Decoration of an α-Resorcylate Nucleus as Part of the Manufacture of a Glucokinase Activator

Article abstract of DOI:10.1021/acs.oprd.8b00175

The need to define a set of processes for the manufacture of a glucokinase activator called for an evaluation of different strategies to differentiate the hydroxyls of an α-resorcylic acid derivative. While direct functionalization proved possible, it did not allow access to crystalline intermediates that offered control over the rejection of process impurities. The strategy taken forward involved the installation of a benzoyl protecting group using careful control of pH in order to achieve useful levels of selectivity. This allowed the remaining α-resorcylate hydroxyl to be functionalized using a Mitsunobu reaction, with liquid-liquid partitioning being used to separate downstream intermediates of interest away from the redox byproducts of this reaction. Downstream challenges that were overcome in order to deliver a commercially viable means of manufacturing the API included developing an amidation reaction with a poorly reactive aminopyrazine coupling partner.

Full text of DOI:10.1021/acs.oprd.8b00175

Subscriber access provided by Kaohsiung Medical University
Full Paper
Decoration of an #-Resorcylate Nucleus as part
of the Manufacture of a Glucokinase Activator
Phillip Hopes, Thomas Langer, Kirsty Millard, and Alan Steven
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.8b00175 • Publication Date (Web): 13 Jul 2018
Downloaded from http://pubs.acs.org on July 15, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
Decoration of an α-Resorcylate Nucleus as part of the Manufacture of a
Glucokinase Activator
8
9
Phillip Hopes,† Thomas Langer,‡ Kirsty Millard,‡ and Alan Steven*,‡
Cyton Biosciences Ltd, 68 Macrae Road, Bristol, BS20 0DD, United Kingdom
Pharmaceutical Technology & Development, AstraZeneca, Charter Way, Macclesfield, SK10 2NA,
United Kingdom
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
†
‡
Alan.Steven@astrazeneca.com
1
ACS Paragon Plus Environment

Organic Process Research & Development
Page 2 of 35
1
2
3
4
TOC graphic goes here
5
6
7
8
9
N
N
Me
O
O
O
O
HO
O
MeO
N
OMe
MeO
OH
N
H
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(64%)
(62%)
N
N
O
OH
OH
monofunctionalization of
benzenoid core using
protecting group approach
AZD1656
506 kg
O
99.5% area, 600 kg
≥
≈
2
ACS Paragon Plus Environment

Page 3 of 35
Organic Process Research & Development
1
2
3
4
ABSTRACT
5
6
7
8
9
The need to define a set of processes for the manufacture of a glucokinase activator called for an
evaluation of different strategies to differentiate the hydroxyls of an αꢀresorcylic acid derivative. Whilst
direct functionalization proved possible, it did not allow access to crystalline intermediates that offered
control over the rejection of process impurities. The strategy taken forward involved the installation of a
benzoyl protecting group using careful control of pH in order to achieve useful levels of selectivity.
This allowed the remaining αꢀresorcylate hydroxyl to be functionalized using a Mitsunobu reaction,
with liquidꢀliquid partitioning being used to separate downstream intermediates of interest away from
the redox byproducts of this reaction. Downstream challenges that were overcome in order to deliver a
commercially viable means of manufacturing the API included developing an amidation reaction with a
poorly reactive aminopyrazine coupling partner.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
KEY WORDS
resorcylate, monobenzoylation, Mitsunobu, glucokinase activator, SNAr, amidation
INTRODUCTION
AZD1656 (1), a compound containing two pyrazine units, has been under development by AstraZeneca
as a glucokinase activator for the treatment of type 2 diabetes.1ꢀ4 Investigations around synthetic route
design for AZD1656 (1) led to the decision to route via acid 2, and to elaborate it to the API through
couplings with chloropyrazine 35 and aminopyrazine 4 (Scheme 1),6 compounds whose development
activities have been disclosed recently. Access to acid 2, an αꢀresorcylate derivative bearing a sideꢀ
chain with a stereogenic center, required a means of differentiating between the hydroxyl groups of
methyl αꢀresorcylate (5), our starting material of choice. This article will discuss how this challenge
was overcome, as part of a wider discussion around the selection of the synthetic route, the development
3
ACS Paragon Plus Environment

Organic Process Research & Development
Page 4 of 35
1
2
3
4
of manufacturable processes and learning gained from the commercialꢀscale manufacture of AZD1656
5
6
7
(1).
8
9
Scheme 1. Disconnections of AZD1656 (1)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
2
1
3
4
MEDICINAL CHEMISTRY APPROACH
The chemistry utilized in order to achieve the selection of a candidate drug by the project is outlined in
Scheme 2. It differentiated the hydroxyl groups of the αꢀresorcylate core by dibenzylating methyl αꢀ
resorcylate (5) before hydrogenolyzing this material using exactly one equivalent of hydrogen gas. This
protocol produced ether 6 contaminated by methyl αꢀresorcylate (5) and residual dibenzylated material,
both of which were removed by chromatographic purification. The chiral sideꢀchain was incorporated
by engaging phenol 6 in a Mitsunobu etherification with (R)ꢀ1ꢀmethoxypropanꢀ2ꢀol (7) so as to afford
ether 8. The API was duly accessed, after the required protecting group removals, through the
sequential appendage of aminopyrazine 4 to acid 9 so as to afford amide 10, and chloropyrazine 3 to
phenol 11. As well as being stepꢀheavy, a consequence of the use of protecting groups to differentiate
between the αꢀresorcylate hydroxyls, the intermediates associated with the medicinal chemistry route
included a number of oils and lowꢀmelting solids whose purification required chromatography.
Scheme 2 Medicinal Chemistry Approach to AZD1656 (1)
ACS Paragon Plus Environment

