Synthesis of the NK1 receptor antagonist GW597599. Part 2: Development of a scalable route to a key chirally pure arylpiperazine

Article abstract of DOI:10.1021/op8002823

GW597599 1 is a novel NK-1 antagonist currently under investigation for the treatment of CNS disorders and emesis. The initial chemical development synthetic route, derived from the one used by medicinal chemistry, involved several hazardous reagents, gav

Full text of DOI:10.1021/op8002823

Organic Process Research & Development 2009, 13, 489–493
Synthesis of the NK1 Receptor Antagonist GW597599. Part 2: Development of a
Scalable Route to a Key Chirally Pure Arylpiperazine
Giuseppe Guercio,*,† Angelo Maria Manzo,† Michael Goodyear,† Sergio Bacchi,† Stefano Curti,† and Stefano Provera‡
Chemical DeVelopment and Analytical Chemistry, GlaxoSmithKline Medicine Research Centre, Via Fleming 4,
37135 Verona, Italy
Abstract:
isolable intermediate in a previous communication,1 to the
corresponding arylpiperazine 14.
GW597599 1 is a novel NK-1 antagonist currently under investiga-
tion for the treatment of CNS disorders and emesis. The initial
chemical development synthetic route, derived from the one used
by medicinal chemistry, involved several hazardous reagents, gave
low yields, and produced high levels of wastes. Through a targeted
process of research and development, application of novel tech-
niques, and extensive route scouting, a synthetic route for
GW597599 has been developed. This paper reports the optimisa-
tion work of the second stage in the chemical synthesis of
GW597599: the development of a pilot-plant-suitable process for
the manufacturing of an optically pure arylpiperazine derivative.
In particular, the new process eliminates the need to purchase and
store dangerous and expensive borane by generating it in situ and
also the need for a chlorinated solvent, thereby improving the
safety of the process and substantially reducing the overall waste.
The final yield and throughput will be significantly enhanced.
2.2. Evaluation of the Process. The chemistry involved the
reduction of the amido group of 13 followed by the isolation
of 14 as its dihydrochloride salt (Scheme 2). The first step was
the free basing of the ketopiperazine mandelate salt in a mixture
of dichloromethane and aqueous sodium hydroxide. The reduc-
tion step was achieved by using a borane solution in THF and
heating the mixture at reflux. A treatment with hydrochloric
acid in refluxing methanol decomposed the intermediate boron
complex, allowing the desired product to be isolated as a salt
upon cooling. On an 11 kg scale, the overall yield for the stage
was 63% theoretical. However the process, even if simple for
a functional group interconversion point of view, had to employ
up to three different reactors, and it produced about 110 kg of
waste for each kilogram of output. Even the need for a
chlorinated solvent seemed to be unavoidable as the free base
of 13 was difficult to recover from the aqueous layer without
using dichloromethane.
1. Introduction
3. Development of a Practical Convergent Synthesis
3.1. Synthetic Strategy: the Borane Replacement. The
development of this stage focussed on the replacement of
the expensive and unstable borane-THF complex along-
side the maximization of the yield.
Substance P (SP) is a member of the neurokinin family that
exerts its pleiotropic role by preferentially binding to the
neurokinin receptor classified as NK1. Substance P and the
NK1-receptors that mediate its activity are present in the brain
stem centres that elicit the emetic reflex. Nausea and vomiting
have been reported as the most distressing side effects of
chemotherapy, and the disruptive effects of these symptoms
on patients’ daily lives have been well documented. In light of
the need for the continued routine use of emetogenic chemo-
therapy, effective prevention of chemotherapy-induced nausea
and vomiting (CINV) is a central goal for physicians admin-
istering cancer chemotherapy. GW597599 1 is a novel NK-1
antagonist currently being developed to relieve these symptoms.
This and other NK-1 antagonists have also been investigated
for the treatment of CNS disorders such as depression. In this
paper we describe the successful efforts to define a synthetic
process for large-scale manufacturing.
