V.; Saville, D.; Tucker, I. G. J. Pharm. Biomed. Anal.
`2001, 25, 741–750. (c) Van der schaaf, P. A.; Schwarzenbach, F.;
`Kirner, H.-J.; Szelagiewicz, M.; Marcolli, C.; Burkhard, A.; Peter, R.
`World Intellectual Property Organization WO/2001/032601, 2001. (d)
`Novoselsky, A.; Glaser, R. Magn. Reson. Chem. 2002, 40, 723–728.
`(e) Borochovitch, R.; Mendelovici, M.; Nidam, T.; Tenengauzer, R.;
`Hrakovsky, J.; Aronhime, J. U.S.Patent 2007:0213404, 2007. (f)
`Srisilla, R.; Potlapally, R. K.; Mamillapalli, R. S.; Gaddam, O. R.
`World Intellectual Property Organization WO/2003/066612, 2003. (g)
`Cord, J.; Chebiyyam, P.; Mamillapalli, R. S.; Krishnamurthi, V.; Seella,
`V. R.; Gaddam, O. R. World Intellectual Property Organization WO/
`2002/026737, 2002.
`(4) Raju, S.; Kumar, R.; Vyas, K.; Rao, D. S.; Sarma, M. R.; Reddy,
`S. V.; Nirmala, M.; Reddy, G. O. Org. Process Res. DeV. 2003, 7,
`962–969.
`(5) Remenar, J. F.; MacPhee, J. M.; Larson, B. K.; Tyagi, V. A.; Ho,
`J. H.; McIlroy, D. A.; Hickey, M. B.; Shaw, P. B.; Almarsson, O.
`Org. Process Res. DeV. 2003, 7, 990–996.
`(6) Vyas, K.; Prabhakar, C.; Rao, D. S.; Sarma, M. R.; Reddy, G. O.;
`Ramanujam, R.; Chakrabarthi, R. U.S. Patent 5,700,820, 1997; Chem.
`Abstr. 1997, 127, 190731.
`(7) Sysko, R. J.; Allen, D. J. M. U.S. Patent 5,248,699, 1994; Chem. Abstr.
`1994, 120, 38134.
`
`10.1021/op800239m CCC: $40.75  2009 American Chemical Society
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 1 of 8
`
`