Page 5 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
CO Me
O
CO Me
O
CO H
2
HO
2
2
MeO
MeO
a
b
(60 %)
(99 %)
OBn
OBn
OBn
6
8
9
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
N
O
O
O
O
MeO
N
H
N
c
e
MeO
N
N
H
(81 %)
(83 %)
10 R = Bn
AZD1656 (1)
N
N
O
d
(92%)
OR
11 R = H
N
O
a) 7, PPh , DIAD, THF; b) 2 M NaOH, MeOH; c) (COCl) , DMF, CH Cl then 4, C H N, 0 C to RT; d) H , 10
°
3
2
2
2
5
5
2
%wt/wt, Pd-C, ethanol, THF; e) 3, Cs CO , MeCN, 120 C (microwave)
°
2
3
N
Cl
N
N
OH
MeO
N
N
H2N
O
3
4
7
SYNTHESIS STRATEGY
When the project entered development, the retention of the αꢀresorcylate core, two pyrazines and a
chiral sideꢀchain as the building blocks of the synthesis was appealing. The use of acid 12 (Figure 1) as
the synthetic precursor to the API held attraction on the grounds that it was known to be a highly
crystalline material whose isolation could be used as a lateꢀstage control point. Regardless, in the spirit
of building up knowledge around alternative route options in early chemical development, exploratory
studies that involved the early introduction of aminopyrazine 4 were undertaken. Its amidation with an
αꢀresorcylate unit with at least one unprotected hydroxyl quickly floundered. The activated carbonyl
functionality of the αꢀresorcylate preferred to react with another such unit to form ester impurities,
rather than with the poorly nucleophilic amine of aminopyrazine 4.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 6 of 35
1
2
3
4
5
6
7
8
9
12
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1 Synthetic precursor to API
INITIAL α-RESORCYLATE FUNCTIONALIZATION STUDIES
Given the intent to use the amidation of aminopyrazine 4 with acid 12 for the API bondꢀforming step,
efforts turned to the identification of an efficient synthesis of the latter. The use of a 3,5ꢀ
dihalobenzonitrile starting material, as an alternative to methyl αꢀresorcylate (5), was briefly considered.
Successive nucleophilic displacements on a compound such as 13 should differ greatly in their
respective rates, allowing the 3ꢀ and 5ꢀ positions to be successfully differentiated. This approach was
tested on material that was associated with a structurallyꢀrelated drug candidate, alcohol 14 (Scheme 3).7
Whilst the initial SNAr to give ether 15 was successful, the attempted union of pyrazinone 16 and aryl
bromide 17 afforded a mixture of ethers 18 and 19 in which the ambidentate nucleophile had
preferentially reacted through nitrogen, rather than the pyrazinone oxygen atom.
Scheme 3 Attempted derivatization of a 3,5ꢀdihalobenzonitrile 13
ACS Paragon Plus Environment

Page 7 of 35
Organic Process Research & Development
1
2
3
4
5
6
a
14
b
7
8
(100%)
(61%)
9
15
ArBr 17
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
13
c
16
18
19
1
:
3
a) NaHMDS, DMF, 23 C; b) NaOH (aq.), EtOH, then HCl (aq.); c) 16, Cs CO , CuI, 2,2,6,6-
°
ꢀ
2
3
tetramethylheptane-3,5-dione, NMP, 125 C then HCl (aq.)
°
Attention returned to methyl αꢀresorcylate (5) as a starting material and the search for a selective
monoalkylation that would allow the chiral sideꢀchain to be introduced.8ꢀ9 In spite of the screening of
different bases and solvents, alkylations with tosylate 20 (prepared from alcohol 73a,10) suffered from
poor conversions or selectivity (monoalkylation vs bisalkylation).11 The same problem was observed
with an assessment of a Mitsunobu etherification option. The propensity for a monoalkylated derivative
like ether 21 to react again, before all of the methyl αꢀresorcylate (5) has been consumed may be
explained by the similarity in the pKa values of the latter and ether 21 (Scheme 4).
Scheme 4 Calculated pKa values (aqueous, T=25 °C, zero ionic strength)12 relevant to monoalkylation
approach
ACS Paragon Plus Environment

