An Improved Convergent Synthesis of AZD7762: A One-Pot Construction of a Highly Substituted Thiophene at the Multikilogram Scale

Article abstract of DOI:10.1021/acs.oprd.6b00364

A multikilogram synthesis of AZD7762 has been achieved using a highly convergent route employing two efficient telescoped sequences to generate the key intermediates. Aminothiophene 11 is formed in a four-step, one-pot addition-elimination-cyclization sequence from cinnamonitrile 9, constructing the trisubstituted thiophene ring with the desired API substitution pattern in place. Cinnamonitrile 9 is derived by elaboration of 3-fluoroacetophenone. Generation of the urea function, followed by deprotection, affords AZD7762 in 49% yield over 5 isolated stages from chiral piperidine 5, a 5-fold increase in yield versus the first generation route, reducing the starting material burden and eliminating the previous requirement for metal-mediated couplings and chromatography.

Full text of DOI:10.1021/acs.oprd.6b00364

Article
pubs.acs.org/OPRD
An Improved Convergent Synthesis of AZD7762: A One-Pot
Construction of a Highly Substituted Thiophene at the Multikilogram
Scale
Matthew Ball,* Martin F. Jones, Fiona Kenley, and J. David Pittam
Pharmaceutical Technology and Development, AstraZeneca, Silk Road Business Park, Macclesfield, Cheshire, SK10 2NA, United
Kingdom
S
* Supporting Information
ABSTRACT: A multikilogram synthesis of AZD7762 has been achieved using a highly convergent route employing two efficient
telescoped sequences to generate the key intermediates. Aminothiophene 11 is formed in a four-step, one-pot addition−
elimination−cyclization sequence from cinnamonitrile 9, constructing the trisubstituted thiophene ring with the desired API
substitution pattern in place. Cinnamonitrile 9 is derived by elaboration of 3-fluoroacetophenone. Generation of the urea
function, followed by deprotection, affords AZD7762 in 49% yield over 5 isolated stages from chiral piperidine 5, a 5-fold
increase in yield versus the first generation route, reducing the starting material burden and eliminating the previous requirement
for metal-mediated couplings and chromatography.
KEYWORDS: thiophene, substituted, one-pot, convergent, synthesis, kilogram
would be required for the subsequent manufacturing campaign
with a delivery target of 5 kg API.
INTRODUCTION
■
AZD7762 was designed as a checkpoint kinase (CHK1)
inhibitor to prevent DNA repair at key checkpoints (or “rest
points”) in the cell cycle. The planned mode of action was to
increase the sensitivity of cancer cells to DNA damaging
treatments, and the candidate drug underwent evaluation in
Phase II clinical trials as a combination therapy against various
solid tumors. The first generation manufacturing route
(Scheme 1)1 started from the commercially available
substituted thiophene 1. This was a modification of the early
discovery route, and although this approach successfully
produced 1.6 kg of API, a number of drawbacks were noted
following the first kilo-lab scale manufacturing campaign.
Bromination of thiophene ester 2 required 5 equiv of bromine
to consume the starting material, yet also led to contamination
of 3 with overbrominated products which were difficult to
remove. Formation of the amide bond of 6 proceeded in only
modest yield and required significant excesses of both the
expensive chiral amino piperidine 5 (2.5 equiv), and
trimethylaluminum (2.6 equiv). Further operational difficulties
were also encountered during the extractive workup involving
treatment with Rochelle’s salt to remove aluminum salts,
whereby gel formation and problematic phase separation were
encountered. The aryl−aryl bond was introduced using a
Suzuki coupling to afford 8, which required chromatography to
remove homocoupled impurities. The level of these impurities
was found to be agitation rate dependent, thus displaying a lack
of robustness and predictability causing concern for future
scale-up. Boc deprotection delivered the final API, but further
purification was necessary which involved screening of the
solution through a cartridge of Quadrapure TU thiourea resin
to remove residual palladium prior to crystallization of the drug
substance. It was clear that an alternative synthetic approach
RESULTS AND DISCUSSION
■
Strategic Approach. The problematic areas of the existing
route were clearly those concerned with the construction of the
fully functionalized thiophene ring, specifically the installation
of the aryl and amide moieties. As an alternative approach, we
sought instead to investigate strategies to generate the
thiophene ring in a cyclization reaction which would lead
unambiguously to the desired 2,3,5-substitution pattern.