`

`Figure 1. DSC overlay of treprostinil diethanolamine (top to bottom) and sample after storage.
`
`Treprostinil (1, UT-15) (Scheme 1) belongs to a class of
`stable analogues of PGI2 called benzindene prostacyclins.10 UT-
`15 (1) is effective in the treatment of pulmonary arterial
`hypertension (PAH), a debilitating and often fatal lung disease,
`and has been approved by the FDA for treatment of PAH.11
`UT-15 is delivered subcutaneously or intravenously via a
`microinfusion device, has a relatively short biological half-life
`and is not degraded upon passage through the lungs.
`The goal of this project was to indentify an oral prostacyclin
`analogue for the treatment of PAH that was bioavailable, soluble
`in water, and easy to deliver. Various salts of UT-15 (1) were
`screened, and the treprostinil diethanolamine salt (UT-15C, 3)
`showed promising physical characteristics for formulation as
`an oral drug.
`Polymorphism. Two polymorphic forms of UT-15C (3),
`Form A and Form B, have been identified to date. Preparation
`of early developmental batches of UT-15C produced Form A.
`However, upon storage, some of Form A partially converted
`to Form B to form a mixture of Forms A and B (based on
`melting point and confirmed by differential scanning calorimetry
`(DSC) and XRPD data; Figures 1 and 2). On the basis of these
`observations, it was hypothesized that Form B was thermody-
`namically more stable and Form A was a metastable form, but
`kinetically crystallized more readily.
`This observation was also further supported by solubility
`and heat of solution results. According to the “Oswald rule of
`stages”,12 often in crystallization processes a metastable form
`crystallizes from the solution initially and transforms to a more
`stable form at a rate specific to the compound, depending upon
`the relative solubility of the two phases in the solvent system.
`This phenomenon is widely observed with many active
`pharmaceutical ingredients (APIs) in the pharmaceutical indus-
`A) and Form B
`try. The melting temperatures of Form A (Tm
`B) were about 103 and 107 °C, respectively, and the
`(Tm
`
`(8) Timko, R. J.; Clements, A.; Bradway, R. J. EP Patent 0,490,648, 1992;
`Chem. Abstr. 1992, 117, 97344.
`(9) Bolandi, A.; Molinari, E. EP Patent 0,022,527, 1982; Chem. Abstr.
`1981, 94, 162743.
`(10) Moriarty, R. M.; Rani, N.; Enache, L. A.; Rao, M. S.; Batra, H.; Guo,
`L.; Penmasta, R. A.; Staszewski, J. P.; Tuladhar, S. M.; Prakash, O.;
`Crich, D.; Hirtopeanu, A.; Gilardi, R. J. Org. Chem. 2004, 69, 1890–
`1902, and references therein.
`(11) (a) Lewis, P. J., O’Grady, J., Eds. Clinical Pharmacology of Prosta-
`cyclin; Raven Press: New York, 1981. (b) Vane, J., O’Grady, J., Eds.
`Therapeutic Applications of Prostaglandins; Edward Arnold: London,
`UK, 1993. (c) Vane, J. R., Bergstrom, S., Eds. Prostacyclin; Raven
`Press: New York, 1979. (d) Moncada, S.; Vane, J. R. Pharmacol. ReV.
`1979, 30, 293–331.
`(12) Ostwald, W. Z. Phys. Chem. 1897, 22, 289.
`
`Figure 2. X-ray powder diffraction (XRPD) pattern comparison
`of treprostinil diethanolamine salt (UT-15C) Form A, Form A
`after storage, and Form B.
`measured heat of fusion for Forms A and B were 109.0 J/g
`(53.955 kJ/mol) and 109.2 J/g (54.054 kJ/mol), respectively.
`The synthesis of UT-15C (3), faced a number of challenges
`during the early development of the final crystallization step.
`The first problem to overcome was the tendency of the
`compound to oil-out (formation of gummy-mass) by finding
`the right solvent ratio. The second obstacle was designing a
`crystallization process that produced the desired form (Form
`B) consistently.
`In light of the above-mentioned issues, it was important to
`develop a more controlled crystallization process to achieve only
`one form and desired morphology from a formulation stand-
`point. This paper describes the problems faced during the
`crystallization development and provides the findings and
`solutions that successfully resulted in a robust crystallization
`process for UT-15C, producing the desired form with desired
`particle properties (Figure 3 shows the overlay of XRPD pattern
`of Form A and Form B). The peaks at 13.7° 2θ and 17.2° 2θ
`were the characteristic values for Forms A and B, respectively,
`in the XRPD analysis.
`Form A is a crystalline material that melts at 103-104 °C.
`Form B is a crystalline form that melts at a higher temperature,
`106-108 °C, and was observed to form under a variety of
`conditions (Figure 4 shows the DSC and thermogravimetric
`analysis (TGA) of Form A and Form B). Evaluation of the
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`•
`
`243
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 2 of 8
`
`