Organic Process Research & Development
Page 8 of 35
1
2
3
4
5
pKa1 = 8.6 0.1
pKa2 = 10.2 0.1
±
±
6
7
5
24
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
20
pKa = 8.7 0.1
±
21
A monoarylation approach was studied in parallel with the alkylation studies described above, and
started with the use of potassium carbonate as base and chloropyrazine 3 as electrophile. In addition to
the risk of overarylation to form bisether 22 (Scheme 5), in an analogous fashion to that described
above, these efforts were soon found to be complicated by the ability of the newlyꢀinstalled pyrazine of
ether 23 to intermolecularly migrate under the basic conditions of the reaction. This led to the
regeneration of a quantity of bisether 22 (now via a mechanism distinct from overreaction) and methyl
αꢀresorcylate (5).3c Gratifyingly, prestirring the methyl αꢀresorcylate (5) with excess carbonate base,
prior to the addition of the electrophile led to useful selectivities, by allowing access to the more basic
(and presumably more nucleophilic) αꢀresorcylate dianion 24 (Scheme 5). After some optimization
which included switching to the use of cesium carbonate as base, the dosed addition of a slight subcess
(0.97 molar equivalents) of chloropyrazine 3 to a solution of methyl αꢀresorcylate (5) in dimethyl
sulfoxide led to a useful level of selectivity for ether 23 versus bisether 22 (Scheme 5). The intention
was then to alkylate the remaining αꢀresorcylate hydroxyl through a Williamson etherification with
tosylate 20, so as to form ether 25. The screening of solvents, carbonate bases and strong organic bases
led to the selection of the use of cesium carbonate in DMSO for the etherification. A drawback of the
telescoped use of these basic conditions was disruption of the equilibrium shown in Scheme 5, resulting
in scrambling of the recently installed pyrazine moiety. This migration led to an increase in the amount
ACS Paragon Plus Environment

Page 9 of 35
Organic Process Research & Development
1
2
3
4
5
6
of bisether 22 in the mixture (to 12% area by HPLC (LCAP)). The migration also liberated methyl αꢀ
resorcylate (5), opening the door to the formation of another bisether 26.13
7
8
9
Scheme 5 Derivatization of methyl αꢀresorcylate (5) using an initial monoarylation.
O
O
O
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HO
pKa1 = 8.6 0.1 –O
–O
pKa2 = 10.2 0.1
±
±
OMe
OMe
OMe
24
5
–
OH
OH
O
a) i-ii
99.6 area%
conversion
O
O
HO
OMe
O
O
23
Ar
O
O
Ar O
+
a) iii-iv
OMe
MeO
OMe
Ar
O
25
22
OMe
22
OAr
O
Ar
+
59% isolated yield
12 LCAP
Ar O
CO Me
O
2
MeO
23
95.4
:
:
22
4.2 area%
O
MeO
26
a) i) Cs CO (2.5 eq.), DMSO, 50 C, ii) 3 (0.97 eq.), dosed addition, iii) 20, 80 C,
°
°
2
3
iv) chromatography
Even when pure, ether 25 was an oil preventing its purification through crystallization. After it had been
chromatographically purified from the mixture of compounds shown in Scheme 5, ether 25 was
successfully hydrolyzed to acid 12, using careful control of the reaction temperature and time to
minimize hydrolysis of the baseꢀlabile azetidine amide and pyrazineꢀether bonds. The sequences shown
in Schemes 5 and 6 were leveraged in order to deliver 3.4 kg of AZD1656 (1).
Scheme 6 Completion of the first development delivery of AZD1656 (1)
ACS Paragon Plus Environment

Organic Process Research & Development
Page 10 of 35
1
2
3
4
5
6
a
b
AZD1656 (1)
7
8
(89 %)
(62 %)
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
25
12
a) i) NaOH (aq), NMP, 10 C, ii) AcOH (aq); b) 4, T3P, N-methylmorpholine, 2-MeTHF, reflux
−
°
LATER α-RESORCYLATE FUNCTIONALIZATION STUDIES
The attractiveness of compound 25 as an intermediate for the commercial route used to supply
AZD1656 (1) was diminished by its being an oil. Thus, our investigation of competing route options
was restarted and focused on the identification of possible crystalline intermediates. The αꢀresorcylate
studies detailed above were already known to generate materials that were mostly oils and so attention
turned to the generation of a crystalline material through the monoprotection of methyl αꢀresorcylate
(5). Such a material would be primed for the introduction of the chiral sideꢀchain through alkylation
before deprotection steps would reveal acid 2 (Scheme 1), a solid that was known to be isolable in spite
of its modest melting point (42 °C). Whilst a synthetic strategy which uses protecting groups has
inherent inefficiencies associated with it, it was hoped these could be mitigated by using a group that
could be removed alongside the hydrolysis of the methyl ester functionality of the αꢀresorcylate ring.
These considerations led to the targeting of the monobenzoylation of methyl αꢀresorcylate (5),9,14
a
known transformation and one that was reported to generate a solid product,15 ester 27 (Scheme 7),
whose crystallization could further enhance the impurity control strategy for the API. Gratifyingly, the
transformation was successful in our hands. The insolubility of ester 27 in the aqueous reaction medium
was no doubt responsible for the limited levels of the bisbenzoyl byproduct 28 observed. Maintaining
ACS Paragon Plus Environment

