Sequential acetic acid-sodium chloride treatment to control salt stoichiometry of a hydrochloride salt

Article abstract of DOI:10.1021/op2003375

The sequential treatment of an intermediate in the synthesis of GSK159797 with a protic acid and sodium chloride was developed to control the stoichiometry of the hydrochloric acid salt of the API. This offered significant advantages over addition of hydrochloric acid, thus avoiding decomposition of an acid-sensitive formamide group. A range of acids were investigated to determine the optimum pK_a range in which to use the acid. The optimum process, using acetic acid, was performed on >100 kg input scale.

Full text of DOI:10.1021/op2003375

Technical Note
pubs.acs.org/OPRD
Sequential Acetic Acid−Sodium Chloride Treatment to Control Salt
Stoichiometry of a Hydrochloride Salt
Darren M. Caine, Ian L. Paternoster, Simon Sedehizadeh, and Peter D. P. Shapland*
Product Development, GlaxoSmithKline, Medicines Research Centre, Gunnels Wood Road, Stevenage SG1 2NY, United Kingdom.
*
S Supporting Information
ABSTRACT: The sequential treatment of an intermediate in the synthesis of GSK159797 with a protic acid and sodium
chloride was developed to control the stoichiometry of the hydrochloric acid salt of the API. This offered significant advantages
over addition of hydrochloric acid, thus avoiding decomposition of an acid-sensitive formamide group. A range of acids were
investigated to determine the optimum pK range in which to use the acid. The optimum process, using acetic acid, was
a
performed on >100 kg input scale.
INTRODUCTION
The choice of the salt for an active pharmaceutical ingredient
API) can be a critical decision for a medicine. The salt plays a
■
(
1
critical role in the biological effect of the medicine since it
determines the solubility in biologically relevant fluids. It also
contributes to the ability to isolate and purify the API and lends
to the stability profile of the parent molecule. Depending upon
2
6
the salt, the API may exhibit multiple polymorphic forms, and
Figure 1. GSK159797 1; structure and calculated pK
acids.
a
of conjugate
the choice of salt can lead to a simplification of the
development process through minimization of polymorphic
forms.
was unstable in the presence of more than one equivalent of
hydrochloric acid. As a consequence, the process to make the
bis-hydrochloride was difficult to render under control, and a
decision was taken to target the mono-hydrochloride. However,
the instability of 1 in the presence of more than one equivalent
of hydrochloric acid was an unwanted complication.
Herein, we report a conceptually novel approach to the
generation of the mono-hydrochloride salt of GSK159797
whereby the hydrochloric acid is not added directly or as a
buffered source of hydrochloric acid but is generated from a
protic acid and sodium chloride washing of an intermediate in
the synthetic sequence. The suitable choice of protic acid allows
us to generate the mono-hydrochloric acid salt without risking
generation of the bis-hydrochloride and also minimizing
decomposition of the formamide group.
Hydrochloric acid salts are the most common for basic
3
pharmaceutical molecules. Hydrochloride salts are most
typically formed through the addition of the requisite amount
of hydrochloric acid to the free base of the API molecule. This
can be carried out under a range of conditions to suit the
maximization of yield, purity, and operational simplicity. On a
few occasions, hydrochloric acid salts have been generated by
the treatment of bases with buffered sources of hydrochloric
4
acid. This is most typically ammonium chloride, but any
protonated ammonium chloride may be capable of transfer of
the hydrochloric acid to the API, where good correlation of
basicities occurs.
Where the stoichiometry of a salt can be variable due to
multiple acidic or basic sites in the API molecule, it is necessary
to control the ratio closely such that a mixture of salts is not
likely to be produced. Where a molecule displays instability in
the presence of extremes of acid or base, great care is required
in the salt-forming step, and conditions must be controlled to
limit decomposition while maintaining the correct stoichiom-
etry of desired salt.
RESULTS AND DISCUSSION
■
The final step of the synthesis of GSK159797 is a hydro-
genolytic debenzylation of benzyl ether 2. Compound 2, in turn
is synthesized by a highly telescoped stage that starts with the
coupling of amine 3 and alkyl bromide 4 (Scheme 1). The
coupling provides intermediate silyl ether 5 which is desilylated
to provide compound 2. At the start of our development
program, the entire process to obtain 1 existed as a single
telescoped set of reactions with no isolations of intermediates
by crystallization. We decided to target the isolation of
compound 2 as the opportunity to develop our novel approach
GSK159797 (Figure 1, 1) is a long acting β-agonist which
was in early development for the treatment of COPD and
5
asthma. After extensive screening, two salt forms were
identified which were considered for further development.