`

`and DICVOL. The indexing method searches for crystal unit
`cells initially containing one molecule per asymmetric unit and
`then proceeds by increasing the number of molecules per
`asymmetric unit until viable solutions are found. The indexing
`begins with the highest orthorhombic symmetry and then
`proceeds to lower symmetries through to monoclinic and
`triclinic. Orthorhombic solutions for each form were indepen-
`dently found that describe all of the measured peaks in each
`experimental XRPD pattern within a 2% error in precision. The
`space group and unit cell dimensions for each form can initially
`be described as:
`Form A: P212121, a ) 45.736 Å, b ) 12.737 Å, c ) 4.704
`Å, volume ) 2740 Å3
`Form B: P212121, a ) 45.212 Å, b ) 12.482 Å, c ) 4.811
`Å, volume ) 2715 Å3
`The unit cell parameters were refined and electron density
`models were evaluated using MAUD. Based on the possible
`indexed unit cells the measured XRPD patterns were fit to find
`solutions which provide the best description of the measured
`data. These unit cell results present the smallest and most precise
`determination of unit cell volumes and improve upon the initial
`precision to within 0.5% resolution limit.
`Form A: P212121, a ) 45.3676 Å, b ) 12.6856 Å, c )
`4.6893 Å, volume ) 2699 Å3
`Form B: P212121, a ) 45.1804 Å, b ) 12.4707 Å, c )
`4.8283 Å, volume ) 2720 Å3
`The initial indexing results indicate that Form B has a smaller
`volume. Upon refinement, unit cell results show the inverse is
`true. However, in each case, the volume differences fall within
`the precision error or resolution limit of the calculation method.
`This indicates that the unit cell volumes are actually nearly
`identical from an XRPD perspective.
`For structures which appear to be so similar (Forms A and
`B), the differences in the large lattice parameters determine
`stability. The largest
`lattice parameter corresponds to the
`weakest bond direction and, therefore, the most likely to fail
`(Donnay-Harker).13 This indicates that Form B is the more
`stable form, but only by a fractional amount. The modified
`Donnay-Harker13 theory predicts the same morphology for
`both Forms A and B, and we observed that Forms A and B
`were similar (needlelike) as predicted (Figure 5). Both forms
`readily dissolve in water with solubilities greater than 500 mg/
`mL (pH 6.95).
`Form A has hydrogen bonds linking cations together along
`the shortest crystallographic c-axis and the anions together along
`the medium crystallographic b-axis (Figure 6). Although these
`hydrogen-bond networks give an indication as to the origin of
`differences between the two forms, it must be pointed out that
`the unit cell values for Forms A and B suggest that the bonding
`networks should be reversed (the unit cell values have higher
`precision than the placement of hydrogen bonds). In Form B
`(Figure 7), the c-axis is longer, indicating a weaker bond
`direction, and in Form A, the b-axis is shorter, indicating a
`stronger bond direction.
`
`(13) Khoo, I. C.; Simoni, F. Physics of Liquid Crystalline Materials; CRC
`Press: Boca Raton, FL, 1991; p 28, ISBN: 2881244815.
`
`Figure 3. Overlay of XRPD pattern of Form A (top) and Form
`B (bottom).
`
`relative thermodynamic relationships of Form A and Form B
`indicated that Form B was the more thermodynamically stable
`form. The energy difference between the two forms was found
`to be about 0.2 J/g (0.1 kJ/mol). The crystal structures of the
`two forms of UT-15C appear to be very similar, and the small
`differences in the large lattice parameters account for the similar
`stabilities of UT-15C Forms A and B. The experimental XRPD
`patterns of Forms A and B were analyzed to provide unit cell
`parameters for each form.
`
`Figure 4. DSC of Form B (top) and DSC and TGA of Form A
`(bottom).
`
`The experimental XRPD patterns of Form A and Form B
`were indexed using the SSCI indexing software (version 1.8.4)
`
`244
`
`•
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 3 of 8
`
`