Page 11 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
the pH within a range of 7.8–8.2 was found to be critical. This was initially achieved by codosing
benzoyl chloride with a separate addition of sodium hydroxide buffered with sodium
dihydrogenphosphate, and restricted bisbenzoylation to ca. 6 LCAP. The use of aqueous solutions of
lithium hydroxide or potassium carbonate offered improvements, and their use in combination produced
a superior result still. Buffering with this mixture allowed ester 27 to be isolated with levels of methyl
αꢀresorcylate (5) and the overreaction product 28 that were both below 2 LCAP.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 7 Synthesis of acid 2 using a benzoyl protecting group strategy
a
b (or e)
(81 %)
R = Me 5
R = H 30
R = H 27
R = Bz 28
29
c
d
(76 % from 27
via Mitsunobu
route)
R' = Me 2
R' = H 31
21
a) BzCl, PhMe, LiOH (aq), K CO (aq); b) (R)-1-methoxypropan-2-ol (7), PPh , DIAD, PhMe, 0 C;
°
2
3
3
c) NaOMe, MeOH; d) KOH (aq); e) 20, Cs2CO3, DMSO
With the αꢀresorcylate 3ꢀ and 5ꢀsubstituents successfully differentiated, through either monoarylation or
monobenzoylation, there was now the need to etherify the remaining hydroxyl with the chiral sideꢀchain.
Whilst the retention of the Williamson ether synthesis remained an option, drawbacks included the
relative expense of cesium carbonate, concerns over its suspension given its high density, and the
negative impact on process mass intensity and throughput of washing out DMSO using multiple water
washes. Further caveats included its use of a potential genotoxic impurity, in the form of tosylate 20,
albeit one whose downstream purging meant that it did not affect the API quality, and benzoyl group
migration under the basic conditions of the reaction. Attention thus returned to the consideration of a
Mitsunobu reaction as a possible means of installing the sideꢀchain. This reaction has found limited
ACS Paragon Plus Environment

Organic Process Research & Development
Page 12 of 35
1
2
3
4
5
6
7
8
9
uses in largeꢀscale production due to the challenge of separating the reaction product from its redox
byproducts, a phosphine oxide and dialkyl azodicarboxylate.16ꢀ21 In our case, we were attracted by the
option of telescoping the product of the etherification into a step that would remove the benzoyl
protecting group. In unveiling the αꢀresorcylate hydroxyl, it was hoped that liquidꢀliquid partitioning
could then be used to draw it into an alkaline phase, away from the Mitsunobu byproducts.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
After some experimentation with the solvent and the order of addition of the reagents, it was found the
addition of diisopropyl azodicarboxylate to a toluene solution of the other components led to the rapid
and clean conversion of phenol 27 to ether 29 (Scheme 7). The reaction proved to be a faster and
higherꢀyielding than the Williamson etherification conditions against which it was being compared
(reaction condition e in Scheme 7), and proceeded without any scrambling of the benzoyl group. DSC
analysis showed there were no thermal stability issues with any of the process or waste streams
associated with the Mitsunobu reaction.22
Using a benzoyl protecting group to access acid 2 potentially posed the challenge of cleanly separating
benzoic acid from the compounds of interest arising from the removal of the two ester groupings of
compound 29 using aqueous base. With this in mind, a transesterification approach was instead targeted
for the debenzoylation. Gratifyingly, the addition of methanolic sodium methoxide, once the redox
byproducts of the Mitsunobu reaction had been filtered off, smoothly converted ester 29 to phenol 21
(Scheme 7). Curiously, any lingering compound 28, an impurity which would reform αꢀresorcylic acid
(30) on downstream processing, transesterified at a slower rate than ester 29. Thus, once the
transesterification was complete, careful extraction with dilute (0.25 M) potassium hydroxide solution
left the methyl esters of the αꢀresorcylate components intact, allowing the potassium salt of phenol 21 to
be partitioned away from compound 28, together with the methyl benzoate coproduct of the
transesterification, residual organicꢀsoluble redox byproducts from the Mitsunobu reaction, and bisether
26, a compound which arose from the reaction of the residual methyl αꢀresorcylate (5) in phenol 27.
ACS Paragon Plus Environment