If this was successful, the complexity and cost associated with
the route could be significantly reduced. Our attentions turned
to reported methodology which involves reaction of β-
chlorocinnamonitriles with α-mercaptoacetic esters in the
presence of base at high temperatures, generating 2-substituted
thiophene esters.2 Our aim, however, was to extend this
methodology to develop a mild, scalable process which
encompassed secondary amides,3 allowing direct access to
thiophene 11 with the desired substitution in place (Scheme 2).
This would also eliminate the need for the inefficient late-stage
amidation of the ester function. This strategy required safe,
efficient generation of both the cinnamonitrile 9 and α-
mercaptoacetic amide 10, the latter of which was not expected
to be crystalline. For this reason, further investigation was
required into a protected surrogate for 10.
Synthesis of Cinnamonitrile 9 (Telescope 1). The new route
began with the synthesis of the required cinnamonitrile 9, using
Arnold homologation chemistry (Scheme 3).4 Preliminary
attempts involved preforming the Vilsmeier reagent by the
addition of phosphoryl chloride to DMF, followed by addition
Received: November 2, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. First-Generation Manufacturing Route to AZD7762
Scheme 2. Second Generation Retrosynthetic Strategy for AZD7762, Trisubstituted Thiophene Synthesis
Scheme 3. Telescoped Synthesis of Cinnamonitrile 9
of 3-fluoroacetophenone (final concentration 0.50 kg/L) to
initially generate iminium 13. Detailed process safety studies
indicated that this procedure would be unacceptably
exothermic for the required scale of manufacture. In order to
remedy this, a greater quantity of DMF was implemented (final
concentration 0.14 kg/L). Crucially, the order of addition was
also altered by beginning with a warm solution of 3-
fluoroacetophenone in DMF (40 °C), prior to the slow
addition of phosphoryl chloride. Performing the addition at this
higher temperature facilitated the rapid consumption of the
Vilsmeier reagent, preventing accumulation and therefore
reducing the possibility for thermal runaway. Once these
changes were in place, the process was inherently safe at the
proposed scales in the required kilolab equipment.
In situ conversion of the intermediate iminium species 13 to
cinnamonitrile 9 [single geometrical isomer; (Z) by 1H NMR/
NOESY] was performed by the addition of solid hydroxylamine
hydrochloride to generate the oxime 14, followed by
spontaneous dehydration in the reaction medium. Unfortu-
nately, a second potentially uncontrollable exotherm was
encountered when hydroxylamine hydrochloride was dosed as
a solid, presumably exacerbated by its notably slow dissolution.
B
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 4. Key Telescoped Synthesis of Thiophene 11, Describing Each Operational Stage
The use of a solution of hydroxylamine hydrochloride in DMF
(14.4 M) allowed more controlled addition, although hydroxyl-
amine hydrochloride was shown to be unstable in neat DMF at
temperatures >30 °C. Care was therefore taken to constitute
the solution at 20 °C immediately prior to use.5 An additional
delayed exotherm was also noted at the end of the addition. In
a similar vein to the Vilsmeier formation, this was rendered safe
by the slow addition of the hydroxylamine solution to the
40 °C reaction mixture, a temperature common to the first
stage. In this manner, the prevention of accumulation of
hydroxylamine hydrochloride and intermediates resulted again
in a basis of control for the exotherm.
At this point the synthesis of cinnamonitrile 9 had been
accomplished with a high degree of control in an easily
operable, one-pot sequence. At the end of reaction, simply
cooling and treating with water facilitated direct crystallization
of 9 in high purity and 65% yield;6 a total of 10 kg of 9 was
generated using this process. Should future demand for this
building block increase significantly, we would also envisage
implementation of the recent continuous-flow based approach
to chlorocinnamonitriles, reported by Buchwald and Pellegatti.7
Synthesis and Delivery of Aminothiophene 11 (Telescope
2). With cinnamonitrile 9 in hand, we began developing the key
cyclization chemistry to generate the thiophene ring (Scheme
4). As outlined previously, in an extension to the existing
methodology, we opted to build the required thiol amide as
opposed to the thiol ester. This required stepwise construction
of the amide 15 followed by introduction of a stable, crystalline
source of sulfur masked with a suitable protecting group. In this
regard we first selected thiobenzoic acid; however, the volatility
and associated odor were deemed unfavorable for use at scale.
The sodium or potassium salt of thiobenzoic acid was also
sought as a potentially less volatile alternative, but when it was
found that this material was not widely available, this approach
was not pursued further. Potassium thioacetate was then
examined; as well as being easily handled, this would have the
benefit of carrying an acetate as a protecting group which might
be easily unmasked prior to cyclization. Unfortunately, attempts
at isolation of chloroacetamide 15 or thioacetate 16 as
crystalline solids were unsuccessful, leading to oils in each case.
Attention then turned to developing a telescoped sequence,
to avoid the isolation of any intermediates en route to 11. We
required a suitable solvent to use throughout; the formation of
α-chloroacetamide 15 was found to be a facile process in 2-
methyl THF, and in the absence of a chromophore, reaction
progress could be followed by gas chromatography. An excess
of chloroacetyl chloride (1.15 equiv) was required to give
complete reaction to 15 in 2 h, and this initially led to
overacylated impurities downstream. This was remedied by
simply agitating with aqueous sodium chloride at the end of the
reaction to hydrolyze excess chloroacetyl chloride, facilitating a
subsequent phase separation leaving 15 within the upper
organic phase. The use of N,N-diisopropylethylamine (DIPEA)
as a base during this stage was found to generate highly colored
reaction mixtures, which persisted through the sequence and
resulted in 11 as a discolored brown solid. After an organic base
screen, pyridine was found to be a superior alternative,
completely eliminating unwanted coloration in the reaction
medium and subsequent solid product. It is postulated that the
higher basicity of DIPEA had facilitated deprotonation of
chloroacetyl chloride at the α-position, generating chloroketene
which can polymerize or react further to give low levels of
colored impurities. With further investigation, this may prove to
be a general consideration when choosing the optimum base
for this type of acylation reaction.
Direct treatment of the resulting 2-MeTHF solution of 15
with potassium thioacetate initially led to sticky deposits of
potassium chloride around the vessel walls, poor mixing, and
trapping of starting materials. Since the solubility of potassium
chloride in 2-MeTHF is poor, we switched to a biphasic system,
dissolving the potassium thioacetate in water prior to addition
to the solution of 15 in 2-MeTHF. This led to solubilization of
the resulting potassium chloride and a clear biphasic mixture.
Again, this has the benefit of a simple phase separation at the
end of reaction; the removal of the aqueous phase resulted in a
2-MeTHF solution of thioacetate 16 which can be used directly
in the subsequent cyclization without isolation.
Since the published cyclization chemistry used sodium
methoxide as a base,2 we envisaged an additional role for this
reagent in the concomitant deprotection (deacylation) of 16 to
give the free thiolate which could take part in reaction with 9
C
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 5. Proposed Mechanism for the Addition−Elimination and Base-Catalyzed Cyclization to 11
(Scheme 5). It became apparent that the order of addition is
not critical;8 it is possible to deacylate 16 in solution with
sodium methoxide and then add cinnamonitrile 9 to initiate
addition−elimination, followed by cyclization. Similarly, it is
possible to add solid 9 to the solution of thioacetate 16 first
at which point no reaction occursfollowed by dropwise
addition of sodium methoxide. It was the latter approach that
was chosen as a basis of safety for exotherm generation and
operational simplicity.
autocatalytic in base. This would be made possible due to the
fact that the proposed nitrile anion 21 is basic enough to
deprotonate the uncyclized compound 19, thus propagating the
catalytic cycle.9 The main impurity which resulted under a
variety of conditions was ∼2−3% of the uncyclized congener
19, as identified by LCMS. This impurity is adequately
removed during the efficient isolation protocol utilizing a
highly convenient, seeded, cooled antisolvent-based crystal-
lization directly from the reaction mixture.
Mechanistically, it is postulated that initial methanolysis of
the thioacetate group leads to a thiolate anion which takes part
in an addition−elimination (Scheme 5). In initial investigations,
a large excess of sodium methoxide (3.5 equiv) was used to
complete the cyclization. This was accompanied by significant
byproduct formation (ca. 1% by area), identified by mass
spectrometry as methoxy impurity 17, later confirmed by
synthesis. We reasoned that a smaller sodium methoxide charge
may minimize the formation of this byproduct, and indeed, in
laboratory studies the reaction proceeded to 11 in >97%
conversion, with as little as 1.25 equiv of sodium methoxide
(wrt 5)significantly lower than the theoretical requirement of
2.0 equiv of base. To improve the conversion rate and
robustness, we finally selected 1.5 equiv of sodium methoxide
(wrt 5) and utilized a slight undercharge of cinnamonitrile 9
(0.9 equiv wrt 5) to minimize the impurity 17. Gratifyingly, the
result was clean conversion to 11 within 1 h at ambient
temperature with no detectable 17. As can be seen from the
proposed mechanism (Scheme 5), after the initial stoichio-
metric deacylation the anionic cyclization itself is potentially
In summary, this operationally simple sequence of trans-
formations constitutes an ambient temperature, one-pot, four-
stage telescope in which chiral amine 5 is transformed into
highly elaborate thiophene intermediate 11 in 71% overall
yield. This approach resulted in the manufacture of 15 kg of
aminothiophene 11 in excellent purity (>97% w/w), with a
significantly reduced cycle time and environmental burden
when compared to the first-generation route.
Synthesis and Delivery of AZD7762. Completion of the
synthetic sequence involved the elaboration to urea 8 in a two-
stage protocol using trichloroacetyl isocyanate (TCANCO),
followed by methanolysis of the trichloroacetyl function. An
alternative direct approach to the synthesis of 8 using
potassium cyanate proved moderately successful but led to a
poorer impurity profile and was therefore discounted. Finally,
the removal of the BOC group with HCl and basification with
triethylamine followed by a controlled water addition resulted
in AZD7762 in good physical form. It should be noted,
however, that during manufacture, prior to crystallization of the
product, it was necessary to install a Harborlite filtration to
D
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 6. Final Synthesis of AZD7762 (Including Yields at Kilolab Scale)
remove traces of a hazy residue. The resulting physical losses
reduced the yield to 78% from the expected >90%. Pleasingly,
investigations confirmed that this residue was not related to the
desired chemistry, but derived from a contaminant in
commercial supplies of TCANCO.10 Its level can therefore
be controlled on specification for future manufacture, removing
the need for harborlite filtration. After the subsequent planned
recrystallization from aqueous methanol, 7 kg of AZD7762 was
produced in good yield and excellent purity (Scheme 6).
temperature at 39 to 41 °C during the addition. The resulting
reaction mixture was stirred at 40 °C for 1 h before analysis for
conversion to iminium 13 by HPLC.
To the resulting reaction mixture was added a solution of
hydroxylamine hydrochloride (45.17 g, 0.637 mol) in N,N-
dimethylformamide (240 mL) dropwise, maintaining the
temperature at 39−45 °C during the addition, followed by a
line-wash of N,N-dimethylformamide (40 mL). After stirring at
40 °C for 15 min, the reaction mixture was sampled for
conversion to 9 before cooling to 15−20 °C and addition of
water (800 mL) dropwise, maintaining the temperature
between 17 and 21 °C. The reaction mixture was then cooled
to 5 °C and stirred at this temperature for a further 20 min
before filtration of the solid, displacement washing with two
separate portions of water (2 × 240 mL), and drying at 40 °C
to afford the title compound as a pale yellow solid (72.16 g,
69% yield). This process was operated in 4 batches of up to 3.3 kg
input 3-fluoroacetophenone each, generating a total of 10.1 kg of 9.
The average yield was 65%.
CONCLUSION
■
In summary a new, highly convergent multikilo synthesis of
AZD7762 has been achieved. This involved two telescoped
sequences, including the key cyclization which uses newly
developed chemistry to generate the thiophene skeleton with
the desired 2,3,5-substitution unambiguously installed. This has
led to 5-fold reduction in the required quantity of chiral
piperidine 5 and an improved overall yield of 49% (versus first
generation route, 9%). In addition, the synthesis is now free of
toxic metals, has simple solid isolations directly from the
reaction mixtures, and requires no chromatography, producing
an API of high purity for clinical development.
1H NMR (400 MHz, DMSO-d6, 300 K): 7.72−7.65 (m, 2H),
7.63−7.56 (m, 1H), 7.49−7.42 (m, 1H), 7.03 (s, 1H). 13C
NMR (100 MHz, DMSO-d6, 300 K): 162.0 (d, J = 245 Hz),
149.3 (d, J = 3 Hz), 135.6 (d, J = 8 Hz), 131.1 (d, J = 9 Hz),
123.3 (d, J = 3 Hz), 118.8 (d, J = 21 Hz), 115.8 (s), 113.8 (d, J
= 24 Hz), 99.3 (s). GC-HRMS Calcd for [M] C9H5NFCl:
181.0095; found [M]+: 181.0090.