These were the mono-hydrochloride and the bis-hydrochloride.
The formation of the bis-hydrochloride was complicated since
(
a) the weakly basic aniline site required more than two
equivalents of hydrochloric acid to enable robust and consistent
stoichiometric control of the salt, and (b) the formamide group
Received: November 24, 2011
Published: February 1, 2012
©
2012 American Chemical Society
518
dx.doi.org/10.1021/op2003375 | Org. Process Res. Dev. 2012, 16, 518−523

Organic Process Research & Development
Technical Note
Scheme 1. Synthetic sequence to prepare GSK159797
to the generation of a hydrochloride salt. This was driven by the
desire to effect a purification of 2 after a number of steps of
chemistry. In particular, a high level of a deformylated
compound (typically >5% peak area by HPLC) was present
in isolated 2. It was thought that the hydrochloride salt would
be tolerated in the debenzylation conditions to provide 1
without complication. The free base of 2 is not crystalline, and
it was felt that the generation of the hydrochloride salt would
enable isolation and purification from the numerous impurities
that are generated during the complex telescoped chemistry.
Similar to 1, compound 2 contains an acid-sensitive
formamide group; thus, we were well aware of the need to
control the exposure to acid during any salt-forming step at the
end of the telescoped sequence to 2.
butanone solution was distilled to remove any dissolved water,
and the dried solution was seeded with the material obtained
from the small-scale precipitations. Pleasingly, this resulted in a
moderate, 71% yield of the mono-hydrochloride of 2 over the
two stages of chemistry and salt formation. The isolated
material was obtained with good purity (typically >98% peak
area by HPLC), and the majority of the more than 40
impurities generated during the telescoped chemistry had been
9
1
removed to the liquors. H NMR analysis of 2 indicated that
no 2-butanone remained.
The recovery of the hydrochloride salt was maximized by
ensuring that the 2-butanone slurry was completely dry, as
determined by Karl Fisher analysis, prior to isolation of the
solids. However, this was found to give considerable levels of
inorganic contamination, determined by residue on ignition
testing, when saturated solutions of sodium chloride were used.
It was found that the level of acetic acid was unimportant and
could be charged between 1.5 and 10 mol equiv. From a cost
perspective, the level of 1.5 equiv was chosen for further work.
The number of washes of saturated sodium chloride solution
also had no impact upon the isolated yield and purity of the
hydrochloride salt. There was also no advantage in yield to be
gained upon performing further iterations of the sequence of
acetic acid treatment and sodium chloride washing.
In order to minimize the inorganic burden, sodium chloride
solutions of varying strength were prepared and examined in
the salt formation, using a stock solution of crude material to
minimize the impact of variation in composition upon the
crystallization (Table 1). All of the reactions gave satisfactory
recovery of product, indicating that a lower concentration of
sodium chloride was well tolerated and minimized the
inorganic content of the solids.
During early simplification of the chemistry to 2 it was found
that very small amounts of crystalline material precipitated from
dichloromethane solutions upon transfer between vessels, but
only under certain processing conditions. The key condition
which led to this precipitation was the omission of a basic wash
to remove any acetic acid at the end of the desilylation step
prior to washing with aqueous sodium chloride. Further
analysis of the crystalline material indicated that it was the
7
mono-hydrochloride 2 of high purity, despite the fact that no
hydrochloric acid had been used in the processing. It was
postulated that the hydrochloride salt was formed by a
combination of the acetic acid providing the proton and the
sodium chloride providing the chloride anion. Clearly, on the
basis of the acidities of the respective acids (hydrochloric acid
8
and acetic acid) this result is counterintuitive, but we felt that
it was deserving of further investigation.
In order to ensure that the process was starting from a point
of control at the end of the desilylation step, we reintroduced
the potassium carbonate wash to ensure that no trace of acetic
acid, from the desilylation step, was present. To this 2-butanone
solution of the free base of crude 2 was added a small excess of
acetic acid. This solution of (presumed) acetate salt was washed
with an aqueous solution of sodium chloride. Analysis of the
aqueous layer indicated minimal losses of 2 from the organic
It should be noted that treatment with acetic acid alone or
washing with sodium chloride solution alone resulted in failure
to isolate any product.