Page 13 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
The addition of more potassium hydroxide basified the phase containing the phenol 21, allowing the
hydrolysis of its methyl ester so as to give acid 2 (Scheme 7). Trace levels of upstream impurities were
still present alongside acid 2, though they could be purged by pHꢀcontrolled liquidꢀliquid extractions,
ahead of the crystallization of 2. Thus, after a pH adjustment to 9.5–10.0, washing the aqueous phase
with tertꢀbutyl methyl ether removed any residual triphenylphosphine oxide and diisopropyl
azodicarboxylate. After further acidification to pH 1.3–1.5, the extraction of acid 2 with tertꢀbutyl
methyl ether reduced accompanying levels of αꢀresorcylic acid 30 (arising from the presence of
compound 28 in ester 29) to ≤0.2 LCAP.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The coupling of acid 2 and chloropyrazine 3 was assessed using a variety of solvents and bases. The
reaction proved highꢀyielding using cesium carbonate in DMSO. Whilst this combination of materials
brought with it the drawbacks outlined above for the Williamson etherification, we were confident that
these could be overcome with future process design studies. Consequently, the route that had been
initiated with the monobenzoylation of methyl αꢀresorcylate (5) had yielded a means of accessing acid
12, the precursor to the API.
ROUTE SELECTION
The existence of both the monobenzoylation approach and that which was initiated by the monoarylation
of methyl αꢀresorcylate (5) meant one option had to be discarded prior to initiating development for
commercialꢀscale production of the API. The route options are outlined in Scheme 8. In spite of the
monobenzoylation approach involving two more transformations (5) than the monoarylation option that
routed via ether 23, the selection of the former was straightforward. The monobenzoylation approach
featured a crystalline precursor (acid 2) to acid 12, highꢀyielding transformations, the ability to partition
away related substances and reaction byproducts through liquidꢀliquid extractions and the opportunity to
telescope the steps associated with the conversion of ester 27 to acid 2.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 14 of 35
1
2
3
4
Scheme 8 Summary of approaches to acid 2
5
6
7
O
CO Me
2
O
CO Me
O
CO H
2
2
MeO
MeO
MeO
8
9
OH
OBz
OH
crystalline
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CO Me
HO
oil
oil
21
29
2
OBz
crystalline
27
CO Me
HO
2
CO Me
CO Me
CO H
2
HO
O
O
2
2
OH
MeO
N
MeO
5
N
N
O
N
N
O
N
N
O
N
N
O
O
O
oil
23
oil
crystalline
12
25
Whilst the monoarylation approach featured fewer chemical transformations than the monobenzoylation
approach, this came at the expense of routing via oils. Another consideration was that the hydrolysis of
the ester functionality of compound 25, so as to afford acid 12, was still going to be challenging to
achieve whilst avoiding significant levels of hydrolysis of the azetidine amide and Cpyrazine–O bonds.
This also ruled out converting compound 21 of the monobenzoylation approach to acid 12 via ether 25
(dotted arrow in Scheme 8).
SCALEUP OF MANUFACTURE OF ACID 2
Not unexpectedly, given the presence of multiple phases, and the potential for bisbenzoylation, the
successful scaleup of the monobenzoylation reaction was envisaged as being challenging.23 Indeed,
initial pilot plant batches produced ester 27 with up to 20 LCAP of overreaction product 28. This was
ACS Paragon Plus Environment

Page 15 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
partly attributed to pH measurements used to control the reagent additions, not aligning with the local
pH at the site of the reaction. After assessing different locations of the reactor pH probe in an attempt to
achieve a correlation with at line measurements, the attachment of the pH probe to the reactor’s
sampling device was selected and used to control the rate of reagent addition. Dosing the benzoyl
chloride as a solution in toluene was found to attenuate its background hydrolysis to benzoic acid and to
make the control of pH easier. This had to be balanced against the increasing tendency of ester 27 to
coalesce from fine particles into a thick, unstirrable mixture as the amount of toluene used was
increased. A path forward was achieved by limiting the amount of toluene (54% wt/wt) used to dilute
the benzoyl chloride and the length of time after the end of the addition before the product was filtered
off. With these modifications in place, a normalized conversion of methyl αꢀresorcylate (5) to ester 27
of 96 LCAP could be typically achieved whilst restricting the amount of overreaction product 28 to
about 2 LCAP. Prolonged vacuum drying failed to reduce the water content of the initially isolated ester
27 below ∼0.5% wt/wt. This level required better control to ensure ester 27 was reproducibly consumed
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
in the ensuing Mitsunobu reaction, thus limiting the amount of αꢀresorcylic acid equivalent that could
potentially track through the rest of the synthesis. This control was achieved and the purity improved by
treating a solution of the initially isolated ester 27 with powdered cellulose before it was crystallized
from a mixture of toluene and isopropyl acetate. In this way, 1.02 metric tons of ester 27 with a water
content of <0.1% wt/wt were manufactured over four batches, and in an overall yield of 81%. At the
end of the Mitsunobu reaction, the addition of 0.07% mol/mol seed induced the crystallization of 75% of
the theoretical amount of both triphenylphosphine oxide and diisopropyl hydrazineꢀ1,2ꢀdicarboxylate as
a readily filtered complex (1:1 mol/mol). The further elaboration of the toluene solution of ether 29
proved straightforward, using the chemistry shown in Scheme 7. In this way, 619 kg of acid 2 were
made with an αꢀresorcylic acid (30) content of only 0.1 LCAP and in an overall yield of 62% from
methyl αꢀresorcylate (5). The purity across four batches was 99.5–99.7 LCAP with the largest impurity
ACS Paragon Plus Environment