(S)-3-{[3′-Amino-5′-(3″-fluorophenyl)thiophene-2′-
carbonyl]amino}-piperidine-1-carboxylic Acid tert-Butyl
Ester (11). Piperidine 5 (120.0 g, 0.599 mol) was dissolved in
2-methyltetrahydrofuran (540 mL). Pyridine (58.14 mL, 0.719
mol) was added, followed by a line-wash of 2-methyltetrahy-
drofuran (60 mL). Chloroacetyl chloride (55.32 mL, 0.689
mol) was added dropwise, maintaining the temperature at 21−
25 °C, followed by a line wash of 2-methyltetrahydrofuran (60
mL). After 2.5 h at ambient temperature, the reaction mixture
was sampled for conversion to 15 by GC before the addition of
a 16% (w/w) aqueous solution of sodium chloride (360 mL).
The mixture was stirred for 30 min before separating off the
aqueous phase.
EXPERIMENTAL SECTION
■
General. Starting materials, reagents, and solvents were
obtained from commercial suppliers and used without further
purification. HPLC analyses were performed with an Agilent
1100 instrument. Intermediate purities were measured by
quantitative NMR analysis using an internal standard of known
purity, typically maleic acid or 2,3,5,6-tetrachloronitrobenzene.
Accurate mass GC-MS or LC-MS methods were used to
confirm elemental formulas. Yields quoted (mass and
percentage) are corrected for purity. The processes described
below are taken from the laboratory process descriptions used
as for kilolab manufacture. A description of the scale on which
these processes were operated is given for each stage.
(Z)-3-Chloro-3-(3′-fluorophenyl)-acrylonitrile (9). To a
solution of 3-fluoroacetophenone (80.0 g, 0.579 mol) in N,N-
dimethylformamide (560 mL) at 40 °C was added phosphoryl
chloride (92.50 mL, 0.990 mol) dropwise, maintaining the
E
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
To the resulting organic phase was added a filtered solution
of potassium thioacetate (102.65 g, 0.899 mol) in water (204
mL), followed by a line-wash of water (36 mL), maintaining the
temperature at 19−26 °C throughout. After stirring overnight
at ambient temperature, the organic phase was sampled for
conversion to 16 by HPLC before separating off the aqueous
phase.
(s), 27.7 (s), 23.2 (s). LC-HRMS Calcd for [M + Na]
C24H26Cl3FN4NaO5S: 629.0566; found [M + Na]: 629.0561.
(S)-3-{[5′-(3″-Fluorophenyl)-3′-ureido-thiophene-2′-car-
bonyl]-amino}-piperidine-1-carboxylic Acid tert-Butyl Ester
(8). To a suspension of 22 (101.45 g, 0.169 mol) in methanol
(516 mL) was added triethylamine (58.15 mL, 0.417 mol).
After 2.5 h at ambient temperature, the mixture was sampled by
HPLC for conversion to 8 before the addition of water (206
mL) over 10 min. After stirring overnight at ambient
temperature, the reaction mixture was heated to 45 °C for 15
min before addition of a second portion of water (1083 mL)
over 2 h. After a further 1 h at 45 °C, the reaction mixture was
allowed to cool to 20 °C and stirred at this temperature for 1 h.
The reaction mixture was filtered, and the solid was washed
with water (206 mL) before drying at 40 °C to afford 8 as a
white solid (75.79 g, 98% yield). This process was operated in 5
batches of up to 4.8 kg input of 22 each, generating a total of 14.2
kg of 8 in quantitative yield.
To the organic phase was added solid 9 (97.93 g, 0.539 mol)
before dropwise addition of a solution of sodium methoxide in
methanol (202 mL @ 25% w/w, 0.899 mol), maintaining the
temperature at 21−24 °C. This was followed by a line wash of
methanol (36 mL). After stirring for 1 h 50 min at ambient
temperature, the reaction mixture was sampled by HPLC for
conversion to 11 before heating to 33 °C, followed by dropwise
addition of water (600 mL). After stirring for 10 min, the
aqueous phase was separated off. To the organic phase was
added isohexane (960 mL) dropwise before removing a small
sample of the reaction mixture, allowing it to cool and
crystallize before returning it to the bulk mixture to seed
crystallization. Dropwise addition of a second portion of
isohexane (480 mL), followed by cooling to 3 °C over 1 h, and
stirring at this temperature overnight caused crystallization of
the product. Filtration, displacement washing the solid with ice-
cold tert-butyl acetate (240 mL), and 2 × ice-cold mixed
solvent system of tert-butyl acetate and isohexane (1:1, 2 × 240
mL) and drying at 40 °C over 3 days afforded 11 as a pale
yellow solid (176.70 g, 70% yield based on 5). This process was
operated in 3 batches of up to 4.8 kg input of 5 each, generating a
total of 15.3 kg of 11. The average yield was 71%.
1H NMR (400 MHz, DMSO-d6, 353 K): 9.86 (s, 1H), 8.24
(s, 1H), 7.60−7.41 (m, 3H), 7.41−7.33 (m, 1H), 7.22−7.15
(m, 1H), 6.36 (br. s, 2H), 3.94−3.68 (m, 3H), 2.97−2.79 (m,
2H), 1.94−1.84 (m, 1H), 1.76−1.55 (m, 2H), 1.47−1.34 (m,
10H). 13C NMR (100 MHz, DMSO-d6, 353 K): 163.6 (s),
163.1 (d, J = 244), 155.4 (s), 154.6 (s), 145.5 (s), 142.3 (d, J =
2 Hz), 136.0 (d, J = 8 Hz), 131.8 (d, J = 9 Hz), 122.2 (d, J = 3
Hz), 120.0 (s), 115.9 (d, J = 22 Hz), 112.6 (d, J = 23 Hz),
110.5 (s), 79.2 (s), 48.3 (s), 46.5 (s), 44.1 (s), 30.1 (s), 28.6
(s), 24.1 (s). LC-HRMS Calcd for [M + H] C22H28FN4O4S:
463.1810; found [M + H]+: 463.1815.
5-(3′-Fluorophenyl)-3-ureidothiophene-2-carboxylic Acid
(S)-Piperidin-3″-ylamide (AZD7762). To a suspension of 8
(75.3 g, 0.163 mol) in methanol (383 mL) was added an
aqueous solution of hydrochloric acid (40.78 mL @ 37% w/w
in water, 0.488 mol) dropwise, maintaining the temperature at
20 to 30 °C. The resulting reaction mixture was heated at 50 °C
for 4 h before analysis by HPLC for conversion to AZD7762.
Triethylamine (85.10 mL, 0.610 mol) was added dropwise
before addition of water (345 mL). A small sample of the
reaction mixture was then removed, allowing it to cool and
crystallize before returning to the bulk mixture in order to seed
crystallization. After stirring for 30 min, water (613 mL) was
added over 1.5 h before stirring at 50 °C for a further 30 min
and allowing to cool to 20 °C with stirring overnight. The
reaction mixture was filtered and the solid washed with water
(153 mL) before drying at 40 °C to afford AZD7762 as a white
solid (56.6 g, 96% yield). This process was operated in 4 batches
of up to 4.3 kg input of 8 each, generating a total of 8.2 kg of
AZD7762; the average yield was 78%. [Note: The yield was
lower than expected due to the Harborlite treatment which was
introduced to remove a haze generated from a contaminant in
TCANCO, a material introduced in the previous stage.10 The
Harborlite treatment successfully removed the impurity and
afforded a typical quality product.] Spectroscopic data are
reported after the final recrystallization.
1H NMR (400 MHz, DMSO-d6, 353 K): 7.49−7.32 (m, 3H),
7.19−7.12 (m, 1H), 7.01 (s, 1H), 6.91 (d, 1H), 6.29 (br.s, 2H),
3.91−3.64 (m, 3H), 2.96−2.77 (m, 2H), 1.92−1.77 (m, 1H),
1.74−1.30 (m, 12H). 13C NMR (100 MHz, DMSO, 353 K):
164.3 (s), 163.1 (d, J = 245), 154.6 (s), 153.9 (s), 142.2 (d, J =
3 Hz), 136.1 (d, J = 8 Hz), 131.6 (d, J = 9 Hz), 121.9 (d, J = 3
Hz), 118.7 (s), 115.6 (d, J = 21 Hz), 112.4 (d, J = 23 Hz),
102.5 (s), 79.2 (s), 48.6 (s), 46.1 (s), 44.1 (s), 30.3 (s), 28.6
(s), 24.1 (s); LC-HRMS calcd for [M + H] C21H27FN3O3S:
420.1752; found [M + H]+: 420.1748.
(S)-3-({5′-(3″-Fluorophenyl)-3′-[3‴-(2⁗,2⁗,2⁗-
trichloroacetyl)ureido]thiophene-2′-carbonyl}-amino)-piper-
idine-1-carboxylic Acid tert-Butyl Ester (22). To a solution of
11 (73.12 g, 0.174 mol) in tetrahydrofuran (800 mL) was
added trichloroacetyl isocyanate (23.23 mL, 0.196 mol),
maintaining the temperature at 20−30 °C during the addition.
After 2.5 h at ambient temperature, the mixture was sampled by
HPLC for conversion to 22 before the addition of isohexane
(1120 mL) dropwise over 1 h. After stirring for a further 1 h,
the reaction mixture was filtered; the solid was washed with
isohexane (160 mL) and dried at 40 °C to afford 22 as a pale
peach solid (101.88 g, 96% yield). This process was operated in 5
batches of up to 3.8 kg input of 11 each, generating a total of 19.7
kg of 22. The average yield was 95%.
1H NMR (400 MHz, DMSO-d6, 353 K): 11.70 (s, 1H),
11.49 (br. s, 1H), 8.24 (s, 1H), 7.80 (d, 1H), 7.57−7.40 (m,
3H), 7.26−7.18 (m, 1H), 3.97−3.67 (m, 3H), 2.95−2.78 (m,
2H), 1.97−1.84 (m, 1H), 1.78−1.53 (m, 2H), 1.51−1.33 (m,
10H). 13C NMR (100 MHz, DMSO-d6, 343 K): 162.3 (d, J =
245 Hz), 161.7 (s), 160.3 (s), 153.7 (s), 148.5 (s), 141.9 (d, J =
3 Hz), 140.5 (s), 134.6 (d, J = 8 Hz), 131.1 (d, J = 9), 121.4 (d,
J = 3 Hz), 119.5 (s), 115.3 (d, J = 21 Hz), 114.7 (s), 112.0 (d, J
= 23 Hz), 91.8 (s), 78.4 (s), 47.4 (s), 45.7 (s), 43.2 (s), 29.2
Recrystallization of AZD7762. AZD7762 (50.0 g, 0.138
mol) in methanol (650 mL) was heated to 30 °C for 30 min
before filtering through a 1.6 μm glass microfiber filter paper
into a second vessel, followed by a line-wash with methanol
(100 mL). The resulting solution was cooled to 10 °C before
the addition of water (250 mL) over 20 min, maintaining the
temperature at 10−15 °C. The reaction was then seeded with
AZD7762 (150 mg, 0.3% w/w), and the contents of the vessel
were allowed to stir at 10 °C for 30 min. The addition of a
F
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
However, this impurity is completely purged in the crystallization
liquors and did not pose any issue on scale-up.
second portion of water (500 mL) over 1 h 30 min, maintaining
the temperature 10−13 °C, followed by stirring for 20 h at 10
°C, completed the crystallization. Filtration, washing with water
(2 × 100 mL), and pulling dry on the filter for 30 min before
drying under vacuum at 40 °C afforded AZD7762 as a white
solid (46.16 g, 92% yield). This process was operated in 3 batches
of up to 2.6 kg of AZD7762 each, generating a total of 6.9 kg of
pure AZD7762. The average yield was 93%.
1H NMR (400 MHz, DMSO-d6, 353 K): 9.88 (br. s, 1H),
8.22 (s, 1H), 7.52−7.36 (m, 4H), 7.19 (m, 1H), 6.35 (br. s,
2H), 3.81 (m, 1H), 2.95 (m, 1H), 2.76 (m, 1H), 2.44−2.56 (m,
2H), 1.82 (m, 1H), 1.67−1.34 (m, 3H). 13C NMR (100 MHz,
DMSO-d6, 353 K): 163.3 (s), 163.1 (d, J = 243 Hz), 155.5 (s),
145.2 (s), 142.0 (d, J = 3 Hz), 136.0 (d, J = 3 Hz), 131.8 (d, J =
9), 122.2 (d, J = 3 Hz), 120.1 (s), 115.8 (d, J = 21 Hz), 112.6
(d, J = 23 Hz), 111.0 (s), 51.6 (s), 47.4 (s), 46.3 (s), 30.9 (s),
25.5 (s). LC-HRMS Calcd for [M + H] C17H20FN4O2S:
363.1286; found [M + H]+: 363.1277.
(7) Buchwald, S. L.; Pellegatti, L. Org. Process Res. Dev. 2012, 16,
1442.
(8) Ball, M.; Jones, M. F.; Kenley, F. R.; Pittam, D. J. PCT Int. Appl.
WO2009133389, 2009.
(9) A similar autocatalytic process is implicated in the synthesis of
benzyl trichloroacetimidate from benzyl alcohol, trichloroacetonitrile,
and catalytic sodium hydride; see: Wessel, H. P.; Iversen, T.; Bundle,
D. R. J. Chem. Soc., Perkin Trans. 1 1985, 11, 2247−50. and Cramer,
F.; Pawelzik, K.; Baldauf, H. J. Chem. Ber. 1958, 91, 1049.
(10) The Harborlite filtration was necessary to remove small
amounts of a hazy white solid from the reaction mixture and resulted
in a lower yield than expected (78% vs > 90% during development)
presumably due to product retention on the filter. The unexpected
solid was later assigned as the oxalyl bridged dimer 24 (by LCMS),
thought to be derived from residual oxalyl chloride present in
commercial trichloroacetyl isocyanate (TCANCO). We would like to
highlight here that residual oxalyl chloride would be a common
problem for many uses of TCANCO and should be analyzed for if
thought to pose a significant risk to product purity.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.oprd.6b00364.
■
S
1H and 13C NMR spectra for compounds 9, 11, 22, 8,
and AZD7762 (PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail: matthew.ball@astrazeneca.com.
■
ORCID
Matthew Ball: 0000-0002-5519-0118
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors would like to thank Mark Harrison and Paul Davey
for mass spectrometry, Karen Rome for analysis, and Lyn
Powell and Andrew Philips for helpful discussions.
REFERENCES
■
(1) Ashwell, S.; Gero, T.; Ioannidis, S.; Janetka, J.; Lyne, P.; Su, M.;
Toader, D.; Yu, D.; Yu, Y. P. Int. Appl. WO2005066163, 2005.
(2) (a) Hartmann, H.; Liebscher, J. Synthesis 1984, 1984, 275−276.
(b) Andersen, H. S.; Olsen, O. H.; Iversen, L. F.; Sørensen, A. L. P.;
Mortensen, S. B.; Christensen, M. S.; Branner, S.; Hansen, T. K.; Lau,
J. F.; Jeppesen, L.; Moran, E. J.; Su, J.; Bakir, F.; Judge, L.; Shahbaz, M.;
Collins, T.; Vo, T.; Newman, M. J.; Ripka, W. C.; Møller, N. P. H. J.
Med. Chem. 2002, 45, 4443−4459.
(3) During the manufacturing campaign, a complementary laboratory
scale (∼1 g) synthesis of 2-(1)-amido-3-amino-5-aryl thiophenes was
reported. This work uses an altered order of reactivity and requires
disodium disulfidea less handlable, more toxic source of sulfurand
higher temperatures than our approach; see: Lassagne, F.; Pochat, F.;
Sinbandhit, S. Synth. Commun. 2007, 37 (7), 1133−1140.
(4) (a) Arnold, Z.; Zemlicka, J. Collect. Czech. Chem. Commun. 1959,
24, 2385. (b) Liebscher, B.; Neumann, B.; Hartmann, H. J. Prakt.
Chem. 1983, 325, 915.
(5) N-Methyl pyrrolidone was also found to be a potential solvent for
hydroxylamine hydrochloride, although this caused complications in
the final crystallization and no further work was progressed.
(6) The reaction also generates ∼25% of the enamine impurity 23,
formed from double addition of the Vilsmeier reagent. This was found
to be remarkably consistent, regardless of reaction conditions.
G
DOI: 10.1021/acs.oprd.6b00364
Org. Process Res. Dev. XXXX, XXX, XXX−XXX