`

`Figure 7. Packing diagram of UT-15C Form B viewed down
`the b-axis.
`
`in various solvents, Form A and Form B did not have noticeable
`differences in solubility. On the basis of these solubility data,
`slurry, and crystallization experiments were conducted to obtain
`Form B exclusively.
`
`Table 1. Solubility of treprostinil diethanolamine (UT-15C)
`at 25 °C
`
`solvent
`
`acetone
`ethanol/acetone (1:5)
`ethanol/acetone (1:6)
`ethanol/acetone (1:7)
`ethanol/acetone (1:8)
`ethanol (EtOH)
`ethyl acetate (EtOAc)
`ethanol/ethyl acetate (1:5)
`ethanol/ethyl acetate (1:6)
`ethanol/ethyl acetate (1:7)
`ethanol/ethyl acetate (1:10)
`1,4-dioxane
`2-propanol (IPA)
`methyl tert-butyl ether (MTBE)
`ethanol/MTBE (1:7)
`tetrahydrofuran (THF)
`toluene
`water
`IPA/MTBE (1:1)
`IPA/MTBE (1:2)
`IPA/MTBE (1:3)
`IPA/MTBE (1:5)
`
`solubility (mg/mL)
`2
`9
`6
`5
`3
`110
`1
`3
`2
`1
`<1
`<3
`9
`<3
`<2
`3
`<2
`>500
`13
`5
`2
`1
`
`Several slurry preparations of UT-15C in various solvent/
`antisolvent ratios and solvent volumes were performed. Initially,
`the conversion from Form A to Form B occurred within 23-26
`h at lower solvent volumes of isopropyl alcohol (4 mL IPA/g)
`at both 1:1 and 1:2 ratios of isopropyl alcohol (IPA) and methyl
`tert-butyl ether (MTBE). No conversion was observed using
`higher solvent volumes 8 mL/g slurry and 12 mL/g slurry
`utilizing the 1:1 and 1:2 IPA/MTBE solvent system (Table 2).
`The two forms were evaluated for their relative thermodynamic
`stability by slurry interconversion experiments conducted in a
`mixture of IPA and MTBE at various temperature conditions
`for several hours.
`Form A was completely converted to Form B as confirmed
`by XRPD (Figure 8), and DSC (Figure 9). Initial studies
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`•
`
`245
`
`Figure 5. Optical microscope images of crystals Form A (top)
`and Form B (bottom).
`
`Figure 6. Packing diagram of UT-15C Form A viewed down
`the c-axis.
`Results and Discussion
`Various methods for obtaining polymorph B were consid-
`ered, and a large number of experiments were conducted using
`several solvents with emphasis on slurry and crystallization
`experiments. Table 1 shows the solubility data of UT-15C (3)
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 4 of 8
`
`

`

`Table 2. Slurry preparation attempts of treprostinil
`diethanolamine Form B using isopropyl alcohol/methyl
`tert-butyl ether (IPA/MTBE) at 25 °C
`solvent
`solid/
`ratio (v/v)
`solvent ratio (w/v)
`-
`-
`1:1
`1:4
`
`1:1
`
`1:2
`
`1:2
`
`1:2
`
`1:3
`
`1:5
`
`1:8
`
`1:4
`
`1:8
`
`1:12
`
`1:12
`
`1:12
`
`slurry
`time (h)
`0
`5.25
`7.25
`23.25
`1
`7
`24
`18.5
`26
`1
`2.5
`6
`23
`1
`23
`1
`5
`24
`1
`5
`24
`
`XRPD
`result
`A + B
`A + B
`A + B
`B
`A + B
`A + B
`A + B
`A + B
`B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`A + B
`
`provide Form B on a consistent basis. As Form B was
`thermodynamically more stable than Form A, it was important
`to isolate Form B and therefore, it was necessary to ensure that
`crystallization occurred slowly and in a controlled manner.
`Seeding with Form B prior to start of crystallization was helpful
`to obtain the desired form. Crystallization using a mixture of
`IPA/MTBE at various ratios was studied but less-stable Form
`A was obtained.
`Development of a new crystallization protocol involved
`investigations of various solvent systems such as ethanol/
`acetone and ethanol/ethyl acetate. Both ethanol/acetone
`and ethanol/ethyl acetate solvent systems provided prom-
`ising results. Various experiments were conducted using
`these two solvent systems, and various parameters were
`investigated to identify a process using either the ethanol/
`ethyl acetate or ethanol/acetone solvent systems that would
`consistently produce Form B. The variables studied
`included: (i) solvent ratio, (ii) seeding with Form B, and
`(iii) cooling rate during crystallization. Using a 1:7 ratio
`of ethanol/acetone and seeding with 1% Form B at 40 °C
`provided Form B with high quality and yield (>90%) as
`confirmed by XRPD and melting point data. Using various
`ratios of ethanol/acetone such as 1:5 provided predomi-
`nantly Form A, and 1:6 provided predominantly Form B;
`however, yields were slightly lower (85-90%) than using
`a ratio of 1:7. When a ratio of 1:8 ethanol/acetone was
`used, a mixture of Forms A and B was obtained as
`confirmed by melting point. When crystallization was
`performed without any seeds of Form B, a mixture of
`
`produced Form B by slurry experiments using a mixture of IPA
`and MTBE, but the process was not reproducible on large scale.
`From these trials, it was concluded that some of the conditions
`were not appropriate for obtaining only Form B and could be
`ruled out (i.e., fast cooling and crashing the compound out at
`low temperature always gave the less stable Form A).
`Concurrent to the slurry experiments, several crystallization
`systems were examined to determine if they would exclusively
`
`Figure 8. XRPD Patterns of UT-15C samples from 1:1 IPA/MTBE, 4 mL/g slurry (top to bottom: initial, 5.25 h, 7.25 h, and
`23.25 h).
`
`246
`
`•
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 5 of 8
`
`