Organic Process Research & Development
Page 16 of 35
1
2
3
4
5
6
7
8
9
(ca. 0.20 LCAP) being the analog 31 arising from the presence of 2ꢀmethoxyethanol in the (R)ꢀ1ꢀ
methoxyꢀ2ꢀpropanol (7) starting material (Scheme 7). Despite its reputation of generating a lot of waste,
it should be noted that, in spite of development efforts, the sequence that utilized a Williamson
etherification (reaction condition e in Scheme 7) had a much less favourable process mass intensity
(207.4 kg/(kg 2)) at the point where it was supplanted by the Mitsunobu route (69.6 kg/(kg 2)).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
SNAr REACTION DEVELOPMENT
A large amount of process development was performed on the stages associated with the conversion of
methyl αꢀresorcylate (5) to acid 2 in the period leading up to route selection. By contrast, the
development of the conversion of acid 2 to AZD1656 (1) had been developed only enough to derisk the
continual manufacture of clinical supplies. With the route selected, there was the opportunity to
consider more wideꢀranging improvements to these latter stages. The assessment of 325 mesh
potassium carbonate as a cheaper and more easily suspended alternative to the cesium carbonate used to
date was initially characterized by excessive reaction times. This was addressed through the addition of
1 relative volume of water which presumably helped to solubilize the base (Scheme 9). More basic
systems, accessed by adding more water or replacing too much of the potassium carbonate charge with
potassium hydroxide liquor, gave rise to byproducts arising from the hydrolysis (not shown) of the
amide or Cpyrazine–O bonds of ether 12. As with all heterogeneous reactions, there was the risk of the
offline analysis of the reaction being performed with a sample that was not representative of the whole
batch. With this in mind, a model based on inꢀline Raman data was built to allow the conversion of acid
2 to ether 12 to be followed during production.24 A small amount (0.1ꢀ0.2 LCAP) of impurity 32 always
formed in the reaction, presumably through the intermediacy of an activated carbonyl species 33
(Scheme 9). Impurity 32 could be conveniently degraded to acid 12 by adding more water at the end of
the reaction to basify the system.
ACS Paragon Plus Environment

Page 17 of 35
Organic Process Research & Development
1
2
3
4
Scheme 9 Transformations associated with SNAr reaction
5
6
7
8
9
O
O
CO Ar
O
2
CO H
O
MeO
2
MeO
OAr'
2
MeO
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
32
side
reaction
OAr'
OAr'
33
2 ArOH
OH
+
water
N
Cl
desired
reaction
2
−
N
CO H
O
N
2
MeO
N
O
3 Ar'Cl
N
N
O
12
O
Conditions: DMSO-water (6:1, v/v), K CO , 57 C then water
°
2
3
The acidification of the batch at the reaction close with dilute aqueous hydrochloric acid resulted in the
uncontrolled precipitation of acid 12. After gathering solubility data for this material as a function of
temperature and solvent (DMSOꢀwater mixtures), a final solvent composition of 56% v/v dimethyl
sulfoxide was targeted. This was achieved by acidifying with hydrochloric acid, whilst taking care to
ensure the final pH was above 2.5 so as to avoid the acidꢀmediated degradation of acid 12. The
potassium chloride generated by the neutralization of the potassium carbonate base was screened off in
order to control the sulfated ash content of the product. After subsequent cooling and seeding, a slurry
of acid 12 formed which filtered rapidly and gave low losses to liquors (1.8% yield loss). After drying,
the material isolated contained low amounts of dimethyl sulfoxide (<0.05% wt/wt), inorganics (0.35%
wt/wt), water (0.16% wt/wt) and related substances (0.56 LCAP). Despite the volumetric efficiency of
this processing option, to date there has not been the opportunity to test it in anything other than a
laboratory setting. Instead, as part of plant manufactures, the acid 12 content of the slurry arising from
the acidification of the batch was dissolved in warm isopropyl acetate by way of liquidꢀliquid
extractions. A water wash rid the extracts of adventitiously extracted dimethyl sulfoxide before
ACS Paragon Plus Environment

Organic Process Research & Development
Page 18 of 35
1
2
3
4
5
6
azeodrying and cooling crystallized acid 12 that was readily filtered, washed and dried. This protocol
allowed 742 kg of acid 12 to be produced in an overall yield of 93%.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AMIDATION
The elaboration of acid 12 to AZD1656 (1) requires its union with aminopyrazine 4. The poor
nucleophilicity displayed by aminopyrazine 4 led to limited success when a slew of different amidation
conditions were screened. Whilst conversions were in general aided by increasing the reaction
temperature, this was often at the expense of competing SNAr reactions at the pyrazine rings of
AZD1656 (Figure 2).25 Activating acid 12 as an acid chloride generated hydrogen chloride which,
without scavenging, was found to open the azetidine ring so as to form alkyl chloride 34 as a byproduct
(Figure 2). The limited delocalization of the azetidine nitrogen lone pair into the amide carbonyl can be
used to rationalize the comparative basicity of the azetidine nitrogen and hence the susceptibility of the
azetidine amide towards this sideꢀreaction. It is notable that this ringꢀopening of azetidine amides has
found preparative use.26
ACS Paragon Plus Environment