The difference in the calculated pK for the conjugate acids
a
of the amine centres (9.11 and 4.16) in 2 indicated a
considerable window for choice of acid to provide the proton in
this approach to formation of the hydrochloride salt without
(
2-butanone) layer to the aqueous wash. The resultant wet 2-
5
19
dx.doi.org/10.1021/op2003375 | Org. Process Res. Dev. 2012, 16, 518−523

Organic Process Research & Development
Technical Note
Table 1. Impact of volume and composition of NaCl wash
upon yield and inorganic content
washes were retained to ensure similar control of water levels in
the 2-butanone during the washing sequence.
The requirement for complete water removal from the 2-
butanone solution of 2 was troublesome. 2-Butanone partitions
with aqueous solutions, but considerable mixing of the organic
and aqueous contents for each layer is observed. This meant
that a significant volume of solvent was required to be removed
by distillation and replaced with fresh solvent to ensure that the
Karl Fisher analysis indicated that an acceptable yield would be
achieved. Any alternative solvent would have to meet a number
of criteria: it would need to enable an aqueous work up after
the alkylation step, be a competent cosolvent in the cesium
fluoride mediated desilylation, enable efficient salt formation
through sequential treatment with acetic acid and aqueous
sodium chloride washing, and enable clean phase separations in
the washing regime.
yield in
total
yield
(%)
inorganic
NaCl (wt
isolated
liquors
content (%w/
a
b
c
entry
%); volume yield (%)
(%)
w)
1
2
3
4
6%; 5 vols
6%; 10 vols
10%; 5 vols
63.8
59.8
66.9
71.6
16.8
13.8
14.4
9.1
80.6
73.6
81.3
80.7
0.99
1.76
0.87
1.41
10%; 10
vols
5
6
29%; 5 vols
68.5
57.0
5.4
73.9
79.5
2.47
29%; 10
vols
22.5
15.89
a
Reactions were run in Radley’s carousel tubes, splitting one 2-
butanone reaction such that each salt formation was based on an input
of 5 g. Solutions of the free base of 2 were treated with acid for 30 min,
washed with 29% w/v NaCl (5 vols) and then three times with 6% w/
v NaCl (5 vols), diluted with 2-butanone (5 vols) and distilled (to
remove 4 vols), diluted with 2-butanone (4 vols), seeded, and distilled
to remove 4 vols) prior to being isolated by filtration. Determined
Eight potential solvents were examined on the basis of their
10
lower solubility in water (Table 3). Of these, the pentanols
b
(
c
by HPLC assay. Determined by residue on ignition (ROI) analysis.
Table 3. Solubility of potential alternative solvents in
10
water.
risking formation of the bis-hydrochloride. We examined a
range of acids within this window, and a couple that were more
acidic than the aniline nitrogen position, to define the scope of
acids and a failure mode of the new salt formation. All of the
acids were examined using a single stock batch of material to
minimize the impact of different impurity profiles. All of the
isolated solids were analyzed by ion chromatography to
determine the salt stoichiometry. The results are shown in
Table 2 and indicate that strong acids that can protonate the
a
entry
solvent
2-butanone
solubility in water (mass %)
1
2
3
4
5
6
7
8
25.6
5.5
2-pentanone
3-pentanone
2-hexanone
4.9
1.49
0.435
1.85
2.14
4.3
2-heptanone
methyl isobutyl ketone
1-pentanol
2-pentanol
9
3-pentanol
5.6
Table 2. Variation of proton source and impact on yield and
stoichiometry of 2
a
All figures for 25 °C.
a
entry
acid
acetic acid
pKa
yield (%) mono·/bis·HCl
and 2-pentanone were found to give the desired hydrochloride
2. Some of the longer-chain and bulkier ketones resulted in
deposition of the intermediate acetate salt or desired hydro-
chloride salt as gums during the washing regime. Further scale-
up and comparison of isolated yields and purity indicated that
1-pentanol was the optimal solvent to replace 2-butanone and
resulted in much reduced solvent use. Indeed, the water
content of the 1-pentanol layer was sufficiently low immediately
after the sodium chloride wash that seeding of product was
possible without distillation. A single, small distillation was
1
2
3
4
5
6
7
4.75
9.24
1.29
2.86
4.47
4.08
76
mono
mono
mixture
mixture
mono
mono
mono
b
ammonium chloride
dichloroacetic acid
chloroacetic acid
55.5
25.3
63.3
67.3
48.6
9.1
p-methoxybenzoic acid
o-methoxybenzoic acid
c
pentafluorophenol
∼7
a
Reactions were run in Radley’s carousel tubes, splitting one 2-
butanone reaction such that each salt formation was based on an input
of 5 g. Solutions of the free base of 2 were treated with acid for 30 min,
washed with 29% w/v NaCl (5 vols) and then three times with 6% w/
v NaCl (5 vols), diluted with 2-butanone (5 vols) and distilled (to
remove 4 vols), diluted with 2-butanone (4 vols), seeded, and distilled
to remove 4 vols) prior to being isolated by filtration. 2-Butanone
solution washed with 29% w/v NH Cl (5 vols) and then three times
with 6% w/v NH Cl (5 vols), diluted with 2-butanone (5 vols) and
1
subsequently performed to increase the isolated yield. H NMR
analysis of 2 indicated that no 1-pentanol was retained.
With an optimized protocol in hand we ran a comparison
between our new process, using acetic acid and sodium chloride
washes and the traditional use of aqueous hydrochloric acid. A
telescoped alkylation and desilylation was split into two
portions. One half was submitted to our new protocol, and
the other was treated with 1.2 equiv of concentrated aqueous
hydrochloric acid and distilled as the new protocol. The lower
molar equivalent of concentrated aqueous hydrochloric acid
(relative to acetic acid in the new process) was deliberately
targeted to reduce the risk of deformylation. Although the
yields were highly consistent between the two, the new
protocol produced the product with lower levels (typically <2%
peak area by HPLC) of the deformylated compound and less
colouration of the isolated material.
b
(
4
4
distilled (to remove 4 vols), diluted with 2-butanone (4 vols), seeded,
and distilled (to remove 4 vols) prior to being isolated by filtration.
Low yield potentially caused by omission of the final distillation.
c
aniline nitrogen lead to mixtures of mono- and bis-hydro-
chlorides and poor control of salt stoichiometry. All of the acids
which fell within the range bordered by the acidities of the two
nitrogens led to good stoichiometric control generating the
monohydrochloride, although the yields were somewhat
variable. The optimum acid was acetic acid in terms of yield,
availability, and cost. In addition ammonium chloride also fared
well as a source of hydrochloric acid for the salt formation. In
this instance, no acetic acid was used, but the sodium chloride
With a new process to 2 in hand it was necessary to verify
that material could be processed to provide the API
(GSK159797, 1) in an appropriate yield and purity. Submission
5
20
dx.doi.org/10.1021/op2003375 | Org. Process Res. Dev. 2012, 16, 518−523

Organic Process Research & Development
Technical Note
Figure 2. Equilibria and partitions set up during sodium chloride washing of acetic acid treated free base of 2.
of the hydrochloride salt of 2 to palladium-catalyzed hydro-
genolysis conditions using N-methylpyrrolidinone and 2-
propanol as solvents enabled clean hydrogenolysis of the
benzyl protecting group to give 1 in good yield and purity. Ion
chromatography analysis of the output 1 showed that retention
of the hydrochloride salt stoichiometry had been achieved. No
additional hydrochloric acid was required to provide stoichio-
metric control of the mono-hydrochloride salt 1. Application of
the new protocol for formation of the hydrochloride of 2
enabled clean production of the API with stoichiometric
control of the salt and also purification at an intermediate stage
preferentially incorporated into the salt. The acetate counter-
ion, not utilized, is rejected into the aqueous layer. Subsequent
washing enhances the chloride content of the organic layer. In
the limiting extreme investigated, the acetate salt was subjected
to only 2.4 mol equiv of chloride which represents an
impressive incorporation of available chloride. It is important
to note that an alternative mechanistic scenario can be
discounted. The level of acetate removed to the aqueous
washes is inconsistent with a scenario in which the freebase of 2
forms an acetate salt, which is retained in the organic layer
along with sodium chloride, after the aqueous washes.
Crystallization would then induce a shift of the equilibrium
through phase separation, leading to isolation of hydrochloride
salt 2.
allowing control of the overall purity of the API.
8
Acetic acid (pK ≈ 4.75 in water; 12.3 in DMSO) is not
a
acidic enough to protonate a chloride ion as it is far less acidic
8
than hydrochloric acid (pK ≈ −8 in water; 1.8 in DMSO).
a
When the free base of 2 is treated with excess acetic acid, a
mono-acetate salt will be the major species formed (Figure 2).
Acetic acid is not anticipated to be acidic enough to generate
any appreciable levels of the bis-acetate salt. The free base of 2
and all of these salt forms are evidently soluble in wet organic
solvents which possess a solubility of >2 mass % in water. When
the solutions of acetate salt are washed with aqueous sodium
chloride, a complex series of equilibria and partition events are
set up. In the organic phase, the free base of 2 enters into
equilibrium between the formed acetate salt and the hydro-
chloride salt. Partitioning of the various species present
between the aqueous and organic phases upon phase separation
is also an important factor in determining the outcome.
In order to gain clarity on the specifics of the salt formation,
we examined a hydrochloride salt formation from purified free
base of 2 in 1-pentanol. The free base was formed by base
washing a typical product derived from this process. By using
purer material we were able to trace the loss of acetate ions to
the aqueous layers during the phase separations. The first wash
contained 70% and the second contained 15% of the available
acetate, which correlated well with the isolated yield of 86% for
this reaction. The salt metathesis apparently occurs during the
phase separations to leave hydrochloride salt 2 in the wet
organic layer prior to distillations. At the point of phase
separation, the organic soluble ammonium salt requires a
counterion, and the most prevalent species, chloride, is
SUMMARY
■
A new protocol for the isolation of 2 was developed through
the sequential treatment of the free base with acetic acid and
aqueous sodium chloride washes. This protocol enabled
isolation of the hydrochloride salt in good yield and purity
after a lengthy telescoped process while avoiding complications
of bis-hydrochloride formation and additional undesired
deformylation. Yields of the telescoped process depended
slightly upon the events of the earlier chemistry steps and
impurity profile, yet the successful isolation of purified 2 proved
to be highly robust and independent of the impurity profile of
crude mixture. The telescoped process was scaled up to 109 kg
input (of 4) and generated 2 in 70−75% yield. Subsequent
processing through the debenzylation to form 1 proceeded
without incident on 60 kg input scale. The application of this
approach rendered complete control of stoichiometry of the
mono-hydrochloride salt 1 to fund toxicological and early
clinical studies.
While the full kinetic and mechanistic details of the salt
formation have not been elucidated, there is sufficient flexibility
demonstrated through the range of acids and range of solvents
to make this approach worth considering for similar lipophilic
amine free bases.
5
21
dx.doi.org/10.1021/op2003375 | Org. Process Res. Dev. 2012, 16, 518−523

Organic Process Research & Development
Technical Note
EXPERIMENTAL SECTION
separated. The organic layer was washed with water (500 mL)
followed by 2% w/v aqueous sodium chloride (500 mL).
To the organic layer was added acetic acid (20 mL, 350
mmol) followed by a solution of cesium fluoride (39 g, 257
mmol) in methanol (500 mL). The mixture was heated at 55
■
1
1
Preparation of 2 Using 2-Butanone.. 2-[4-((R)-2-
Hydroxy-2-phenylethylamino)phenyl]ethylamine bis-hydro-
chloride (107.9 kg) was dissolved in water (440 L). Isopropyl
acetate (550 L) was added. A 32% w/v aqueous sodium
hydroxide solution (95 L) was added over at least 30 min. The
organic layer was washed with water (550 L) and then was
distilled at atmospheric pressure to a volume of ∼380 L.
To this solution was added N,N-dimethyl acetamide (DMAc,
°
C for 4 h. The reaction was quenched with 37% w/v aqueous
potassium carbonate solution (500 mL) and cooled to 30 °C.
The organic layer was washed with 10% w/v aqueous sodium
chloride solution (500 mL) and then treated with acetic acid
(
18.5 mL, 324 mmol). The mixture was washed with 10% w/v
2
(
75 L) followed by 2-bromo-(R)-1-tert-butyldimethylsiloxy-1-
aqueous sodium chloride solution (2 × 500 mL).
The solution was diluted with 1-pentanol (1 L), seeded with
3-formamido-4-benzyloxyphenyl)ethane (109 kg) and potas-
1
2
sium carbonate (40.5 kg). The mixture was heated at 90 °C
for 17 h and then cooled to 50 °C. Water (820 L) was added,
and the mixture was cooled further to room temperature. 2-
Butanone (820 L) was added, and the layers were separated.
The organic layer was washed with 17:40:340 (v/w/v) acetic
acid/sodium acetate/water (550 L) followed by 29% w/v
aqueous sodium chloride (550 L). The organic layer was
diluted with 2-butanone (270 L) and then was distilled at
atmospheric pressure to a volume of approximately 820 L,
followed by addition of more 2-butanone (270 L). The mixture
was heated to 37 °C, and a solution of cesium fluoride (44.1
kg) in methanol (550 L) was added. Heating at 37 °C was
continued for 7.5 h, and then the mixture was cooled to 30 °C.
The reaction was quenched with 44% w/v aqueous potassium
carbonate solution (550 L), and water (110 L) was added. The
organic layer was washed with 29% w/v aqueous sodium
chloride solution (550 L) and then treated with acetic acid
2
(0.2 g), and aged for 30 min. The mixture was distilled under
reduced pressure (100 mbar) to a volume of about 1.4 L and
then cooled to room temperature. The solids were collected by
filtration, washed with 1-pentanol (2 × 300 mL) followed by
ethyl acetate (300 mL), and dried in vacuo to give 2 as a
colourless solid (86.09 g, 71% yield). H NMR in accord of
structure (400 MHz, DMSO-d ) δ (ppm): 2.70−2.89 (m, 2H);
1
6
2
.95 (m, 1H); 3.01−3.14 (m, 4H); 3.14−3.23 (m,, 1H); 4.71
(
m, 1H); 4.81 (m, 1H); 5.17* (s, 1H); 5.23 (s, 1H); 5.46 (d, J
4.4 Hz, 1H); 5.50 (m, 1H); 6.10 (d, J = 3.2 Hz, 1H); 6.59 (d,
J = 8.3 Hz, 2H); 6.94 (d, J = 8.3 Hz, 2H); 7.03 (d of d, J = 8.6,
.0 Hz, 1H); 7.12 (d, J = 8.6 Hz, 1H); 7.25 (m, 1H); 7.30−7.36
m, 3H); 7.36−7.42 (m, 4H); 7.50 (d, J = 7.3 Hz, 2H); 8.26 (d,
J = 2.0 Hz, 1H); 8.35 (d, J = 1.7 Hz, 1H); 8.45* (d, J = 11.0 Hz,
=
2
(
13
1
H); 8.63 (br, 2H); 9.64* (m, 1H); 9.67 (s, 1H). C (100
MHz, DMSO-d ) δ (ppm):30.6, 48.5, 51.5, 53.5, 68.0, 69.7,
6
7
1
1
0.7, 112.4, 118.4, 121.5, 124.0, 126.0, 126.9, 127.1, 127.6,
27.8, 128.0, 128.4, 129.1, 134.0, 136.7, 144.2, 146.9, 147.5,
60.3. HRMS (ES +ve) calcd for C H N O : 526.2700,
(
20.2 L). The mixture was washed with 29% w/v aqueous
sodium chloride solution (550 L) and then 6% w/v aqueous
sodium chloride solution (3 × 550 L).
The solution was diluted with 2-butanone (550 L) and then
distilled at atmospheric pressure to a volume of about 660 L. 2-
Butanone (440 L) was added, and the mixture was seeded with
32
35
3
4
found: 526.2694.
*Peaks are due to ∼11.5 mol % of the minor rotamer due to
the presence of the formamide group.
Preparation of 1. A mixture of N-{2-[4-((R)-2-hydroxy-2-
phenylethylamino)phenyl]ethyl}-(R)-2-hydroxy-2-(3-formami-
do-4-benzyloxyphenyl)ethylamine monohydrochloride (2, 60
kg), N-methylmorpholine (3.6 L), and 5% Pd/C catalyst
2. The mixture was further distilled to a volume of about 770 L.
More 2-butanone (330 L) was added, and the mixture was
cooled to room temperature. The solids were collected by
filtration, washed with 2-butanone (3 × 110 L), and dried in
(
Engelhard 167, 50% wet with water; 0.6 kg) in N-
1
vacuo to give 2 as a colourless solid (99 kg, 75% yield). H
methylpyrrolidinone (NMP, 150 L) was stirred under 1.5
atm of hydrogen at 22 ± 2 °C until the reaction was judged
complete by HPLC analysis. The mixture was filtered through a
Eurofiltec filter to remove the catalyst and then through a Cuno
Zeta carbon filter to remove dissolved palladium followed by a
polishing filter. The filter system was washed successively with
NMP (60 L), followed by 2-propanol (90 L).
NMR in accord of structure (400 MHz, DMSO-d ) δ (ppm):
6
2
3
.70−2.89 (m, 2H); 2.95 (m, 1H); 3.01−3.14 (m, 4H); 3.14−
.23 (m, 1H); 4.71 (m, 1H); 4.81 (m, 1H); 5.17* (s, 1H); 5.23
(
s, 1H); 5.46 (d, J = 4.4 Hz, 1H); 5.50 (m, 1H); 6.10 (d, J = 3.2
Hz, 1H); 6.59 (d, J = 8.3 Hz, 2H); 6.94 (d, J = 8.3 Hz, 2H);
.03 (d of d, J = 8.6, 2.0 Hz, 1H); 7.12 (d, J = 8.6 Hz, 1H); 7.25
7
(
7
8
m, 1H); 7.30−7.36 (m, 3H); 7.36−7.42 (m, 4H); 7.50 (d, J =
The combined filtrates were stirred and heated to 69 ± 3 °C.
Water (30 L) was added followed by 2-propanol (180 L) added
at a rate to ensure the temperature remained at 69 ± 3 °C. Seed
crystals of 1 (0.06 kg) were added. The mixture was aged for 60
min before 2-propanol (390 L) was added over 1.5 to 2 h at 69
± 3 °C. The resulting slurry was stirred at 69 ± 3 °C for 1 h
and then cooled to room temperature over 4 h and aged at 20
°C for 4 h.
The solids were collected by filtration and washed
successively with 10:1 v/v 2-propanol/water (120 L) and
then 2-propanol (180 L). The solids were dried in vacuo at 50
°C to give 1 as colourless crystals (39.9 kg, 79%) yield. H
NMR in accord of structure (400 MHz, DMSO-d ) δ (ppm):
.3 Hz, 2H); 8.26 (d, J = 2.0 Hz, 1H); 8.35 (d, J = 1.7 Hz, 1H);
.45* (d, J = 11.0 Hz, 1H); 8.63 (br, 2H); 9.64* (m, 1H); 9.67
(
s, 1H). HRMS (ES +ve) calcd for C H N O 526.2700,
3
2
35
3
4
found 526.2694.
*
Peaks are due to ∼11.5 mol % of the minor rotamer due to
the presence of the formamide group.
Preparation of 2 in 1-Pentanol.. 2-Bromo-(R)-1-tert-
1
1
butyldimethylsiloxy-1-(3-formamido-4-benzyloxyphenyl)ethane
(
4, 100 g, 215 mmol), 2-[4-((R)-2-Hydroxy-2-
13
phenylethylamino)phenyl]ethylamine bis-hydrobromide salt
12
1
(
3, 99 g, 237 mmol), and potassium carbonate (119 g, 861
mmol) were charged to a reactor. N-Methylpyrrolidinone
NMP, 500 mL) was added, and the mixture was heated at
6
(
2.73−2.89 (m, 2H); 2.95 (m, 1H); 3.01−3.14 (m, 4H); 3.15−
3.24 (m, 1H); 4.72 (m, 1H); 4.82 (m, 1H); 5.46 (d, J = 4.7 Hz,
1H); 5.48 (m, 1H); 6.03 (d, J = 3.4 Hz, 1H); 6.59 (d, J = 8.6
Hz, 2H); 6.89 (d, J = 8.1 Hz, 1H); 6.91−6.98 (m, 3H); 7.01*
1
10−115 °C for 5 h and then cooled to 50 °C. Water (900 mL)
was added followed by 1-pentanol (500 mL), the mixture was
cooled further to room temperature, and the layers were
5
22
dx.doi.org/10.1021/op2003375 | Org. Process Res. Dev. 2012, 16, 518−523

Organic Process Research & Development
d, J = 8.6 Hz, 1H); 7.14* (s, 1H); 7.25 (t, J = 7.3 Hz, 1H);
Technical Note
(
7
(7) The identity of the crystalline material as being the acetate salt
was ruled out by analysis of the H NMR spectrum. Ion
chromatography confirmed the presence of one molar equivalent of
1
.33 (d of d, J = 7.3, 7.6 Hz, 2H); 7.39 (d, J = 7.6 Hz, 2H); 8.13
d, J = 1.5 Hz, 1H); 8.29 (d, J = 1.7 Hz, 1H); 8.53* (d, J = 11.0
Hz, 1H); 8.57−9.08 (br, 2H); 9.36* (d, J = 11.0 Hz, 1H); 9.60
(
chloride.
1
3
(8) Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456−463.
9) Tables of impurities generated in the alkylation of 3 with 4 and
desilylation of 5 (structures, relative retention times of, and typical
levels observed) are included in the Supporting Information.
(
s, 1H); 9.92* (s, 1H); 10.10 (s, 1H). C (100 MHz, DMSO-
(
d6) δ (ppm): 30.6, 48.5, 51.5, 53.3*, 53.5, 67.7*, 68.1, 70.7,
1
1
1
12.4, 114.8, 116.0*, 118.4, 119.4*, 121.5, 123.0, 124.0, 125.9,
26.0, 126.9, 128.0, 129.1, 132.2, 132.7*, 144.2, 146.3, 147.5,
60.0, 163.4*. Anal. Calcd for 1: C, 63.8; H, 6.2; N, 8.9; Cl, 7.5;
(
10) Haynes, W. M., Ed. Aqueous Solubility and Henry’s Law
Constants of Organic Compounds. In CRC Handbook of Chemistry and
Physics, 91st ed, Internet Version 2011; CRC Press/Taylor and
Francis, Boca Raton, FL, 2011.
found: C, 63.7; H, 6.4; N, 8.9; Cl, 7.4. HRMS (ES +ve) calcd
for C H N O : 436.2231, found: 436.2241.
25
29
3
4
(
11) Two processes for the preparation of 2 are provided. The first,
*
Peaks are due to ∼11 mol % of the minor rotamer due to
using 2-butanone, was used on the largest scale. The second, using 1-
pentanol, was the preferred process for further scale-up yet was not
implemented due to cessation of development activities.
(12) The use of finely divided “325 mesh” potassium carbonate is
important for the success of the alkylation. The use of “granular”
potassium carbonate resulted in a slower, incomplete reaction with
greater levels of impurities.
the presence of the formamide group.
ASSOCIATED CONTENT
■
*
S
Supporting Information
1
13
H and C NMR, LC/MS, accurate mass and HPLC data of 1
and 2. This information is available free of charge via the
Internet at http://pubs.acs.org.
(
13) During development work, 3 was changed from the bis-
hydrochloride salt to the bis-hydrobromide salt. This avoided the
formation of the chloro-analogue of 4, which was substantially less
reactive and led to lower conversions in the alkylation step.
AUTHOR INFORMATION
■
Corresponding Author
peter.z.shapland@gsk.com.
*
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We acknowledge the Analytical Sciences department at
GlaxoSmithKline, Stevenage for running numerous Karl Fisher
and ion chromatography analyses in the development of this
work.
REFERENCES
■
(
1) Stahl, P. H.; Wermuth, C. G., Eds. Handbook of Pharmaceutical
Salts: Properties, Selection, and Use; Wiley-VCH: Weinheim, 2011.
(
2) Brittain, H. G., Ed. Polymorphism in Pharmaceutical Solids; Marcel
Dekker: New York, 1999.
3) Haynes, D. A.; Jones, W.; Motherwell, W. D. S. J. Pharm. Sci.
005, 94 (10), 2111−2120.
4) (a) Wojciechowska-Nowak, M.; Boczon
Warzajtis, B. J. Mol. Struct. 2007, 840, 44−52. (b) Padi, P. R.; Akundi,
(
2
(
́
, W.; Rychlewska, U.;
̇
S. P.; Suthurapu, S. K.; Kolla, N. K.; Kotagiri, V. K.; Neelam, U. K.;
Baddam, S. R.; Manne, N. (Dr. Reddy’s Laboratories, Ltd., India, and
Dr. Reddy’s Laboratories, Inc., United States). Process for the
preparation of Cinacalcet. WO 2008/058235 A2, 2008. (c) Castellano,
S.; Kuck, D.; Sala, M.; Novellino, E.; Lyko, F.; Sbardella, G. J. Med.
Chem. 2008, 51, 2321−2325. (d) Jones, K.; Newton, R. F.; Yarnold, C.
J. Tetrahedron 1996, 52, 4133−4140. (e) Brambilla, R.; Friary, R.;
Ganguly, A.; Puar, M. S.; Sunday, B. R.; Wright, J. J.; Onan, K. D.;
McPhail, A. T. Tetrahedron 1981, 37, 3615−3625.
(
5) (a) Linsell, M. S.; Jacobsen, J. R.; Khossravi, D.; Paborji, M.;
Zhang, W. (Theravance, Inc., United States). Crystalline β2
Adrenergic Receptor Agonist. WO 2004/011416 A1, 2004.
(
b) Stergiades, I.; Yost, E.; Hubbard, C.; Zhang, W. (Theravance,
Inc., United States). Crystalline Form of β2 Adrenergic Receptor
Agonist. WO 2004/106279 A2, 2004. (c) Caine, D. M.; Paternoster, I.
L.; Shapland, P. D. P. (Glaxo Group Ltd, U.K.). Chemical process for
preparing and new crystalline form of N-{2-[4-((R)-2-hydroxy-2-
phenylethylamino)phenyl]ethyl}-(R)-2-hydroxy-2-(3-formamido-4-
benzyloxyphenyl)ethylamine monohydrochloride. WO 2005/095328
A1, 2005.
(
6) pK ’s were calculated using ACD, v11, Advanced Chemistry
a
Development, Inc., Toronto, Ontario, Canada, http://www.acdlabs.
com.
5
23
dx.doi.org/10.1021/op2003375 | Org. Process Res. Dev. 2012, 16, 518−523