`

`Figure 9. DSC thermogram of treprostinil diethanolamine (UT-15C; mixture of Forms A and B).
`
`Table 3. Summary of various crystallization conditions and
`the resulting form(s) as determined by XRPD
`ratio
`Form A
`v/v
`or B
`1:5
`A major
`1:6
`B major
`1:7
`B
`1:7
`A major
`A + B
`1:8
`A + B
`1:5
`A + B
`1:6
`1:7
`B
`1:7
`A major
`1:10
`A major
`
`seed of
`Form B (%)
`1
`1
`1
`no seed
`1
`1
`1
`1
`no seed
`1
`
`entry
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`
`solvents
`EtOH/acetone
`EtOH/acetone
`EtOH/acetone
`EtOH/acetone
`EtOH/acetone
`EtOH/EtOAc
`EtOH/EtOAc
`EtOH/EtOAc
`EtOH/EtOAc
`EtOH/EtOAc
`
`Forms A and B was obtained with greater percentage of
`Form A (Table 3).
`From these experiments, it was obvious that two conditions
`were required to produce Form B: (i) seeding with Form B to
`induce nucleation and (ii) maintaining a 1:7 ratio of ethanol/
`acetone while controlling temperature. Nucleation was induced,
`at a 1:7 ratio of ethanol/acetone, by seeding to the clear solution
`with 1% Form B at 40 °C. At these conditions, most of the
`seeds remain undissolved to induce nucleation. Once nucleation
`started, cooling was stopped for ∼1 h to allow the crystallization
`to progress at low supersaturation. Hence, the crystallization
`occurred under controlled conditions. As the crystalliza-
`tion proceeded, more crystals started to grow, and at this stage
`the solution was kept at 35 °C for 14 h and then cooled slowly
`to ambient at a rate of 3 °C/h. This crystallization process
`provided Form B consistently. In later experiments, once
`nucleation was induced by adding 1% seed of Form B at 40
`°C to a clear solution of UT-15C, cooling was stopped for ∼1
`h to allow the crystallization to progress at low supersaturation.
`At this stage the mixture was cooled to ambient conditions at
`a cooling rate of 1 °C /h, and this crystallization process
`provided Form B consistently.
`Various ratios of ethanol/ethyl acetate such as 1:5 or 1:6
`provided a mixture of Forms A and B, and yields were slightly
`lower (85-90%) than using a ratio of 1:7 (Table 3). A ratio of
`ethanol/ethyl acetate, 1:7, produced Form B as the major form
`
`as confirmed by XRPD and yields were in the range of
`90-95%. Crystallization of UT-15C using a 1:7 ratio of either
`ethanol:acetone or ethanol/ethyl acetate combined with 1%
`polymorph B seeding at 45-50 °C provided Form B. However,
`one noteworthy point is the crystallization from ethanol/acetone
`provided a denser material as compared to a light, fly-away
`material obtained from crystallization using ethanol/ethyl ac-
`etate. Table 3 summarizes the results of various conditions used
`to obtain Form B.
`Finally, ethanol/acetone was selected as the solvent system
`for final recrystallization of UT-15C because of the acceptable
`physical properties of the final material. As scale-up of this
`crystallization process continued, the yields and form purity
`improved, but the particle obtained was rod-shaped (Figure 10)
`which was not ideal for formulation development.
`A dry-milling operation was not desirable for this compound
`because UT-15C (3) is a highly potent prostaglandin mimic,
`
`Figure 10. Particle shape image before heptane slurry.
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`•
`
`247
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 6 of 8
`
`

`

`generally consistent with the particle sizing data (Figure
`12). This observation provided a hypothesis that
`the
`particle characteristics change after initial crystallization
`(1:7 ratio of ethanol/acetone) and subsequent slurry (Figure
`11). The particle characteristics obtained after heptane
`slurry were more suited for formulation development.
`In conclusion, two crystalline forms of UT-15C were
`studied. There was only 0.1 kJ/mol in enthalpy difference
`between the two crystalline forms. A solubility difference was
`observed between forms, but this solubility difference is not
`expected to influence bioavailability from a solid-dosage form.
`Form B was the more stable form, and both ethanol/acetone
`and ethanol/ethyl acetate as crystallization solvents consistently
`provided Form B of UT-15C. The use of ethanol/acetone for
`crystallization provided a denser material as compared to
`crystallization from ethanol/ethyl acetate. Finally, ethanol/
`acetone was selected as a solvent system for the final crystal-
`lization of UT-15C because of the acceptable physical properties
`of the final material. Performing a heptane slurry operation on
`crystallized UT-15C gave the desired granular and compact
`particle shape as compared to the rodlike material before the
`heptane slurry.
`
`3. Experimental Section
`X-ray powder diffraction (XRPD) analyses were performed
`by SSCI Incorporation using Shimadzu XRD-6000 X-ray
`powder diffractometer. Thermogravimetric analyses (TGA) and
`differential scanning calorimetry (DSC) were performed using
`TA Instruments 2950 thermogravimetric analyzer and TA
`Instruments 2950 differential scanning calorimeter, respectively.
`XRPD, TGA, and DSC analyses were performed by SSCI
`Incorporation. Particle size analysis and microscopy of trepro-
`stinil diethanolamine (UT-15C) were performed by Cirrus
`Pharmaceutical Inc. Melting temperatures were determined
`using a capillary tube melting point apparatus and are uncor-
`rected. The following experimental procedure was used for
`large-scale production of Form B of UT-15C.
`3.1. Experimental Procedure using Ethanol/Acetone
`Solvent System. To a suspension of UT-15 (1200 g) in
`ethanol (1000 mL) was added a solution of diethanolamine
`(352 g) in ethanol (2200 mL). The mixture was heated to
`a clear solution at 50 °C. The warm solution was filtered
`to remove any suspended, insoluble materials, and the
`clear solution was transferred to 50-L jacketed reactor. It
`was heated to 65 °C, and acetone (34 L) was slowly added
`in portions while maintaining the temperature of the
`reaction solution between 45-55 °C. At this temperature,
`the clear solution was stirred for 0.5 h. The temperature
`of the solution was decreased to 40 ( 2 °C over a period
`of 2 h. At this temperature, the reaction was seeded with
`Form B (12 g), and the solution was stirred for 2 h at 40
`( 2 °C. The temperature of the reaction was decreased
`to 35 °C over 2 h and then stirred at 35 °C overnight.
`The following day, the reaction temperature was decreased
`to 22 °C over a period of 4 h. The reaction mixture was
`stirred at this temperature overnight. The third day, the
`reaction mixture was cooled to 10 °C over a period of
`1 h in order to maximize recovery of UT-15C. At 10° C,
`
`Figure 11. Particle shape image after heptane slurry.
`
`Figure 12. Particle size distribution before heptane slurry (top)
`and after heptane slurry (bottom).
`
`and any acute exposure would be highly undesirable. The
`subsequent drug product formulation incorporated a wet granu-
`lation process, and as a result, a milling step was not required
`to control particle size after synthesis of the API.
`UT-15C obtained after the synthesis/crystallization from
`ethanol/acetone was subjected to a heptane slurry at 75-78 °C
`to obtain the desired particle habit. On the basis of small-scale
`experiments this process was scaled up to kilogram scale, and
`the nature of material obtained from the heptane slurry was
`granular and denser (Figure 11) as compared to the rod-shaped
`material before the heptane slurry (Figure 10).
`Microscopically, before heptane slurry, UT-15C ap-
`peared to be composed of fine, rodlike particles and large
`aggregates of particles; after heptane slurry it appeared
`to be composed of large agglomerates with a low level of
`fines (Figure 12). The microscopic appearances were
`
`248
`
`•
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 7 of 8
`
`

`

`UT-15C was isolated by filtration using an Aurora filter.
`The cake was washed with acetone (18 L), and the UT-
`15C was dried in the Aurora filter for 2 h under house
`vacuum. It was then transferred into trays for air-drying
`overnight. At this stage, the melting point of the free-
`flowing UT-15C (106 °C) indicated Form B was pre-
`dominantly present (g98.5%). This was also confirmed
`by XRPD data. The weight of the UT-15C was 1414 g
`(94%). This product was slurried in heptane (12 L) at
`50-55 °C for 14-16 h to change its crystal morphology.
`After the heptane slurry and drying, the melting point of
`this batch increased to 106-108 °C.
`3.2. Experimental Procedure using Ethanol/Ethyl Ac-
`etate Solvent System. A 50-L cylindrical reactor equipped
`with a heating/cooling system, a mechanical stirrer, a
`condenser, and a thermocouple was charged with a
`solution of UT-15 (1250 g, 3.20 mol) in ethyl acetate (40
`L), and a solution of diethanolamine in ethanol (337 g of
`diethanolamine was dissolved in 5.1 L of ethanol, 3.20
`mol). While stirring, the reaction mixture was heated to
`60-75 °C for 0.5-1.0 h to obtain a clear solution. The
`clear solution was cooled to 50 ( 5 °C. At this temper-
`ature, some seeds of Form B of UT-15C (∼12 g) were
`added to the clear solution. The suspension of polymorph
`B was stirred at this temperature for 1 h. The suspension
`was cooled to 20 ( 2 °C overnight (over a period of
`16-24 h). The resulting UT-15C was collected by
`filtration using an Aurora filter equipped with a filter cloth,
`and the collected solid was washed with ethyl acetate (2
`× 8 L). TheUT-15C was transferred to a HDPE/glass
`container for air-drying in a hood, followed by drying in
`a vacuum oven at 50 ( 5 °C under high vacuum for 1-2
`h. The weight of the UT-15C was 1 507 g (95%), mp
`104-105 °C. XRPD data indicated Form B was predomi-
`nantly present (g98.5%).
`
`3.3. Experimental Procedure for Heptane Slurry. A 50-L
`cylindrical reactor equipped with a heating/cooling system,
`a mechanical stirrer, a condenser, and a thermocouple was
`charged with a slurry of UT-15C (3071 g, obtained from
`the EtOH/EtOAc solvent system, mp 104-105 °C) in
`heptane (36 L). The suspension was heated to 70-75 °C
`for 16 h. The suspension was cooled to 22 ( 2 °C over
`a period of 1-2 h.UT-15C was collected by filtration
`using an Aurora filter. The cake was washed with heptane
`(15-30 L) and the material was dried in the Aurora filter
`for 1 h.UT-15C was transferred to trays for air-drying
`overnight in a hood, followed by vacuum drying at 50 °C
`under high vacuum for 1 h toobtain 3.0 kg of UT-15C
`(99%), mp 105.0-106.5 °C
`3.4. Experimental Procedure for Preparation of
`Form B Seed. To a suspension of UT-15C(8.5 g, mixture of
`Forms A and B) in MTBE (40 mL) was slowly added IPA in
`four portions (4 × 40 mL) over a period of 2 h while stirring.
`The slurry was stirred at ambient temperature (∼22-25 °C),
`and the conversion of Form A to Form B was observed by
`recording the XRPD after regular intervals (an aliquot was taken
`from the slurry, filtered, and dried for XRPD). After stirring
`for 24 h, the slurry was filtered in vacuo, dried at room
`temperature to obtain Form B (8 g, mp 106-108 °C, confirmed
`by XRPD)
`Acknowledgment
`We thank SSCI Inc. for their technical advice during
`the course of this study and providing us the results on
`XRPD, DSC, and TGA analyses. We also thank Cirrus
`Pharmaceuticals for providing us the results on particle
`size analysis.
`
`Received for review September 26, 2008.
`
`OP800239M
`
`Vol. 13, No. 2, 2009 / Organic Process Research & Development
`
`•
`
`249
`
`IPR2020-00769
`United Therapeutics EX2009
`Page 8 of 8
`
`