Page 19 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
N
N
O
O
MeO
N
N
H
N
N
O
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
O
N
N
O
O
MeO
N
H
N
N
O
34
H
N
Cl
O
Figure 2 Sites of side reactions when screening conditions for conversion of acid 12 to AZD1656 (1)
and structure of alkyl chloride 34
A successful reagent combination used nꢀpropylphosphonic anhydride (T3P) and Nꢀmethylmorpholine
in 2ꢀmethyltetrahydrofuran. Whilst this enabled early development deliveries of AZD1656 (1) (Scheme
6), the continued use of T3P posed a number of drawbacks. These included the need and expense
associated with the use of a ~70% mol/mol excess of the reagent to completely consume acid 12.27 As a
listed chemical weapons precursor,28 the storage, use and disposal of T3P needs to be fully documented,
whilst the nꢀpropylphosphonic acid waste generated by its use requires treatment prior to discharge into
watercourses. Finally, impurities formed in a process using T3P retarded the rate of crystallization of
the desired polymorph (vide infra), necessitating its later recrystallization.
With the need to development a longꢀterm solution for the amidation step, attention turned to the
activation of acid 12 as an acid chloride 35 (Scheme 10). Whilst the use of thionyl chloride in this
regard was one of the better performers in the initial screen of conditions, the conversion was still only
ACS Paragon Plus Environment

Organic Process Research & Development
Page 20 of 35
1
2
3
4
5
6
7
8
9
modest due to the formation of significant quantities of the acid anhydride intermediate 36 (Scheme 10),
a material that showed limited reactivity towards aminopyrazine 4. The generation of hydrogen chloride
as a byproduct also brought the need to scavenge it adequately so as to avoid the formation of alkyl
chloride 34 (Figure 2).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In spite of these challenges, the economics of using thionyl chloride as a reagent were persuasive. It was
recognized from the outset that the presence of even small amounts of an acyl pyridinium species 37
could provide a valuable conduit for improving the conversion of acid 12 to acid chloride 35 (at the
expense of acid anhydride 36), as shown in Scheme 10, and accelerating the conversion of acid chloride
35 to AZD1656 (1).29 It was also appreciated that through the judicious choice of solvent,
(super)stoichiometric amounts of pyridine itself could simultaneously be used to scavenge the hydrogen
chloride generated as a byproduct of the acid chloride formation, limiting its ability to open up the
azetidine ring. The examination of literature pKa data immediately drew us to the potential use of
31
acetonitrile as the reaction solvent (pKa HCl = 10.3,30 pKa pyH = 12.5 ). Gratifyingly, this combination of
+
pyridine additive (3 eq.) and solvent met with immediate success and a readily stirred reaction mixture.
Raman monitoring revealed the rapid consumption of acid 12 to give an intermediate, presumably
anhydride 36. Levels of the latter then fell away to low levels at the expense of growth in levels of
another material, presumably acid chloride 35, which was now able to persist given the ever decreasing
levels of acid 12. The addition of aminopyrazine 4 at this point led to a high reaction conversion to
AZD1656 (1) in under 30 min and the material isolated contained only ca. 200 ppm of alkyl chloride 34.
Process optimization involved the addition of the activated species to a solution of aminopyrazine 4.
This controlled overreaction to form Oꢀacyl impurity 38 (Scheme 10), and allowed a reaction conversion
of 99 LCAP to be reliably achieved. Interestingly, the reaction conversion was found to be adversely
affected when particularly dry acetonitrile (20 ppm) was used. This water had a role in hydrolysing the
excess thionyl chloride (20% mol/mol) present at the end of the activation stage of the reaction,
ACS Paragon Plus Environment

Page 21 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
preventing the excess reagent from transforming aminopyrazine 4 into an Nꢀsulfinyl imine 39 (Scheme
10).32 Gratifyingly, the reactions went to completion with a 10% molar excess of aminopyrazine 4 and
plantꢀgrade acetonitrile (water content 100ꢀ200 ppm). Workup development included washing out the
pyridine at the end of the reaction using sulfuric acid, rather than hydrochloric acid, in order to mitigate
the risk of azetidine ringꢀopening.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 10 Working mechanism summarizing key transformations consistent with conversion of acid
12 to AZD1656 (1)
35
12 RCOOH
(until 12 completely
38
consumed)
4
py
35 or 37
py
4
36
37
AZD1656 (1)
12 RCOOH
a) i) SOCl , py (3 eq.), MeCN, 20 C, ii) add to 4 in MeCN, 20 C
°
°
2
excess
SOCl2
4
39
CRYSTALLIZATION OF AZD1656 (1)
Screening had unearthed five polymorphs (including hydrates) on top of Form VI, the
thermodynamically most stable polymorph that had been requested by drug product colleagues. Form
VI could be accessed by slowly cooling a methyl isobutyl ketone (MIBK) solution of the API in the
ACS Paragon Plus Environment