Diborane gas, which can easily be evolved from borane-THF
complex, has a very low auto ignition temperature of about
38-52 °C and a wide explosive range in air (0.8-90% vol).2
The reduction of the carbonyl moiety was performed in
refluxing THF at 65-67 °C. In the case of a vessel at
atmospheric pressure and open to a scrubber, the diborane gas
would be able to easily escape at temperatures above 38 °C,
possibly causing an explosion. In addition the borane-THF
complex must be transported and stored below 8 °C due to its
instability. This special handling measure increases the costs
and the difficulties of using such a reagent. Finally, it is an
expensive reagent considering that it is routinely sold at a
concentration of 1 M solution in THF (26£/L on 5000 L scale)
and produces a large amount of waste.3
2. Initial Medicinal Chemistry Route
The synthetic route followed to synthesise the first few grams
of GW597599 is shown in Scheme 1.
2.1. Synthetic Objective. The aim of this paper is to
describe the reduction of chiral derivative 13, identified as an
Several alternative reduction conditions were tried, but they
mostly failed or gave very little conversion. In particular,
LiAlH4, sodium metal, and Red-Al gave only partial reduction
to the desired product and degradation, while NaBH4 alone,
(1) Guercio, G.; Bacchi, S.; Goodyear, M.; Carangio, A.; Tinazzi, F.; Curti,
S. Org. Process Res. DeV. 2008, 12, 1188.
* To whom correspondence should be addressed. E-mail: giuseppe.2.guercio@
gsk.com.
(2) Urben, P. G., Ed. Bretherick’s Handbook of ReactiVe Chemical
Hazards, 6th ed.; Butterworth-Heinemann: Boston, 1999; p 1937.
(3) Burkhardt, E. R.; Corella, J. A., II. U.S. Patent 6,048,985, 2000.
† Chemical Development.
‡ Analytical Chemistry.
10.1021/op8002823 CCC: $40.75  2009 American Chemical Society
Published on Web 02/24/2009
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
489

Scheme 1. Synthetic route used by the medicinal chemistry department
NaBH4 in diglyme, NaBH4/methanol, LiBH4, amine-borane
complexes, borane dimethyl sulfide, and sodium acyloxyboro-
hydride4 gave almost no reaction. The literature covers the
possibility of generating diborane in situ using NaBH4 in
combination with several Lewis acids. A plethora of possibilities
were tried, obtaining variable amounts of the desired product:
I2, H2SO4, TFA, TiCl4, SnCl4, FeCl3, AlCl3, ZnCl2, BF3 ·Et2O,
TMSCl.5 The one that gave the highest conversion was the
couple NaBH4 plus BF3 ·Et2O complex. Thus, the work was
focused in optimising the relative ratio of the two reagents, the
temperature, and the reaction concentration.
BF3 ·Et2O6 in refluxing THF were required to ensure the
reduction in acceptable yield.
Table 1. DoE performed at 55-57 °C
BF3 ·Et2O (equiv)
NaBH4 (equiv)
conv. (%) at 5 h
3.00
1.75
0.50
1.75
1.75
3.00
3.00
0.50
1.75
0.50
1.00
4.00
4.00
2.50
2.50
4.00
2.50
2.50
1.00
1.00
10.5
74.9
27.9
61.2
57.4
96.4
66.5
16.0
8.2
0.0
Scheme 2. Initial synthetic approach
A first set of trials performed by using an Endeavor7 system
allowed us to find out that both the amount of solvent (THF)
and the temperature played a key role; low dilution and high
temperature being required to get a good conversion. Then a
design of experiment (DoE, central composite design with two
central points, Table 1, Figure 1) was planned in order to
optimise the equivalents of NaBH4 and BF3 ·Et2O in combina-
tion with the temperature. The data indicates the importance of
the Lewis acid amount to drive the conversion close to
completion.
3.2. Initial Optimisation. From the very beginning of the
work it appeared clear that excesses of both NaBH4 and
(4) Thomas, J. B.; Atkinson, R. N.; Vinson, N. A.; Catanzaro, J. L.;
Perretta, C. L.; Fix, S. E.; Mascarella, S. W.; Rothman, R. B.; Xu,
H.; Dersch, C. M.; Cantrell, B. E.; Zimmerman, D. M.; Carroll, F. I.
J. Med. Chem. 2003, 46 (14), 3127.
(6) Sengupta, S.; Sahu, D. P.; Chatterjee, S. K. Indian J. Chem. 1994,
33B, 285.
(7) Endeavor, Argonaut, is a system made of eight reactor vessels with
independent temperature and pressure controls. It works on a maximum
scale of ∼5 mL, and it permits feeding a reactant in the closed vials,
monitoring the pressure trend versus time.
(5) For a general review about methods of generating diborane in situ
from sodium borohydride see: Souza, M. V. N.; Vasconcelos, T. R. A.
Appl. Organomet. Chem. 2006, 20, 798.
490
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

theoretical and chiral purity of above 99% ee. Further process
development of this step was therefore needed.
3.3. Process Improvements. The residual presence of base
and the high temperature were confirmed as the causes of both
low yield and racemisation.
A promising idea seemed to be the direct reduction of 13
combined with the possibility of performing the chemistry at
the lowest possible temperature. However, the poor solubility
of 13 in THF led us to investigate alternative solvents that
dissolved the organic salt 13 followed by addition of the solution
of this salt to a suspension of NaBH4 in THF at 0 °C. Although
1,3-dioxolane appeared suitable, it was difficult to replace it
with methanol in the next step, since 34 volumes were necessary
to remove it completely due to the poor azeotrope composition.
This problem was solved by adding a substoichiometric amount
of tetrabutylammonium bromide8 to help the dissolution of the
mandelate salt in THF at room temperature by increasing the
ionic character of the solvent. This solution was then added
dropwise to a suspension of NaBH4 in THF at 0 °C, followed
by slow addition of an excess of BF3 ·Et2O at 15 °C.
After 18-20 h at room temperature a 90% HPLC conversion
was observed. Then methanol was added at 0 °C to quench the
borane complexes and the small amount of unreacted NaBH4
and BH3.THF complex possibly present. The mixture was then
distilled down to 6 residual volumes at atmospheric pressure.
We postulate that the acidity of methanol during the lengthy
distillation allowed quenching the amine borane complexes
without the need to add acid to the mixture. The resulting
suspension was cooled to 0 °C for 2 h to precipitate all the
insoluble NaBF4 which was then filtered off.
Figure 1. Effect of NaBH4 and BF3 equivalents on conversion.
The best conditions were found to be ∼3.5 equiv of NaBH4
and ∼3 equiv of BF3 ·Et2O in 5 volumes of THF. However
some potentially explosive diborane gas might be generated at
55-57 °C. Thus, we started to investigate the possibility of
performing the step in a closed system. Another set of
experiments was designed with the aim of confirming those
data. The output of this second DoE study (central composite
design with two central points) showed that conversion above
90% could in fact be achieved in a closed vessel with 3 equiv
of NaBH4 and 2.7-3 equiv of BF3 ·Et2O at 55 °C only. After
the reaction completion and nitrogen venting, the unreacted
diborane should be decomposed to boric acid and hydrogen by
adding water.
However, the technical constraint of the designated pilot-
plant facilities dictated that we use the atmospheric pressure
procedure. In order to ensure that the atmospheric pressure
procedure could be safely carried out on pilot-plant scale, a
dedicated reaction calorimetric experiment was carried out by
passing the effluent gas through a scrubber containing an
aqueous sodium hydroxide solution. The scrubber liquors were
analysed for boron, and this suggested that a maximum level
of 2% v/v diborane should have been expected in the effluent
gas. To ensure safety of operation a nitrogen flow was applied
to the pilot-plant equipment in order to dilute diborane to below
its lower explosive limit of 0.8% v/v.
The following procedure was applied on scale: the mandelate
salt 13 was suspended in dichloromethane and washed with
aqueous sodium hydroxide to obtain the free base. After a swap
from dichloromethane to THF, the organic solution was treated
with NaBH4 and BF3 etherate, to reduce the amide moiety.
Overall this process still presented a couple of obstacles. The
first one was the partial racemisation of ketopiperazine deriva-
tive 13 due to the presence of residual sodium hydroxide in
the organic phase. The second was the reaction temperature,
high enough to both further increase the racemisation and to
increase the hazards associated with the production of diborane.
The process was scaled up twice on 80 kg scale with variable
results in respect to theoretical yields, 61.9% and 66.7%, and
chiral purity, 99.1% ee and 95.7% ee. The preliminary work
on 1 kg scale had set our expectations at a yield of 75%
The clear organic solution was finally treated with HCl 5-6
N in IPA dropwise at 0 °C to crystallise 14,9 and the mixture
was left at 0 °C for 4 h. The yield of this process was 75-80%
theoretical, and the observed ee was 99.8-100%.
3.4. Process Understanding via 19F NMR. The possibility
of having a residue of fluorinated species in the mother liquors,
and particularly fluorides, was considered prior to scaling up
the synthesis of this intermediate. The absence of fluorides was
fundamental in order to avoid possible damage to plant reactors.
Confirmation of this absence was found experimentally by using
19F NMR. All the 19F NMR experiments described in this work
were performed using a Varian INOVA 400 MHz spectrometer
(19F NMR operating frequency 376 MHz) equipped with an
ATB probe simultaneously tuned to both 1H and 19F nuclei. 1H
NMR experiments were recorded using a Varian INOVA 600
MHz.
A sample of 14 was analyzed by means of 1H NMR: only
traces of impurities and about 1% M of 2-propanol were found
in the spectrum recorded in DMSO-d6.
(8) Yadav, V. G.; Yadav, G. D.; Vyas, J. R. Chim. Oggi 2000, 18, 39.
(9) The collected cake was washed with methanol to eliminate the alcohol
derivative (see structure below) resulting from the reduction of residual
mandelic acid.
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
491

Figure 2. 19F NMR spectrum of the mother liquors coming from the synthesis of 14 recorded at 376 MHz. The four expanded plots
show the observed 19F resonances.
The mother liquors, after precipitation and filtration, were
also analyzed by means of both 1H and 19F NMR to verify the
possible presence of fluorinated species.
observed but could not be quantified with respect to the aromatic
19F resonance of 14, possibly because of incomplete dissolution.
A few drops of DMSO-d6 were necessary to achieve complete
dissolution of the sample. A new 19F NMR spectrum was then
recorded, allowing the determination of a 7.5/100 molar ratio
BF4-/14.
1
These mother liquors showed, in the H NMR spectrum
recorded with the addition of CD3OD, the presence of methanol,
2-propanol, and THF mainly, plus some residuals of 14 and
related impurities. Resonances coming from 2-propanol and
THF possibly involved in complexation with BF3 were also
observed through NMR signals, very similar to those of the
“parent” solvent but shifted in the spectrum.
The possible source of fluorine was identified as the
BF3 ·Et2O complex due to the well-known propensity of BF3
to complex with alcohols and ethers. The 19F NMR spectrum
of the BF3 ·Et2O was recorded in CD3OD. This spectrum
showed two 19F resonances: the minor one at -155.3 (broad)
due to the BF3 complex with Et2O and the major one at -158.5
(broad) due to the complex with CD3OD. These data allowed
us to be sufficiently confident that in the mother liquors the
relevant species were in fact BF4- and the complex between
BF3 and methanol. The 19F NMR additional couple of peaks at
-153.3/-153.4 (20/80 ratio) can probably be assigned to
another BF3 complex with either THF or 2-propanol.
Having confirmed the absence of fluorides to avoid possible
damage to the glass components of the pilot plant, the process
was ready for the scale-up.
3.5. Final Refinements. In the final development stage three
further refinements were considered necessary. First, to improve
the safety of the process, the BF3 ·Et2O complex was replaced
with the BF3 ·THF complex. In fact the diethyl ether could lead
to peroxide formation, and its low boiling and flash points
presented a problem, whereas the complex BF3 ·THF gave
comparable results upon use test, and it was commercially
available in bulk.
The 19F NMR spectrum showed the typical 19F signals due
to 14 plus two similar products in the region of aromatic fluorine
signals (about -112 ppm). In addition, intense peaks were
observed in the region around -150/-160 ppm. Among these,
the presence of the species BF4- was clearly identified due to
the peaks at -154.9/-154.8 ppm. Those peaks accounted for
two resonances with a ratio of 80/20 due to the natural
abundance of the two boron isotopes. This led to the observation
of distinct 19F signals for 11BF4-/10BF4-. Two other species were
also found at -153.3/-153.4 (20/80 ratio) and at -158.3 (single
broad peak). Figure 2 shows the full 19F NMR expanded plots
of the regions of interest.
None of these signals can be ascribed to inorganic F- which
was expected in the region -110/-130 ppm. Had it been
present, HF should be in an equilibrium with F- and FHF-,
with chemical shift expected in a variable range (-168/-200
ppm).10-12
FHF- a HF a F-
(1)
Second, it was also clear that the reaction mixing had the
potential to become another issue: different batches of NaBH4
with different particle size distribution gave variable conversion.
Using the granular product, the conversion was only 60% after
24 h, while the powder gave a conversion higher than 90% in
a few hours, suggesting the need to examine the solubility
effects. Finally, the same batch of NaBH4 gave different
Further studies were carried out both to confirm the possible
presence of fluorinated species in the solid 14 and to give more
information on the nature of those unknown species.
Solid 14 was dissolved in CD3OD, to record the 19F NMR
spectrum under the same conditions applied to the mother
liquors. The presence of the BF4- signal was surprisingly still
492
•
Vol. 13, No. 3, 2009 / Organic Process Research & Development

Experimental Section
(2S)-2-(4-Fluoro-2-methylphenyl)piperazine Dihydro-
chloride Salt 14. Tetrabutylammonium bromide (0.02 kg) and
13 (1 kg) were added to THF (6.5 L) at 25 °C and then heated
at 40 °C while stirring until dissolution. This solution was added
to a suspension of NaBH4 (0.38 kg) in THF (1.5 L) at 25 °C.
The suspension was stirred for 2 h at 25 °C, and then BF3 ·THF
(2.3 L) was added dropwise over about 50 min. The resulting
mixture was stirred at 35 °C for at least 18 h. Methanol (3 L)
was added over about 1 h, keeping conditions inert and the
internal temperature at 35 °C. The mixture was refluxed for
1 h, concentrated to 4.0 L at atmospheric pressure, and diluted
with 2-propanol (4 L) before being cooled to 0 °C for 3-4 h.
The slurry was filtered and the cake washed with 2-propanol
(2 × 1 L). The collected mother liquors were treated with
methanol (3 L) first, followed by HCl 5-6 N in 2-propanol
(1.4 L). The resulting suspension was kept at reflux for about
1 h, then cooled down and stirred at 0 °C for 6 h. Finally, it
was filtered, and the cake was washed with methanol (4 × L).
Both solids were dried under vacuum at about 50 °C to obtain
an overall yield of 519 g of 14 (corrected yield of 81.5%; ratio
14:enantiomer, 99.3:0.7).
Figure 3. Rushton turbine and mixing model.
Scheme 3. Final manufacturing process
conversions, depending on the reactor configuration and the
stirrer shape. An engineering assessment was carried out with
the aid of computer modelling techniques, and it suggested that
adopting a definite reactor configuration would ensure homog-
enization of NaBH4 particles by mechanical shearing, resulting
in a reproducible conversion. The prescribed measures were:
the use of a dish base vessel; the use of a high-shear impeller
like a Rushton13 turbine; the use of four baffles (Figure 3); the
need to work at relatively low vessel occupancy (liquid height
at max equal to tank diameter) and to use a high stirring speed.
All the chemical (Scheme 3) and engineering suggestions
were successfully applied in the final manufacturing route.
The new process produced a total of 23 kg of waste for each
kilogram of output (a reduction of almost 80% of the waste
relative to the original process), and delivered 46.7 kg of
intermediate 14 in high yield (81.5%) and a chiral purity greater
than 99% ee.
1H NMR (600 MHz, DMSO-d6) δ 10.3 (4 H, br s), 8.03 (1
H, dd, J ) 9.5, 5.7 Hz), 7.16 (2 H, m), 5.00 (1 H, dd, J ) 12.0,
3.1 Hz), 3.64 (1 H, m), 3.55 (3 H, m), 3.51 (1 H, dd, J ) 13.7,
3.5 Hz), 3.38 (1 H, t, J ) 12.8 Hz), 2.43 (3 H, s).
13C NMR (151 MHz, DMSO-d6) δ 162.20 (1 C, d, JCF
)
246.6 Hz), 140.22 (1 C, d, JCF ) 8.5 Hz), 129.30 (1 C, d, JCF
) 8.5 Hz), 128.22 (1 C, d, JCF ) 3.1 Hz), 117.42 (1 C, d, JCF
) 21.4 Hz), 113.29 (1 C, d, JCF ) 21.4 Hz), 51.33 (1 C, s),
44.42 (1 C, s), 41.22 (1 C, s), 38.93 (1 C, s), 19.22 (1 C, d, JCF
) 1.2 Hz).
MS (ES+) m/z 195.1 (MH+).
HPLC Phenomenex LUNA C18; mobile phase A: 0.05%
TFA/water and B: 0.05% TFA/acetonitrile; gradient: 0 min 0%
B to 30 min 95% B; flow 1 mL/min; detector UV DAD @210
nm; retention time 14 13.2 min. Purity >95.6% a/a.
HPLC Chiralcel OD (4.6 mm × 250 mm); Mobile phase
n-hexane/ethanol, 90:10 v/v; flow 1 mL/min; detector UV @265
nm; retention times 14 7.8 min, opposite enantiomer 6.7 min,
ratio 99.3:0.7.
4. Conclusion
In conclusion a simple and efficient reduction process for
the arylpiperazine of interest was developed and successfully
scaled up in pilot plant. Key breakthroughs were the definition
of a process that prevented the loss of optical purity and the
development of safer conditions when handling the diborane
species.
Acknowledgment
We thank Giulio Camurri and Luca Martini for HPLC
method development, Francesco Tinazzi for the support in the
DoE setup and interpretation, and Francois Ricard and Daniela
Ciccarone for the engineering suggestions.
(10) Emsley, J.; Gold, V.; Lowe, B. M.; Szeto, W. T. A. J. Chem. Soc.,
Dalton Trans. 1988, 1271.
(11) Hudlicky, M. J. Fluorine Chem. 1985, 28, 461.
Supporting Information Available
(12) Christe, K. O.; Wilson, W. W. J. Fluorine Chem. 1990, 46, 339.
(13) Rushton turbines are radial impellers having a higher power number
(NP) than other common impellers. The power number is a dimen-
sionless number which describes the power supplied by an impeller.
An empirical correlation relates NP to the effects of fluid viscosity.
Considering the same fluid physical properties, stirrer speed, and
impeller size, a Rushton turbine, having a higher NP, will supply more
power to the fluid and, in general, is able to supply more intense mixing
which should translate to a higher particles shear. Thus, on the basis
of theoretical knowledge, the use of a Rushton turbine was recom-
mended in the present case, requiring high intense mixing, in particular
strong radial flow, and high particles shear.
Process safety evaluation of 14 and chiral HPLC of (2S)-
2-(4-fluoro-2-methylphenyl)piperazine dihydrochloride salt 14.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Received for review November 4, 2008.
OP8002823
Vol. 13, No. 3, 2009 / Organic Process Research & Development
•
493