Process development challenges to accommodate a late-appearing stable polymorph: A case study on the polymorphism and crystallization of a fast-track drug development compound

Article abstract of DOI:10.1021/op0501287

The case of disappearing/late-appearing stable polymorphs and their impact is well-understood by scientists in the pharmaceutical industry. This paper discusses an instance where a more stable crystal form was discovered during the development of a fast-track drug candidate. Challenges in adapting to the discovery of the new crystal form during this accelerated drug development program and approaches to develop a robust crystallization process are discussed.

Full text of DOI:10.1021/op0501287

Organic Process Research & Development 2005, 9, 933−942
Process Development Challenges to Accommodate A Late-Appearing Stable
Polymorph: A Case Study on the Polymorphism and Crystallization of a
Fast-Track Drug Development Compound
Sridhar Desikan,*,† Rodney L. Parsons, Jr.,† Wayne P. Davis,‡ James E. Ward,§ Will J. Marshall,| and Pascal H. Toma
Pharmaceutical Research Institute, Bristol Myers Squibb Company, New Brunswick, New Jersey 08903, U.S.A.
Abstract:
as a well-documented case study on how the discovery of a
more stable polymorph can have a potentially disastrous
effect on the supply of an essential drug product.2,3 In this
paper, we discuss another instance where the issue of a late-
appearing stable polymorph was encountered, but with no
significant impact on the overall development of the drug.
The challenges faced by the drug development team to adapt
quickly to overcome this challenge in a fast-track drug
development scenario will be discussed. The API’s propen-
sity to form solvates with a number of solvents and
the impact of this on the identification and selection of
a crystallization system will be peripherally addressed.
Also, the use of in-process analytical tools to gain better
understanding of crystallization processes will be high-
lighted.
The case of disappearing/late-appearing stable polymorphs and
their impact is well-understood by scientists in the pharma-
ceutical industry. This paper discusses an instance where a more
stable crystal form was discovered during the development of
a fast-track drug candidate. Challenges in adapting to the
discovery of the new crystal form during this accelerated drug
development program and approaches to develop a robust
crystallization process are discussed.
Introduction
Polymorphs of a molecular material can be envisioned
as minima in the energy landscape of a single component
system. Metastable polymorphs would constitute local minima
in the energy landscape with the thermodynamically stable
form being the absolute minimum at a given temperature
and pressure. The search for absolute minimum and energy
differences between local minima of drug substances is the
goal of material and formulation scientists in the pharma-
ceutical industry. While significant efforts are made by drug
development groups to identify and characterize crystal forms
early in development, there are many instances where new
crystal forms have been discovered later in development
during process scale-up. The late emergence of thermody-
namically stable crystal forms is often explained by Ost-
wald’s law of stages which states that the least stable crystal
form is likely to crystallize first. Dunitz and Bernstein1
discuss the phenomena of disappearing polymorphs in de-
tail. They elaborate how in many instances, control of
polymorphs is related more to the control of crystallization
conditions. According to Dunitz and Bernstein,1 “...once a
polymorph has been obtained, it is always possible to obtain
it again; it is only a matter of finding the right experimental
conditions.” However, consistent, controlled manufacture of
less stable crystal forms can be quite challenging. The
polymorph issues encountered with ritonovir (Norvir) serves
Compound A is a nonnucleoside reverse transcriptase
inhibitor of human immunodeficiency virus type-1 (HIV-
1). The first crystal form of compound A was discovered
through sublimation experiments and unambiguously char-
acterized through single crystal analysis as a neat (solvent-
free) phase (designated as Form I). This neat form was
chosen for further development. The chemical process used
for the drug substance synthesis has been previously
described4,5 and is illustrated in Scheme 1. The final chemical
step of the process involved a dephenethylation conducted
in neat formic acid; after reaction workup the crude drug
substance (Compound A) was crystallized from a toluene/
heptane mixture.
(2) Letter to Health Care ProViders, Abbott Laboratories. http://www.fda.gov/
medwatch/safety/1998/norvir.htm (accessed May 17, 2005).
(3) Chemburkar, S. R. et al. Dealing with the impact of Ritonovir Polymorphs
on the Late Stages of Bulk Drug Process Development. Org. Process Res.
DeV. 2000, 4, 413-417.
(4) Magnus, N. A.; Confalone, P. N.; Storace, L.; Patel, M.; Wood, C. C.; Davis,
W. P.; Parsons, Rodney, L., Jr. General Scope of 1,4-Diastereoselective
Additions to a 2(3H)-Quinazolinone: Practical Preparation of HIV Thera-
peutics. J. Org. Chem. 2003, 68, 754-761.
(5) Magnus, N. A.; Confalone, P. N.; Storace, L. A new asymmetric 1,4-ad-
dition method: Application to the synthesis of the HIV nonnucleo-
side reverse transcriptase inhibitor. Tetrahedron Lett. 2000, 41(17), 3015-
3019.
* Corresponding author: Telephone (732) 227-6133. Fax (732) 227-3782.
E-mail: sridhar.desikan@bms.com.
† Pharmaceutical Research Institute, Bristol Myers Squibb Company.
‡ Current address: Hovione LLC, East Windsor, NJ 08520.
§ Current address: Mettler-Toledo AutoChem Inc., 7075 Samuel Morse Drive,
Columbia, MD 21046.
| Current address: DuPont R& D, Experimental Station, Wilmington, DE.
Current address: Abbott Laboratories, Abbott Park, IL 60064.
(1) Dunitz, J. D.; Bernstein, J. Disappearing Polymorphs. Acc. Chem. Res. 1995,
28, 193-200.
10.1021/op0501287 CCC: $30.25 © 2005 American Chemical Society
Published on Web 11/02/2005
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
933

Figure 1. Differential scanning calorimetry thermogram of Form I drug substance.
Scheme 1. Synthesis scheme for the preparation of Compound A
Crystallization from toluene/heptane affords a mixed
toluene/heptane solvate which undergoes a thermal conver-
sion to compound A with a variable level of amorphous
phase material. To produce drug substance with the requisite
purity and prepare material of a uniform crystalline phase,
the crude drug substance was recrystallized from methanol
to provide a crystalline, stoichiometric methanol solvate. The
isolated methanol solvate was designated as Form II. Form
I was obtained by a thermal conversion of Form II by drying
the wet cake of Form II at 90 °C for 4 h followed by further
drying at 120 °C for 3 h. The transformation of Form II to
I was monitored using differential scanning calorimetry
(DSC). During the development of this drug candidate as
many as 29 batches of Form I material were produced by
this thermal conversion process at scales up to 150 kg. Figure
1 shows the DSC thermogram of the material obtained from
the thermal conversion process where a single melting
endotherm with melt onset at 181 °C corresponding to Form
I was observed. During the drying of the 30th batch, a
significant change in melting behavior was observed in the
process control samples analyzed by DSC. Material taken
from the dryer showed a new lower melting endotherm with
melt onset of 174 °C (Figure 2a, b). At the end of the drying
process, a material with single melting endotherm with melt
onset of 174 °C was isolated (Figure 3). This material was
designated as Form III. Once Form III had been generated,
Form I material could no longer be produced via the thermal
desolvation process except by extended heating at melt
temperatures of Form III. Powder X-ray pattern of the
material collected during batch 30 during the drying process
shows a mixture of two forms, thus indicating a transition
from one to the other (Figure 4). Scanning electron micro-
graphs of Forms I and III show significant difference in
morphology of the forms, indicative of differences between
934
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development

Figure 2. Differential scanning calorimetry thermogram of the thermal conversion process. New lower melting endotherm was
observed. (Top) Scanning rate 10 °C/min. (Bottom) Scanning rate 2 °C/min.
crystal forms (Figure 5). It should be noted here that Form
I was obtained from desolvation of the methanol solvate as
clearly evident from the surface heterogeneities shown in
Figure 5a. However, morphology of Form III crystals show
fused nonsolvated crystals which may indicate a solvent-
mediated transformation from methanol solvate to neat Form
III. Precise mechanism of transformation to Form III and
drying conditions which led to this transformation is not
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
935

Figure 3. Differential scanning calorimetry thermogram of Form III drug substance.
Table 1. Thermodynamic interrelationships between Form I
and III
known. This underscores the risk of isolating crystals from
suboptimal, poorly controlled processes such as thermal
conversion in the dryer. Once more stable Form III crystals
were obtained from the thermal conversion process in the
dryer and Form I could no longer be isolated, development
work focused on efforts to identify a procedure to isolate
Form III directly from a crystallization process.
Discussion on the Crystal Forms of Compound A.
Compound A is a nonhygroscopic neutral molecule that has
very low aqueous solubility (less than 10 µg/mL) and high
permeability (Caco2 permeability > 150 nm/s). It can be
considered as a Class II compound according to the Biop-
harmaceutics Classification System (BCS) with dissolution-
limited bioavailability.6 It is soluble in most organic solvents
with a tendency to form solvates in many of the common
solvents. During an extended polymorphism screening study
several solvates including methanolate, ethanolate, and
propanolate (from 2-propanol) were discovered. To date only
two anhydrous crystal forms have been discovered (Forms I
and III). Form I was discovered initially during develop-
ment, and the discovery of Form III has been described
in the previous section. After the discovery of Form III,
its single-crystal structure was obtained and confirmed to
be a neat form as well. Table 1 summarizes the thermody-
namic properties of Forms I and III. The main diagonal
represents melting behavior of forms. The upper half matrix
describes the thermodynamic relationships, based on the
observations outlined in the lower half matrix. Based on
Burger’s heat of melting rule,7 Forms I and III are enantio-
*Burgers Heat of Melting rule: This rule states that if the polymorph with
the higher melting point exhibits lower heat of fusion, the system is enantiotropic,
otherwise, it is monotropic.7
tropically related with N-3 (Form III) more thermodynami-
cally stable than N-1 (Form I) at temperatures below 120
°C. The transition temperature is inferred to be between 120
and 174 °C (Tm of lower melting Form III). At ambient
conditions, far from the transition temperature, Form III is
likely to be significantly more stable than Form I. This may
explain the disappearance of Form I once Form III was
isolated.
936
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development

Figure 4. Powder X-ray diffraction pattern of material obtained during the drying process (Top) Form III. (Middle) Form I.
(Bottom) Material obtained during the drying process. A mixture of Form I and III was observed by PXRD analysis.
Table 2. Structural parameters for Forms I and III
Form III. Furthermore, Form III contains relatively large
channels (∼3.5 Å in diameter) along the c-axis, which might
explain the large difference in densities between the two
forms (Figure 8). Form I exhibits a higher density, contra-
dicting the density rule7 that states, “If one modification of
a molecular crystal has a lower density than the other, it
may be assumed to be less stable at absolute zero”.
Exceptions to this density rule have been reported in the
literature in about 10% of the molecular crystals exhibiting
polymorphism.8 Burger and Ramberger explain that in
instances where directed forces such as hydrogen bonds are
present it may not be possible to conclude a correlation
between density and potential energy.7 The lattice energy
calculations using the universal force field confirm that Form
I has a more stable crystal packing and hence a higher
density.
parameters
Form I
Form III
crystal system
space group
monoclinic
P21
tetragonal
P4122
crystal habit
a
b
c
â
V
acicular
16.810 Å
9.080 Å
18.541 Å
92.52°
rhomboid
11.593 Å
11.593 Å
67.066 Å
90°
2827.3 Å3
8
9013 Å3
24
Z
Z′a
4
3
density
1.479 g/cm3
418.4 kcal/mol
1.391 g/cm3
449.6 kcal/mol
lattice energyb
a Z′ ) number of molecules per asymmetric unit. b Universal Force Field.
The crystal structures for Forms I and III have been
determined using single-crystal X-ray crystallography. The
structures are very different in terms of molecular packing.
Table 2 summarizes the structural parameters. The hydrogen
bonding in Form I is made up of two sets of dimers along
the a-axis. The dimers are held together through a π-π
stacking along the c-axis and N-H- - -Cl-C interactions
along the b-axis (Figure 6). Form III is also held together
by dimers but through an infinite chain along the c- and
b-axes, whereas van der Waals interactions are along the
a-axis (Figure 7). Table 3 summarizes the intermolecular
distances in each form. It is clear from this table that Form
I contains more and stronger intermolecular interactions than
Direct Crystallization Process of Form III. Since
Compound A exhibited tendencies to form solvates in many
organic solvents, the search for an acceptable solvent for
direct crystallization of anhydrous form III resulted in few
leads. After careful screening, n-propanol was chosen as a
(6) Amidon, G. L.; Lennernas, H.; Shah, V. P.; Crison, J. R. A theoretical basis
for a biopharmaceutic drug classification: the correlation of in vitro drug
product dissolution and in vivo bioavailability. Pharm. Res. 1995, 12, 413-
420.
(7) Burger A.; Ramberger, R. On the Polymorphism of Pharmaceutical and
Other Molecular Crystals. I. Theory of Thermodynamic Rules. Mikrochim.
Acta [Wein] 1979, II, 259-271.
(8) Burger A.; Ramberger, R. On the Polymorphism of Pharmaceutical and
Other Molecular Crystals. II. Applicability of Thermodynamic Rules.
Mikrochim. Acta [Wein] 1979, II, 273-316.
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
937

Figure 5. Scanning electron micrographs of Form I and III material. (a) Form I. (b) Form III.
Figure 6. Form I unit cell.
solvent for crystallization process development. Due to the
high solubility of Compound A in n-propanol, use of an
antisolvent (water) was required to achieve the required yield
target and to provide a slurry with acceptable flow properties.
938
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development

Table 3. Intermolecular distances
heterogeneous atom distances (Å)
interaction
type
Form I
Form III
H-bonding
“Dimers”
2.827
2.742
2.834
2.791
2.857
2.857
2.874
2.898
2.826
2.771
4.084
-Cl- - -H-N-
π-π staking
3.384
3.347
3.409
3.412
3.47
none
3.56
Figure 8. Form III channel along c-axis.
of the experimental batches, a new crystal form with
needlelike morphology was observed. Analysis by PXRD
indicated the presence of a new phase different from that of
Form III. Further characterization of the material confirmed
the material to be n-propanol solvate. Formation of solvate
despite substantial amount of Form III seeds present was
investigated as this is not desirable for isolating neat Form
III crystals. Efforts were also undertaken to study the phase
behavior of this solvate in n-propanol/water solvent systems
to develop crystallization protocols for isolating neat Form
III directly from crystallization (next section). This was
solved using in-process monitoring using in situ optical
microscopy (Particle Vision Measurement probe, Lasentec
Inc.). With the in situ microcopy, it was clearly evident that
during the water addition, a quasi-emulsion phase was formed
with the hydrophobic layer enriched with drug substance (n-
propanol phase) and the aqueous phase containing small
amounts of the drug substance. Unlike stable emulsions, the
quasi-emulsion phase observed is not thermodynamically
stable with easy separation of hydrophobic/hydrophilic
Figure 7. Form III unit cell.
The n-propanol/water crystallization system was developed
and scaled up to 1-L scale to demonstrate process feasibility.
The crystallization protocol involved the use of Form III
seeds (with a loading of up to ∼5% by weight) to ensure
growth on existing crystals of Form III and to avoid
formation of undesired crystal forms. However, during one
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
939

Figure 9. In situ optical micrograph during the crystallization process. A quasi-emulsion phase was observed.
phases. The presence of the highly hydrophobic drug
substance in high concentrations (30-50% w/v) resulted in
the phase separation of the otherwise fully miscible n-pro-
panol-water mixtures resulting in the formation of the quasi-
emulsion phase described above (Figure 9).
Phase Behavior of Compound A in n-Propanol/Water
System. To investigate the formation of the quasi-emulsion
phase and to understand the formation of an n-propanol
solvate, detailed experiments were carried out in 20-mL
laboratory-scale equipment. An excess of Form III drug
substance was slurried in n-propanol/water mixtures at
varying concentrations and temperatures. Evidence for phase
separation of the organic and aqueous phase was visually
observed and noted. The corresponding solid phase was
analyzed by PXRD to determine the subsequent material
crystalline phase. Figure 10 shows the phase behavior
Figure 10. Phase diagram of the crystal forms in the crystal-
information of the drug substance in the solvent system
chosen for crystallization. It is clear that at high n-propanol
concentrations and at lower temperatures, n-propanol solvate
is the stable form. At predominantly n-propanol-rich condi-
tions (>1:1 n-propanol/water ratio), at temperatures above
35 °C, a clear liquid phase corresponding to complete
dissolution of drug substance was observed. As the water
composition is increased, the quasi-emulsion region was
observed. The composition range over which the quasi-
lization solvent system.
emulsion phase was observed was found to increase at higher
temperatures. This could be attributed to the fact that at
higher temperatures more drug substance is soluble in
n-propanol rendering the solution more hydrophobic. At
sufficiently high water compositions, the n-propanol solvate
was observed to desolvate to the anhydrous Form III.
940
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development

Figure 11. In situ optical micrograph during the crystallization process. Needlelike crystals (n-propanol solvate) were observed.
Development of Direct Crystallization Process. On the
basis of the phase behavior information, several crystalliza-
tion protocols were devised and evaluated for direct isolation
of the desired Form III. As shown in Figure 10, in Protocol
1, the temperature of seeding was set at 45 °C and 5% seeds
of neat Form III crystals were added as a slurry in a
n-propanol/water mixture. While some liquid-phase separa-
tion and quasi-emulsion formation was still observed, the
effect was not as pronounced as in earlier cases. In Protocol
2, the seeding temperature was further lowered to ap-
proximately 30 °C. In this case, while no significant emulsion
formation was observed, n-propanol solvate crystals were
observed during crystallization as evidenced by in-process
microscopy (Figure 11). As noted previously the n-propanol
solvate crystals were converted to Form III upon further
water addition. In both Protocols 1 and 2, the crystallization
slurries were held at 30 °C for 8 h before filtration to ensure
complete conversion of any undesired n-propanol solvate.
Protocol 2 was successfully scaled up to 300-kg scale (1000
L) in the manufacturing facility. As an alternative to both
Protocols 1 and 2, a new reverse-addition process (Protocol
3) was developed. In this process, drug substance dissolved
in n -propanol was slowly added to the aqueous phase
containing seeds of Form III. This modified scheme incor-
porates all the elements of a robust process where Form III
is the stable solid phase at all times during the crystallization
process. As noted in Figure 12, some agglomerate crystals
of Form III were observed, but this issue was easily
overcome by use of intense mixing during crystallization and
by sonication of the slurry.
Effect of Crystal Form on Dosage Form Development.
Compound A was formulated as 50-mg strength tablets using
Form I. Analysis of tablets stored under accelerated storage
conditions (40 °C/75% RH) indicated complete conversion
to Form III in dosage form. The formulation was modified
to accommodate the new crystalline form and was found to
be equivalent to Form I through a relative bioavailability
study in humans. While polymorphs of drug substances may
result in differences in bioavailability, in this instance there
was no difference in the in vivo performance.
Summary
Knowledge of stability of the crystal forms and their
interrelationships is critical for developing robust crystal-
lization processes. While it may be tempting to use expedi-
tious processes such as thermal conversion during initial drug
development, they may have much greater overall negative
impact on development timelines. Use of on-line analytical
tools such as in-process microscopy is critical for developing
knowledge on the crystallization behavior. In this paper, the
case of a late-appearing, thermodynamically stable poly-
morph during a fast-paced drug development program and
challenges encountered are discussed. We hope to have
provided another convincing example for pharmaceutical
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
941

Figure 12. In situ optical micrograph during direct crystallization process (Protocol 3).
scientists on the need for early and thorough investigation
of crystal forms.
are thanked for their assistance in physical characterization.
Mr. Benjamin Smith and Mr. Rich Becker of ASI Mettler
Toledo (Lasentec Inc.) are thanked for their technical
assistance with in-process microscopy.
Acknowledgment
We acknowledge the contributions from the project team
of DuPont Pharmaceuticals Company where the work was
conducted. Mr. Darin Norwood and Ms. Barbara Stephens
Received for review July 25, 2005.
OP0501287
942
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development