Organic Process Research & Development
Page 22 of 35
1
2
3
4
5
6
7
8
9
presence of micronized Form VI seed (>55 m2/kg input) such that the batch was kept supersaturated
with respect to Form VI but undersaturated with respect to a metastable (but kinetically more accessible)
form.33 When levels of chloropyrazine 3 and compound 40 (Figure 3), the latter arising from the
presence of αꢀresorcylic acid (30) in acid 2 were below levels that slowed the rate of crystallization of
Form VI, this polymorph could be crystallized in an isolated yield of 86%. Slow cooling was also
important in that it reduced the entrainment of MIBK in the crystals. The enantiomeric excess of the
AZD1656 (1) prepared (97.6 LCAP), prior to the crystallization of Form VI was in accord with that of
the (R)ꢀ1ꢀmethoxyꢀ2ꢀpropanol (7) input (97.5 LCAP), consistent with a complete inversion of
stereochemistry during the Mitsunobu reaction. The crystallization of Form VI reduced the amount of
the undesired enantiomer to 1.3ꢀ1.4 LCAP. As the initial crystallization was performed in a facility that
was not fully certified for the isolation of API destined for clinical trials, it was necessary to redissolve
the material in MIBK, screen it and to crystallize it again in just such a facility. When the amidation
process was not capable of producing Form VI directly (as was the case with the T3Pꢀmediated
amidation), such a recrystallization was required anyway. Whilst it may not be a necessity for any
future manufactures performed in an appropriately certified facility, in this case it allowed the isolation
of the API in a yield of 86%.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
40
Figure 3 Structure of API impurity 40
CONCLUSIONS
ACS Paragon Plus Environment

Page 23 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Various strategies were investigated in order to differentiate between the hydroxyl groups of an αꢀ
resorcylate derivative. The option pursued involved the monobenzoylation of methyl αꢀresorcylate (5),
a slurryꢀtoꢀslurry reaction that relied on the insolubility of the monobenzoyl ester product 27 in the
reaction medium and maintaining accurate pH control on changing scale and equipment in order to limit
bisbenzoylation. The chiral sideꢀchain was emplaced using the Mitsunobu reaction: the solubility of a
downstream intermediate in alkali allowed it to be partitioned away from the organicꢀsoluble
byproducts. These reaction conditions generated significantly less waste than a Williamson
etherification variant. Thionyl chloride and pyridine in acetonitrile were used to achieve the successful
amidation of an aminopyrazine 4 displaying low reactivity. These conditions offered a substantial costꢀ
saving over the use of T3P. They also allowed access to the desired polymorph of the drug substance, a
form that was thermodynamically preferred but slowꢀgrowing and disfavored by the presence of
impurities, such that the recrystallization of the API was no longer a necessity. The processes described
herein allowed the production of 506 kg of AZD1656 (1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Experimental Section
General. NMR spectra were recorded on a Bruker instrument with tetramethylsilane (TMS) as internal
reference. Chemical shifts are expressed in ppm (δ) relative to TMS, coupling constants (J) are in Hz.
Infrared spectra were recorded with diamond ATR sampling on an FTIR spectrometer using powdered
samples or oils. Melting points were recorded using a Q2000 differential scanning calorimeter from TA
Instruments. High resolution mass spectra were recorded on a Waters Synapt UPLCꢀMS instrument.
Unless stated otherwise, commercial grade materials were used and stated temperatures refer to the
temperature of the batch.
Preparation of ester 2734
ACS Paragon Plus Environment

Organic Process Research & Development
Page 24 of 35
1
2
3
4
5
6
7
8
9
A solution of benzoyl chloride (479 kg, 3.41 kmol) in toluene (658 L) was added over no more than 2 h
to a slurry of methyl αꢀresorcylate (5) (259 kg, 1.54 kmol) in water (2,600 L), which was being stirred at
20 °C. The pH was adjusted to 8.0±0.2 prior to the start of (ca. 55 L) and during (ca. 1525 L) the
addition by dosing a freshlyꢀprepared aqueous solution containing both lithium hydroxide (3.8% wt/wt)
and potassium carbonate (4.8% wt/wt). At the end of the addition of the benzoyl chloride solution, the
batch was maintained for a further 30 min using the addition of the alkali (ca. 65 L) to maintain the same
pH. The slurry was then filtered and the cake was washed with water (640 L). The material was
dissolved in isopropyl acetate (1755 L) and stirred at 55 °C with Vitacel FAC 200 (21 kg) and activated
charcoal (8 kg of 2CW CECA). After 15 min., the slurry was filtered and washed through with
isopropyl acetate (275 L). The batch volume was then reduced by vacuum distillation (275 mBarA).
Once 800 kg of distillate had been collected, toluene was added continuously (1155 L) whilst a further
800 kg of distillate were collected. The batch was cooled to 12 °C over at least 2 h. After a further 1 h,
the slurry was isolated by filtration and the cake washed with toluene (555 L). Vacuum drying (50
mBarA) at 60 °C afforded ester 27 as a white solid (340 kg, 81%). Mp 139 °C. IR 3431, 1706, 1599,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
1331, 1272, 1246, 1144, 1106, 1086, 1071, 1009, 998, 764, 706 cmꢀ1. H NMR (DMSOