Development of a Practical and Efficient Synthesis of CP-945, 598-01, a CBi Antagonist for the Treatment of Obesity

Article abstract of DOI:10.1021/op800255j

Development of an efficient bond-forming sequence and optimization of reaction conditions are described for the synthesis of CP-945, 598-01 (1-HC1), a CB, antagonist in clinical studies for the treatment of obesity. Reordering of the bond-forming sequence

Full text of DOI:10.1021/op800255j

Organic Process Research & Development 2009, 13, 186–197
Development of a Practical and Efficient Synthesis of CP-945,598-01, a CB1
Antagonist for the Treatment of Obesity
John A. Ragan,*,‡ Dennis E. Bourassa,‡ Jon Blunt,¶ Darragh Breen,0 Frank R. Busch,‡ Eric M. Cordi,‡ David B. Damon,§
Nga Do,‡ Alanya Engtrakul, Denis Lynch,0 Ruth E. McDermott,‡ Joseph A. Mongillo, Maria M. O’Sullivan,0 Peter R. Rose,§
and Brian C. Vanderplas§
Chemical Research and DeVelopment, Materials Science, and Analytical Research and DeVelopment, Pfizer Global Research
and DeVelopment, Eastern Point Road, Groton, Connecticut 06335, U.S.A., Supply Chain, Pfizer Ltd., Ramsgate Road,
Sandwich, Kent CT139NJ, U.K., and Pfizer Global Manufacturing, Ringaskiddy API Plant, County Cork, Ireland
Abstract:
Development of an efficient bond-forming sequence and optimiza-
tion of reaction conditions are described for the synthesis of CP-
945,598-01 (1 ·HCl), a CB1 antagonist in clinical studies for the
treatment of obesity. Reordering of the bond-forming sequence
provided a more efficient synthesis and avoided the use of
phosphorous oxychloride. A telescoped reaction sequence (4 f 9)
was developed to avoid a problematic isolation. Product isolations
were developed so as to provide efficient throughput by minimizing
solvent volumes and avoiding slow filtrations.
Figure 1. Structures of CB1 antagonists.
1. Introduction
The endocannabinoid system (ECS), and specifically the CB1
receptor, plays a pivotal role in energy homeostasis.1 As such,
stimulation of the ECS promotes food intake and energy storage
and may be chronically overactive in obese subjects.2,3 In
contrast, blockade of the CB1 receptor decreases food intake
and increases energy expenditure, leading to a reduction in body
weight.4-7
CB1 receptor antagonists (Figure 1) may provide effective
therapy options for the management of metabolic disorders, such
as obesity. Several CB1 receptor inverse agonists/antagonists
Figure 2. Structure of CP-945,598-01 (1 ·HCl).
are in clinical development including the diarylpyrazole
SR141716A8 (rimonabant, 2) and the acyclic amide MK-03649
(taranabant, 3).
† Submitted in memory of Dr. Chris Schmid, a valued friend and colleague
whose many contributions to the field of process chemistry continue to benefit
this growing community of scientists.
1-(8-(2-Chlorophenyl)-9-(4-chlorophenyl)-9H-purin-6-yl)-4-
(ethylamino)-piperidine-4-carboxamide hydrochloride (CP-
945,598-01, 1·HCl) (Figure 2) was identified by Pfizer Dis-
covery chemists as a promising CB1 antagonist for the treatment
of obesity10,11 and nominated for development in 2002. This
paper describes the original synthesis used to prepare CP-
945,598-01, the identification of a more efficient bond-forming
sequence, and optimization of the reaction conditions for this
synthesis to support scale-up from laboratory (<50 g) to pilot
plant (20-50 kg) to commercial pilot scale (>100 kg)
campaigns.
* Author to whom correspondence may be addressed. E-mail:
john.a.ragan@pfizer.com.
‡ Chemical Research and Development, Pfizer Global Research & Development.
§ Materials Science, Pfizer Global Research & Development.
AnalyticalResearchandDevelopment,PfizerGlobalResearch&Development.
¶ Supply Chain, Pfizer Ltd.
0 Pfizer Global Manufacturing, Ringaskiddy API Plant.
(1) Pagotto, U.; Marsicano, G.; Cota, D.; Lutz, B.; Pasquali, R Endocr.
ReV. 2006, 27, 73–100.
(2) Matias, I.; Gonthier, M.-P.; Orlando, P.; Martiadis, V.; De Petrocellis,
L.; Cervino, C.; Petrosino, S.; Hoareau, L.; Festy, F.; Pasquali, R.;
Roche, R.; Maj, M.; Pagotto, U.; Monteleone, P.; Di Marzo, V. J. Clin.
Endocrinol. Metab. 2006, 91, 3171–3180.
(3) Engeli, S.; Boehnke, J.; Feldpausch, M.; Gorzelniak, K.; Janke, J.;
Batkai, S.; Pacher, P.; Harvey-White, J.; Luft, F. C.; Sharma, A. M.;
Jordan, J. Diabetes 2005, 54, 2838–2843.
(8) Rinaldi-Carmona, M.; Barth, F.; Heaulme, M.; Shire, D.; Calandra,
B.; Congy, C.; Martinez, S.; Maruani, J.; Neliat, G.; Caput, D.; Ferrara,
P.; Soubrie, P.; Breliere, J. C.; Lefur, G. FEBS Lett. 1994, 350, 240–
244.
(4) Jbilo, O.; Ravinet-Trillou, C.; Arnone, M.; Buisson, I.; Bribes, E.;
Peleraux, A.; Penarier, G.; Soubrie, P.; Le Fur, G.; Galiegue, S.;
Casellas, P. FASEB J. 2005, 19, 1567–1569.
(9) Lin, L. S.; Lanza, T. J., Jr.; Jewell, J. P.; Liu, P.; Shah, S. K.; Qi, H.;
Tong, X.; Wang, J.; Xu, S. S.; Fong, T. M.; Shen, C.-P.; Lao, J.; Xiao,
J. C.; Shearman, L. P.; Stribling, D. S.; Rosko, K.; Strack, A.; Marsh,
D. J.; Feng, Y.; Kumar, S.; Samuel, K.; Yin, W.; Van der Ploeg,
L. H. T.; Goulet, M. T.; Hagmann, W. K. J. Med. Chem. 2006, 49,
7584–7587.
(5) Thornton-Jones, Z. D.; Kennett, G. A.; Benwell, K. R.; Revell, D. F.;
Misra, A.; Sellwood, D. M.; Vickers, S. P.; Clifton, P. G. Pharmacol.,
Biochem. BehaV. 2006, 84, 353–359.
(6) Vickers, S. P.; Webster, L. J.; Wyatt, A.; Dourish, C. T.; Kennett,
G. A. Psychopharmacology (Berlin, Ger.) 2003, 167, 103–111.
(7) Hildebrandt, A. L.; Kelly-Sullivan, D. M.; Black, S. C Eur. J. Phar-
macol. 2003, 462, 125–132.
(10) Griffith, D. A. U.S. Patent 2004157839, 2004.
(11) Griffith, D. A. Manuscript submitted to J. Med. Chem.
186
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development
10.1021/op800255j CCC: $40.75
2009 American Chemical Society
Published on Web 12/22/2008

Scheme 1. Synthesis used in first cGMP campaign
2. Discussion
(8) was effected with triethylamine in refluxing isopropanol;
this coupling required higher temperatures than the analogous
coupling in the earlier synthesis (7 to 1 in Scheme 1) but
occurred cleanly to generate a new intermediate (9). The
cyclization of 9 to 1 was mediated by H2SO4 in methanol, and
the product was isolated by direct filtration of the sulfuric acid
salt. This material was then converted to free base (Na2CO3 in
aq acetone) and then to the HCl salt (HCl in isopropanol) for
comparison to reference standards. Several attempts were made
to perform the cyclization with HCl in place of H2SO4, in the
hope that the HCl salt could be isolated directly. These efforts
were unsuccessful. We initially attributed this failure to the
higher water content of concd HCl relative to concd H2SO4.
However, subsequent experiments showed that dilute aq H2SO4
solutions led to successful cyclizations; therefore, it appears to
be related to acid strength, or other, unidentified, factors.
The first scale-up of the new bond-forming sequence was
performed in the pilot plant in Groton, Connecticut, and
delivered 19 kg of API. The chemistry utilized is shown in
Scheme 3 and included several modifications from the first
laboratory-scale experiments. The cyclization of 9 to 1 was
effected by concd H2SO4 in DMSO at 90-100 °C. The product
was isolated as the free base by an extractive workup involving
neutralization with aq NaOH and product extraction into
2-methyltetrahydrofuran. Solvent displacement with toluene by
vacuum distillation led to precipitation of 1 as the crystalline
free base. The salt formation was performed in two stages:
addition of concd HCl to the free base in THF generated the
HCl salt as a mixture of three less stable polymorphs. A reslurry
in aq isopropanol then converted this mixture to Form A, the
desired, thermodynamically stable polymorph.
The synthesis utilized for the later Discovery lots and first
cGMP (current Good Manufacturing Practice) campaign is
shown in Scheme 1. The cGMP campaign proceeded remark-
ably well for a compound at this early stage of development,
providing 13.8 kg of active pharmaceutical ingredient (API).
All but one of the steps proceeded in >90% yield; the sole
exception was the chlorination to generate chloropyrimidine 7.
In addition to the modest yield (62%), the use of phosphorous
oxychloride presented handling and potential safety challenges
that we felt might be avoided by a reordering of the bond-
forming steps.
In the third step in Scheme 1, the sulfuric acid-mediated
dehydration of chloropyrimidine 5 generated 8-(2-chlorophe-
nyl)-9-(4-chlorophenyl)-9H-purin-6-ol (6), in which the chlo-
ropyrimidine had undergone hydrolysis to the pyrimidinone.
Other, milder conditions for this cyclization led to mixtures of
the chloropyrimidine and pyrimidinone, so for the first bulk
campaign it was deemed expedient to use conditions that
completely hydrolyzed the chloropyrimidine, thus simplifying
product isolation and analysis. This required a subsequent
chlorination, which was effected with phosphorous oxychloride.
The piperidine side chain was then installed by coupling of
piperidine 8 with the regenerated chloropyrimidine (7) to deliver
1. We reasoned that if the piperidine was installed prior to the
acid-mediated cyclization, this could obviate the need for the
rechlorination step.
The first variant of this bond-forming sequence was suc-
cessful on laboratory scale, as shown in Scheme 2. The two-
step synthesis of chloropyrimidine 5 was unchanged from the
previous route. Coupling of 5 with the piperidine side chain
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
187

Scheme 2. Alternative bond-forming sequence
Scheme 3. First scale-up of the new bond-forming sequence
The next series of bulk campaigns were performed in the
pilot plant in Sandwich, Kent, UK. Over the course of several
campaigns in 2006, over 150 kg of API were delivered in this
facility. During the course of these campaigns and associated
laboratory development work, optimizations were made to each
step in the synthesis. The process improvements that arose from
the scale-up knowledge gained in Sandwich were critical in
improving our process from the first scale-up experience
(Scheme 3) to the ultimately successful transfer of the process
to our commercial manufacturing group in Ireland (Scheme 4).
These improvements are summarized below for each step.
2.1. Preparation of 5-Amino-6-chloropyrimidine Hydro-
chloride·HCl. This step was modeled closely after the original
literature report.12 We observed formation of only one significant
impurity in this step, the bis-adduct 10, which arises from addition
of a second molecule of 4-chloroaniline to the dichloropyrimidine
core. Although this impurity was well-controlled by downstream
crystallizations, minimizing its formation would increase reaction
efficiency. Experiments in which the HCl stoichiometry was varied
showed a negative linear correlation between the levels of this
impurity and the equivalents of HCl, ranging approximately 10-
fold from 2 to 3% levels with 0.4 equiv down to 0.2-0.3% levels
with 2 equiv. Based on these studies, the HCl stoichiometry was
increased from 0.4 to 2.0 equiv.
Optimization of the product isolation led to a switch in the
final rinse solvent from hexanes (or heptanes) to ethyl acetate.
(12) Greenberg, S. M.; Ross, L. O.; Robins, R. K. J. Org. Chem. 1959,
24, 1314–1317.
188
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Scheme 4. Optimized API synthesis utilized in commercial pilot-plant campaign
This provided shorter drying times, as the increased water
miscibility lead to more efficient displacement of residual water
from the filter cake. We confirmed that the potential N-acylated
impurity 11 (Figure 3) was not present in batches prepared using
the ethyl acetate rinse.
in methanol solution. We found no advantage to using free base
(4) in the acylation step. In fact, the hydrochloride salt was
considerably more soluble in DMAC or NMP, and was thus
preferred as a substrate for the next reaction.
Although all bulk campaigns run to date have utilized the
free base 4-chloroaniline, we have also demonstrated that the
aniline HCl salt can be substituted with no adverse impact on
product yield or purity. It should be noted that the HCl salt of
4-chloroaniline is a higher melting and much more easily
handled solid.
Figure 3. Structures of known or potential impurities in
formation of 4.
We and earlier chemists on the project had initially assigned
the step 1 product (4) as the free base. This was based on the
literature precedent,12 in which the free base is isolated following
a recrystallization from methanol. However, subsequent analysis
during the optimization studies described above showed that we
were in fact isolating the mono-HCl salt. The discrepancy with
the literature arises from our isolation method, which is to directly
filter the product from the aq HCl-ethanol reaction medium. If this
material is recrystallized from 50% aq methanol (per the literature
report, 70 mL/g), the free base is obtained. The same conversion
to free base was also realized by reslurrying the HCl salt in 20
mL/g of 50% aq methanol. These data show that the initially
isolated product is in fact the mono-HCl salt of 4, and that
recrystallizing from aq methanol converts the salt to the free base;
this likely reflects the relatively weak basicity of the aminopyri-
midine, as no external base is added during this conversion. This
caused some concern as to the solid state stability of 4·HCl.
However, a sample challenged in a vacuum oven (25-30 mmHg)
at 90 °C for one week showed no loss of HCl, indicating that in
the solid state, the salt is stable and robust, in contrast to its behavior
Preparation of acylated amino-chloropyrimidine 5. The
acylation of the 5-aminopyrimidine (4·HCl) with 2-chloroben-
zoyl chloride was performed in a polar, aprotic solvent, initially
dimethylacetamide (DMAC), and eventually N-methylpyrroli-
dinone (NMP). The reaction performed equally well in either
solvent, and the only significant driver for the switch to NMP
was to minimize the total number of solvents used in the
synthesis, since NMP was utilized in a later step (cyclization
of 9 to 1), thus simplifying the testing for residual solvents.
Kinetic analysis of this reaction showed it to be a standard,
second-order nucleophilic acylation (first-order in both ami-
nopyrimidine 4 and acid chloride). Because of the kinetics, it
was difficult to drive the acylation to completion without using
an excess of one of the reagents. Given the low cost of the
acid chloride and its simple purge following aqueous quench
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
189

Table 1. Predicted yield (%) of acylation product 5 as a
piperidine is commercially available (albeit more expensive than
the acid chloride),13 it was used in slight excess (1.2-1.3 equiv).
function of temperature and water
% H2O in starting material (4·HCl)
T (°C)
0.5
1
2
4
5
15
25
35
99
99
99
99
92
95
97
98
83
85
87
89
65
70
74
77
and product crystallization, we chose to use this reagent in slight
excess (1.2-1.3 equiv). The kinetic experiments also examined
hydrolysis of the acid chloride, a potential competing side
reaction if water is present. This is also a second-order reaction,
first order in both acid chloride and water. At relatively modest
water content relative to 4·HCl (e.g., 0.5-4.0 wt %, or 8-65
mol % relative to 4), the observed rate is comparable to the
rate of the desired acylation. Table 1 shows the predicted
reaction yield as a function of water content and temperature
(0.5 to 4.0 wt % H2O). On the basis of these data, the hydrolysis
was controlled by specifying the water content of 4·HCl at not
more than (NMT) 1.0%. These results also suggest that there
is no advantage to running the acylation below room temperature.
Another aspect of this reaction observed on scale-up to >20
kg was the slow filtration of the acylated product (5) following
the aqueous quench. One particularly onerous campaign (76
kg scale) required 14 days for filtration, rinsing, and drying of
this material. Experiments were pursued to identify other
crystalline forms of the product that might crystallize from other
solvent systems and offer improved filtrations. With no clear
leads identified from these experiments, we chose to pursue a
telescoped process in which 5 would not be isolated, but instead
taken directly into the piperidine coupling. This would allow
us to bypass the problematic filtration of 5 altogether, and was
ultimately chosen as the best solution to this problem (Vide
infra).
2.4. Development of a Telescoped Process: Single-Pot
Conversion of 4 ·HCl to 9. Due to the long filtration and drying
times encountered with isolation of amide 5, we sought to
develop a telescoped process in which this intermediate was
not isolated. Two obvious challenges were apparent: first,
identification of a suitable solvent for both reactions, as the
n-propanol solvent used in the piperidine coupling was clearly
not suitable for the acylation. A solvent screen was performed
for the piperidine coupling (5 + 8 f 9), and NMP was
identified as a suitable solvent for this reaction, providing
comparable yields, purities, and reaction times as n-propanol.
Other solvents examined included THF, 2-methyl-THF, aceto-
nitrile and toluene; none of these were suitable due to the limited
solubility of 4·HCl in these solvents. The second challenge was
to quench the excess acid chloride present following the
acylation, as this would react with piperidine 8 if present during
its addition. This was easily achieved by addition of a small
volume of water to the NMP solution upon complete acylation.
Selection of a suitable base turned out to be more challeng-
ing. The amine bases used in the previous piperidine coupling
(e.g., triethylamine or i-Pr2NEt) were of limited utility in the
telescoped process, as the piperidine coupling stalled after
75-80% conversion, even with use of excess piperidine. This
is presumably due to formation of the hydrochloride salt of the
piperidine, which is of comparable basicity to these amines (note
that by the end of the piperidine coupling, 3 equiv of HCl are
present in the reaction mixture). Some success was realized with
stronger, inorganic bases such as NaOH or Na2CO3 in
aqueous-organic solvent mixtures such as tert-amyl alcohol,
NMP, or THF. However, reproducibility on scales >5 g and
filtration rates were problematic. The best results were obtained
with organic amines that were more basic than triethylamine.
1,4-Diazabicyclo[2.2.2]octane (DABCO) and tetramethylguani-
dine (TMG, Me2NC(dNH)NMe2) were both effective in this
regard, providing >98% conversions. TMG was chosen for
further optimization. The optimized process involved dissolution
of 4·HCl in NMP (5 L/kg) and addition of 1.2 equiv of
2-chlorobenzoyl chloride (2-CBC). Upon complete reaction
(typically 3-4 h), water was added to quench the excess acid
chloride (0.75 L/kg, which is a 10-fold excess relative to the
0.2 equiv of excess acid chloride). This addition was somewhat
2.3. Preparation of Piperidine-Substituted Pyrimidine 9.
A screen of several solvents and bases found that this
piperidine-chloropyrimidine coupling was conveniently ef-
fected with organic amine bases in alcohol solvents. Triethy-
lamine and isopropanol were used initially. We later changed
conditions to i-Pr2EtN (Hu¨nig’s base) in n-propanol, which
offered a modest increase in reaction rate. Upon development
of the telescoped process, we returned to NMP so as to be
compatible with the acylation (Vide infra).
The kinetics of this reaction were also studied and were
found to be second order overall, first order in both chloropy-
rimidine (5) and piperidine (8). As with the acylation, this meant
that the reaction slowed dramatically as it approached comple-
tion, and use of an excess of one reactant was required to drive
the reaction to completion in reasonable times. Because the
(13) Griffith, D. A. WO Patent 2004037823, 2004.
190
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

exothermic, and was done at a rate to keep the reaction
temperature below 25 °C. Kinetic studies showed that the acid
chloride hydrolysis was very fast under these conditions; the
solution would be held for 60-90 min to ensure complete
hydrolysis. TMG (3.5 equiv) was then added, followed by
piperidine 8. The resulting solution was then heated to 90-100
°C for 4-6 h.
and addition rates led to a process in which the crude reaction
solution was diluted with a 75:25 (v/v) THF/water solution at
ambient temperature, followed by slow addition of water to
induce crystallization. This procedure worked well on labora-
tory-scale (see Experimental Section), but upon a pilot scale-
up in the kilo laboratory we encountered a very slow filtration
(>24 h on 2 kg scale). We also noted a difference in the
crystallization induction during this run: in our laboratory pilots,
crystallization typically initiated about one-third of the way
through the water addition, whereas in the kilo laboratory
campaign, crystallization did not occur until complete water
addition, cool down, and seeding (by mechanical scratching of
an aliquot). These observations were consistent with the use of
equipment which was not seeded (i.e., the reactor was scrupu-
lously clean and had not previously been used for this
chemistry), whereas in our laboratory pilots we were in
glassware and hoods that had seen this chemistry many times.
It appeared that we had produced a supersaturated solution of
product in the kilo laboratory campaign, and by the time seed
crystals were introduced, a rapid and uncontrolled crystallization
resulted.
Solubility measurements provided further clarity around what
had occurred. Figure 4 shows the thermodynamic solubility
(right-hand, solid line) and kinetic solubility (left-hand, dotted
line) at 30% water composition. These data were generated by
slowly heating a defined concentration of 9 (Form D) in THF/
water and recording the temperature at which dissolution was
complete; this is represented by the thermodynamic solubility
curve. The solution was then slowly cooled, and the temperature
at which crystallization occurred was noted as the kinetic
solubility (these experiments were performed in laboratories and
equipment that had previously been used to run this process,
so fall into the “seeded” category relative to the dichotomy
discussed above). The red point indicates the region in which
seed addition should lead to a controlled crystallization and good
filtration rates.
In contrast, Figure 5 shows the situation in the Sandwich
kilo laboratory. At the point of complete addition of water we
were at 42% water composition and 22 °C. At this point, our
mechanical induction of crystallization (by scratching the
solution aliquot) was well to the left of the metastable zone
that lies between thermodynamic and kinetic solubilities. Thus,
the solution was significantly oversaturated, and when crystal-
lization was initiated, it occurred in a rapid and uncontrolled
fashion.
The difference in filtration rates was also reflected in
photomicrographic analysis of the slow-filtering solids from the
kilo laboratory campaign versus a representative, fast-filtering
laboratory sample (Figure 6).
On the basis of this analysis, we decided to modify our
process to incorporate a seeding protocol, such that we could
ensure that crystallization would occur while the solution was
in the metastable zone. Four solid-state forms of 9 had been
characterized: Form A (anhydrous), Form B (n-propanol solvate,
not relevant in the current process), Form C (dihydrate), and
Form D (THF solvate). It was determined that Forms A and C
were soluble in the 30% aq THF system, and thus were not
suitable as seed crystals. Form D was less soluble, and could
Although this telescoped process worked extremely well in
terms of effecting the desired bond formations, we were now
faced with a complex solution containing the desired product
as well as numerous salts (e.g., TMG-hydrochloride, TMG-2-
chlorobenzoate, and some piperidine (8)-hydrochloride). Our
first solution to this was an extractive workup, in which the
reaction solution was diluted with THF and water. At this point
a homogeneous solution was observed. Addition of toluene led
to phase separation, with much of the NMP residing in the
lower, aqueous phase. The product-rich toluene phase was
separated, and a second portion of toluene was added to the
organic phase, inducing a second separation of residual water
and NMP. Finally, the organic phase (consisting primarily of
product 9, some of the excess piperidine 8, toluene, and THF)
was distilled to remove the THF. During this distillation the
product began to crystallize from solution. Upon cooling, the
product could be isolated as an easily filtered white solid.
Subsequent analysis showed that significant levels (15-20%)
of piperidine 8 were also present in this material (due to its
lack of chromophores, this impurity could not be detected by
HPLC, but NMR and GC/MS were able to quantify its levels).
We found that a simple reslurry in 10 L/kg of isopropanol at
reflux was effective at removing this impurity, and also offered
significant purges for several other process-related impurities
(see Table 2).
This telescoped process was incorporated into a campaign
conducted in the Sandwich pilot plant, and offered a dramatic
improvement in processing times. In contrast to the worst
filtration encountered on 76 kg scale (14 days for isolation of
the acylation product), the telescoped process provided a two
day processing time for both the acylation and piperidine
coupling on 80 kg scale, a time savings of 12 days.
While the telescoped process offered considerable improve-
ment in throughput, it became apparent during planning
meetings for process optimization and design of experiment
(DoE) work that we had an intimidating number of process
parameters to consider during the reaction workup (e.g., addition
volumes of water, THF, toluene addition #1, toluene addition
#2, rate of heating during distillation, etc.). Having such a large
number of parameters suggested that we could face challenges
in technology transfer to our commercial manufacturing site,
and could have an intrinsically nonrobust process. Additionally,
the phase separation was quite difficult to observe during the
extractive workup, and a large volume of toluene (38 L/kg)
was required. We knew that a more attractive product isolation
method would involve addition of an antisolvent to the reaction
solution such that the product crystallized directly and could
be isolated without requiring the multiple operations in the
extractive workup.
Several antisolvents were screened, and we ultimately
focused on water/THF mixtures. Optimization of the volumes
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
191

Table 2. Impurity summary and control strategy
be formed from Forms A and/or C by reslurrying in THF. The
material from the Sandwich kilo laboratory campaign (ca. 2
kg) was used to generate Form D seed crystals for the
commercial pilot campaign in Ireland.
As reflected in Figures 4 and 5, solubility ranges varied with
both temperature and %H2O composition. Through laboratory
experiments and DoE work, we settled on a water content of
30% and a temperature of 40-50 °C (target 45 °C) as our target
metastable range in which seed crystals could be introduced
without redissolving, and well to the right of the metastable
zone limit (i.e., the point at which an uncontrolled crystallization
could occur).
This modified, seeded process was incorporated during
the first run in our commercial manufacturing pilot plant
on 75 kg scale. Upon completion of the telescoped reaction
sequence, 3:1 THF/water was added, the solution was
heated to 45 °C, and the appropriate volume of water was
added to bring the solution to 30%. Form D seed crystals
(0.5 wt %) were then added, followed by gradual cooling
to 5 °C over 3 h. Using this protocol, reasonable filtration
and rinse times were realized, with the caveat that cleaning
of the filter-dryer between batches was necessary to
prevent blinding of the filter (i.e., a “heel effect” was
observed).
192
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Figure 4. Kinetic vs thermodynamic solubilities of 9 (Form D) in 30% aq THF.
Figure 5. Kinetic vs thermodynamic solubilities of 9 (Form D) in 42% aq THF.
toxicology studies had qualified it at levels of NMT 0.5% in
final API. Nevertheless, the modest reduction in hydrolysis
product in NMP relative to DMSO (from 1 to 2% to <0.2%)
offered an improvement in product yield. It also raised the
interesting possibility that NMP is acting as a “dummy amide
hydrolysis substrate” for the equivalent of water generated in
the cyclization, and is in effect serving as a mild dehydrating
agent. The acid-mediated hydrolysis of NMP is known (com-
plete hydrolysis is realized by refluxing NMP in concd HCl
for 8 h, and product is isolated as the hydrochloride salt of 3-N-
methylaminobutyric acid).14 Although we did not observe this
ring-opened NMP hydrolysis product, it is possible that upon
neutralization and product isolation this material would cyclize
to regenerate NMP, or would simply remain dissolved in the
basic aqueous mother liquors.
2.5. Preparation of CP-945,598 (1). The first two cam-
paigns using the reordered bond-forming sequence (Scheme 2)
effected this cyclization with sulfuric acid in hot DMSO.
Subsequent studies of a wider range of acids and solvents
determined that methanesulfonic acid (MSA) in NMP provided
a faster reaction with lower levels of hydrolysis impurity 12
(Figure 7). This impurity was not of tremendous concern, as it
purges relatively well in the isolation of 1 (both as free base
and again in the salt formation). Additionally, it is an active
metabolite formed postdosing in human plasma, and our
(14) McElvain, S. M.; Vozza, J. F. J. Am. Chem. Soc. 1949, 71, 896–900.
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
193

Figure 6. Photomicrographs of slow (left) and fast (right) filtering solids (both images are at 50× magnification).
a two-stage salt formation: addition of concd HCl to the free
base in THF generated the HCl salt as a mixture of three less
stable polymorphs. A reslurry in aq isopropanol converted this
mixture to Form A, the desired, thermodynamically stable
polymorph.
Although reliable, this two-stage salt formation was less
preferred than a single-pot method, which would improve
throughput and decrease solvent waste streams. We found that
salt formation in aq isopropanol reliably delivered Form A in
a slurry (to solution)-to-slurry conversion. The free base was
combined with isopropanol (30 L/kg) and water (2 L/kg) and
heated to 60 °C, at which point a slurry was observed.
Hydrochloric acid (12 N, 1.3 equiv) was then added, at which
point dissolution occurred. After 15-30 min, the HCl salt began
to crystallize from solution, which was then cooled gradually
to 20 °C over 1-2 h, and the product was collected by filtration.
It is standard cGMP practice to perform a final filtration of
the API (or a near precursor) to remove any fibers or particulate
matter that may have been introduced in earlier operations. We
refer to this as a “speck-free filtration.” In the salt formation
described above, the speck-free filtration was performed during
a recrystallization of the free base (1) from THF and toluene.
The free base was dissolved in THF and water, passed through
a microfilter, and then displaced with toluene to precipitate the
anhydrous free base. This material was then handled under
speck-free conditions prior to being used in the isopropanol salt
formation. It bears noting that all solvents introduced after the
speck-free filtration must themselves be speck-free filtered prior
to use.
Figure 7. Structure of API hydrolysis product.
In the H2SO4/DMSO reaction, product was isolated by
basification with aq NaOH and extraction of free base 1 into
Me-THF. Concentration of the organic phase and displacement
with toluene brought about crystallization of 1 as the free base
in satisfactory yields (e.g., 78% on 50 kg scale). Although
effective, this extractive isolation procedure suffered from the
requirement for large volumes of organic solvents (up to 50
L/kg), reducing the efficiency and environmental sustainability
of the process. We addressed this concern when we switched
to the MSA-NMP conditions through identification of a direct
crystallization protocol in which the reaction mixture was diluted
with isopropanol (5 L/kg) and neutralized with aq NaOH (10
volumes, 3.5 equiv). This induced crystallization of 1 as its free
base, and reduced the solvent usage from 50 to 16 L/kg. The
isolated yield for this process was typically 90-95%.
During tech transfer discussions with our colleagues in
Pfizer’s commercial manufacturing pilot plant in Ringaskiddy,
Ireland, they expressed a strong desire to move the speck-free
filtration to a point during the final salt formation, thus avoiding
the need to handle a solid intermediate under speck-free
conditions. This change was also driven by the fact that the
free base recrystallization was no longer needed as a purity
control point, as the Step 3R isopropanol reslurry was now our
primary control gate for impurities. Thus, we developed a third
salt formation procedure in which the free base hydrate (the
product of Step 4 prior to the THF/toluene recrystallization)
was dissolved in aq THF and heated to effect dissolution. Speck-
2.6. Preparation of CP-945,598-01 (1 ·HCl). Several pro-
tocols were developed to deliver the hydrochloride salt as the
desired polymorph (Form A); these procedures changed over
time, reflecting the scale of API manufacture, equipment train
utilized, and stage of development. The early campaigns utilized
194
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Table 3. Commercial pilot plant campaign in Ringaskiddy,
and atmospheric pressure chemical ionization (APCI). Reactions
were monitored by HPLC analysis using an Agilent 1100 series
HPLC equipped with a Zorbax SB-C18 column (4.6 mm × 50
mm), gradient elution from 95:5 to 20:80 (0.5% HClO4 in
water)/(acetonitrile) over 11 min, hold for 1 min, then gradient
from 20:80 to 95:5 over 5 min; flow rate ) 1.0 mL/min, 40
°C, 210 nm detection. Retention times: 4 (6.7 min), 5 (8.9 min),
9 (6.3 min), 1 (7.1 min).
Ireland
step number (product)
yield (%)
1 (4·HCl)
2/3 (9)
86
80
89
94
95
55
3R (9)
4 (1)
5 (1·HCl)
overall yield
Reagents were purchased from commercial suppliers and
used as received. 4-Chloroaniline, 2-chlorobenzoyl chloride,
TMG, methanesulfonic acid, and N-methylpyrrolidinone were
purchased from Aldrich. 5-Amino-4,6-dichloropyrimidine was
purchased from Fluka. 4-Ethylamino-4-carboxamidopiperidine
(8)15,16 was purchased from Fontarome Chemicals, Milwaukeee,
Wisconsin.
free filtration was done at this stage, followed by a solvent
displacement with isopropanol to a defined specification of
residual THF and water (determined by NMR and Karl Fischer
titration, respectively). The appropriate volume of water was
then reintroduced, followed by concd HCl. Seed crystals were
also introduced to ensure formation of the desired polymorph.
This procedure represents the current salt formation and is
included in the Experimental Section.
2.7. Impurity Control. Detection, identification, and control
of impurities in the API are of primary importance during the
development of any drug candidate. Table 2 summarizes several
key impurities that have been identified during the development
process. It is interesting to note that most of the impurities
shown are formed at some point in the synthesis, with the
exception of the chloro and bis-chloro isomers introduced with
the 2-chlorobenzoyl chloride. Because a rather modest purge
of these impurities was observed in the isopropanol reslurry
(Step 3R), our control strategy requires controlling the isomer
levels at point of purchase through well-defined purchase
specifications.
Step 1: Preparation of 6-Chloro-N -(4-chlorophenyl)-
4
4,5-pyrimidinediamine Hydrochloride (4 ·HCl) from 4-Chlo-
roaniline. A three-neck, 12 L round-bottom flask equipped
with an overhead stirrer, nitrogen inlet, and reflux
condenser was charged with 5-amino-4,6-dichloropyrimi-
dine (1.0 kg, 6.1 mol), 4-chloroaniline (930 g, 7.3 mol,
1.2 equiv) and 2B ethanol (5.8 L) (2B ethanol is anhydrous
ethanol denatured from toluene). A portion of 3 N
hydrochloric acid (4.0 L, 12 mol, 2.0 equiv) was added,
and the slurry was heated to reflux for 36 h. The reaction
mixture was cooled, and the solids were collected by
filtration, rinsing with 3.75 L of ethyl acetate. After
vacuum drying overnight at 40 °C, the product was
obtained as a white-to-tan solid, mp 192-196 °C dec (1.5
kg, 84% yield).
IR (thin film): cm-1 3316, 3224, 2632, 1650, 1604, 1586,
1558, 1491, 1231; 1H (DMSO-d6): δ 9.29 (s, 1H), 8.2-7.9 (br
s, 2H), 7.88 (s, 1H), 7.80 (d, J ) 9, 2H), 7.35 (d, J ) 9, 2H).
13C NMR (DMSO-d6): δ 149.2, 144.7, 139.4, 138.1, 129.0,
127.0, 126.0, 122.8. MS (EI): 255 (100). Anal. Calcd for
C10H9Cl3N4: C, 41.19; H, 3.11; N, 19.22; Cl, 36.48. Found: C,
41.35; H, 3.00; N, 19.23; Cl, 36.34.
3. Conclusions
Although the current bond-forming sequence was defined
quite early in the development process, numerous changes to
the reaction conditions and isolation strategies occurred during
the course of the project, as described in this paper. The
identification of key issues that require experimental redesign
was very much an interactive process involving continuous
dialogue among process chemists, analytical chemists, material
scientists, supply chain personnel, and Pfizer’s commercial
manufacturing group. Each of these groups had their own
perspective on the project and the associated bulk campaigns,
and the optimal solution for one group may present significant
technical challenges for another group. Maintaining an open
dialogue was critical for arriving at a process that best met the
needs of all the groups involved in the project. This optimization
strategy has also led to a final API process that is robust and
consistently delivers a high-quality API.
Preparation of 6-Chloro-N -(4-chlorophenyl)-4,5-pyri-
4
midinediamine Hydrochloride (4·HCl) from 4-Chloroa-
niline Hydrochloride. A single-neck, 25-mL flask equipped
with a reflux condenser and nitrogen inlet was charged
with 4,6-dichloropyrimidine (2.02 g, 12.3 mmol), 4-chlo-
roaniline hydrochloride (2.45 g, 14.9 mmol, 1.2 equiv),
and 2B ethanol (11.7 mL). A portion of 3 N hydrochloric
acid (8.1 mL, 24 mmol, 2.0 equiv) was added, and the
reaction mixture was heated to reflux for 24 h. The slurry
was cooled, filtered, and rinsed with water. After drying
in a vacuum oven for 4 h, the product was obtained as a
white-to-tan solid (3.15 g, 88% yield). Physical and
spectral data matched that of the material prepared by the
aniline free base method (above).
Our largest bulk campaign to date was completed in Pfizer’s
commercial manufacturing pilot plant in Ringaskiddy, Ireland.
The outcome of this campaign was quite successful, as
summarized in Table 3. A total of 371 kg of API was delivered
in four batches, utilizing the route shown in Scheme 4.
Experimental Section
All reactions were run in standard air-dried glassware with
magnetic stirring under a static atmosphere of nitrogen unless
otherwise noted. Mass spectral data were obtained on a
Micromass ZMD mass spectrometer with flow injection analysis
(15) Dow, R. L. U.S. Patent 2004110453, 2004.
(16) Ragan, J. A. WO Patent 2006043175, 2006.
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
195

Step 3R: Reslurry Purification of 1-[5-(2-Chlorobenzoy-
lamino)-6-(4-chlorophenylamino)-4-pyrimidinyl]-4-et hy-
lamino-4-piperidine Carboxamide (9). A four-neck, 1-L
round-bottom flask equipped with an overhead stirrer, reflux
condenser with nitrogen inlet, and a temperature probe was
charged with 9 (35.33 g) and isopropanol (353 mL, 10 mL/g).
The slurry was heated in an oil bath to an internal temperature
of 80 °C and held at that temperature for 22 h, after which it
was then cooled to ambient temperature and stirred for 2 h.
The solids were collected by filtration, rinsing with 200 mL of
isopropanol. After drying in a vacuum oven at 25-30 mm and
46 °C, the product was obtained as a white powder, mp
218-221 °C (Form A, anhydrous) (29.61 g, 84%).
The following data are for Form D, the THF solvate of 9,
which is obtained by refluxing in 7:3 THF/water for 3-6 h:
1H (DMSO-d6): δ 9.70 (s, 1H), 8.17 (s, 1H), 7.99 (s, 1H),
7.67 (dd, J ) 7, 2, 2H), 7.59-7.57 (m, 2H), 7.54-7.47 (m,
2H), 7.32 (dd, J ) 7, 2, 2H), 7.26 (br s, 1H), 6.96 (s, 1H),
3.76-3.72 (m, 2H), 3.47 (br t, J ) 10, 2H), 2.33 (t, J ) 7,
2H), 2.01-1.98 (m, 1H), 1.87-1.82 (m, 2H), 1.56-1.52 (m,
2H), 0.99 (t, J ) 7, 3H). 13C NMR (DMSO-d6): δ 178.5, 165.9,
160.8, 158.0, 155.3, 139.6, 136.4, 132.2, 131.0, 130.7, 129.7,
129.1, 128.0, 126.4, 122.5, 100.5, 59.8, 43.4, 37.5, 33.0, 16.4.
MS (CI): 528 (100). Anal. Calcd for C25H27Cl2N7O2 ·C4H8O:
C, 58.00; H, 5.87; N, 16.33. Found: C, 57.58; H, 5.72; N, 16.72.
Step 2/3. Preparation of 1-[5-(2-Chlorobenzoylamino)-
6-(4-chlorophenylamino)-4-pyrimidinyl]-4-ethylamino-4-pi-
peridine Carboxamide (9). A four-neck, 300-mL round-bottom
flask equipped with an overhead stirrer, addition funnel with
rubber septum, temperature probe, and a reflux condenser with
a nitrogen inlet was charged with NMP (100 mL) and 6-chloro-
N4-(4-chlorophenyl)-4,5-pyrimidinediamine hydrochloride (4·
HCl) (20.0 g, 68.6 mmol). 2-Chlorobenzoyl chloride (10.4 mL,
14.4 g, 82.1 mmol, 1.20 equiv) was charged to the addition
funnel. At ambient temperature, the acid chloride was added
over 60 min, during which time the internal temperature
remained at 20-25 °C. A 5 mL portion of NMP was used to
rinse the addition funnel. The reaction mixture was stirred for
4 h after the addition, at which point HPLC analysis of an
aliquot showed complete conversion to 4-chloro-5-(2-chlo-
robenzamido)-6-(4-chlorophenylamino)pyrimdine (5). Water
(15 mL) was charged to the addition funnel, and added slowly
to the reaction mixture such that the temperature remained <25
°C (addition required ∼5 min). The reaction mixture was stirred
for an additional 75 min and then treated with TMG (30.2 mL,
27.8 g, 241 mmol, 3.5 equiv), which was added from the
addition funnel over 30 min. 4-Ethylamino-4-carboxamidopi-
peridine (8) (15.3 g, 89 mmol, 1.3 equiv) was added in a single
portion, and the mixture was heated in an oil bath to an internal
temperature of 95 °C for 5 h. Analysis of an aliquot by HPLC
showed >98% conversion to 1-[5-(2-chlorobenzoylamino)-6-
(4-chlorophenylamino)-4-pyrimidinyl]-4-ethylamino-4-piperi-
dine carboxamide (9). The reaction mixture was cooled to
ambient temperature. In a separate flask, a solution of THF (220
mL) and water (73 mL) was prepared.
Step 4. Preparation of 1-[8-(2-Chlorophenyl)-9-(4-chlo-
rophenyl)-9H-purin-6-yl]-4-ethylamino-4-piperidine Car-
boxamide (1). A 50-L, glass-lined reactor was charged with
1-[5-(2-chlorobenzoylamino)-6-(4-chlorophenylamino)-4-pyri-
midinyl]-4-ethylamino-4-piperidine carboxamide (9) (1.80 kg,
3.14 mol) and NMP (4.98 L). The resulting solution was stirred
at 25 °C for 20 min. Methanesulfonic acid (906 g, 9.43 mol,
3.0 equiv) was then added in three equal portions of 302 g each
over approximately 15 min. The reaction mixture was heated
to 100 °C and held at that temperature for 4 h. The reaction
mixture was then cooled to 35 °C over 45 min and sampled
for HPLC analysis (acceptance criteria NMT 1.0% starting
material, result was 0.36%). The reaction mixture was stirred
at 25 °C for 2 h while the analysis was conducted. 2-Propanol
(9 L) was then added over 25 min, followed by NaOH (377 g,
9.42 mol, 3.0 equiv) dissolved in 18 L of water. The caustic
solution was added over 35 min. The pH of the solution was
checked to ensure complete acid neutralization (pH was 8.5).
The resulting slurry (solids precipitated during the base addition)
was stirred for 2-4 h, then filtered on a 14-in. Nutsche filter
fitted with a poly cloth. The solids were washed with 7:3 (v/v)
isopropanol/H2O (18 L) and pulled dry, then transferred to a
vacuum oven (45 °C, 25-30 mm) to yield the product as a
white to off-white solid, mp 141-143 °C (Form C, monohy-
drate) (1.4 kg, 84%).
Approximately half of this solution was added to the reaction
mixture, which was then transferred to a 1 L Erlenmeyer flask
equipped with a magnetic stir bar, rinsing with the remainder
of the THF/water solution. The diluted reaction mixture was
then treated with an additional 257 mL of water, added dropwise
from an addition funnel. The resulting slurry was stirred for
4 h. The solids were collected by filtration, rinsing the flask
and washing the filter cake with 1:1 THF/water (100 mL). After
air-drying for 4 h, crude product was obtained as an off-white
solid (37.8 g, 104%). This material was taken directly into the
isopropanol reslurry (next procedure).
Data for 4-chloro-5-(2-chlorobenzamido)-6-(4-chlorophe-
nylamino)pyrimdine (intermediate 5, isolated in a different
1
experiment): MS (CI): 393 (100); H (DMSO-d6): δ 10.0 (s,
IR (thin film) cm-1 3475, 3286, 2976, 1678, 1495, 1334,
1265, 1091; 1H (DMSO-d6): δ 8.23 (s, 1H), 7.70 (dd, J ) 8, 2,
1H), 7.51-7.41 (m, 4H), 7.33-7.30 (m, 2H), 7.00 (br s, 1H),
1H), 9.11 (s, 1H), 8.40 (s, 1H), 7.93 (dd, J ) 7, 2, 1H),
7.66-7.40 (m, 7H).
196
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

5.0-3.5 (br m, 4H), 2.39-2.36 (m, 2H), 2.11-2.09 (m, 1H),
1.89-1.84 (m, 2H), 1.65-1.62 (m, 2H), 1.02 (t, J ) 7, 3H).
cooled to 70 °C, and a final portion of isopropanol (5.73 L)
was added. In a 2 L Erlenmeyer flask, 3 N HCl was prepared
by addition of water (859 mL) and concd HCl (152 mL).
Approximately 20% (226 mL) of this HCl solution was added
to the reaction vessel over 15-20 min. The reaction mixture
was further cooled to 60 °C, and a sample was pulled for KF
and residual THF analysis (1H analysis showed a 172:1 (0.6%)
molar ratio of isopropanol to THF; KF analysis showed 0.13%
water). Seed material (1·HCl, Form A) (7 g, 1 wt %) was then
added, followed by the remaining 3 N HCl (approximately 785
mL) over 15-20 min. The solution was stirred at 60 °C for
4 h, then cooled to 25 °C over 1.5-2 h. Solids were collected
by filtration, rinsing with isopropanol (1.79 L). After drying
for 16 h in a vacuum oven, the product was obtained as a white
solid, mp 275-276 °C dec (676 g, 88%).
13C
NMR (DMSO-d6): δ 178.4, 153.8, 153.5, 152.5, 146.0,
133.8, 133.72, 133.67, 133.6, 132.7, 130.1, 129.9, 129.7, 129.6,
128.0, 119.4, 60.0, 37.6, 33.0, 26.1, 16.4. MS (CI): 510 (100).
Anal. Calcd for C25H25Cl2N7O·H2O: C, 56.82; H, 5.15; N,
18.55. Found: C, 56.86; H, 5.21; N, 18.36.
IR (thin film): cm-1 3332, 2949, 1691, 1578, 1562, 1493,
1
1332, 1291, 1259, 1090; H (DMSO-d6): δ 9.47 (br s, 2H),
8.30 (s, 1H), 8.11 (s, 1H), 8.00 (s, 1H), 7.71 (dd, J ) 7, 1, 1H),
7.53-7.43 (m, 4H), 7.34-7.30 (m, 2H), 5.2-3.5 (br m, 4H),
2.87-2.86 (m, 2H), 2.47-2.43 (m, 2H), 2.03-1.98 (m, 2H),
1.25 (t, J ) 7, 3H). 13C NMR (DMSO-d6): δ 170.1, 153.7, 153.5,
152.6, 146.5, 133.9, 133.8, 133.5, 132.8, 130.2, 129.8, 129.7,
129.6, 128.1, 119.5, 112.5, 63.9, 38.5, 30.6, 12.3. MS (CI): 510
(100). Anal. Calcd for C25H25Cl2N7O·HCl (Form A, anhy-
drous): C, 54.91; H, 4.79; N, 17.93; Cl, 19.45. Found: C, 54.77;
H, 4.66; N, 17.83; Cl, 19.54.
Step 5. Preparation of 1-[8-(2-Chlorophenyl)-9-(4-chlo-
rophenyl)-9H-purin-6-yl]-4-ethylamino-4-piperidine Car-
boxamide Hydrochloride (1·HCl). To a clean, dry, and
nitrogen-purged 100-L glass-lined reactor was charged THF
(6.73 L), water (358 mL), and 1-[8-(2-chlorophenyl)-9-(4-
chlorophenyl)-9H-purin-6-yl]-4-ethylamino-4-piperidine car-
boxamide (1) free base monohydrate (716 g, 1.35 mol). The
solution was warmed to 35-45 °C until homogeneous and was
then transferred to a clean, dry, and nitrogen-purged glass-lined
50 L reactor through an inline 0.5 µm Pall filter, rinsing with
an additional portion of THF (1.43 L). Isopropanol (7.16 L)
was added, and the mixture was heated to 80-90 °C to distill
off THF to a pot volume of approximately 7 L. The reaction
mixture was then cooled to 70 °C and an additional 7.16 L of
isopropanol was added. The distillation process was repeated.
A third portion of 7.16 L of isopropanol was added, and the
distillation was repeated a third time. The reaction mixture was
Acknowledgment
We gratefully acknowledge Jason Mustakis and Joe Rainville
for initial reagent/solvent screens for steps 3 and 4, and Charlie
Santa Maria, Melissa King, Ciaran Byrne, Julia Cronin, and
Dave Griffith for productive discussions and guidance through-
out the course of this project.
Received for review October 8, 2008.
OP800255J
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
197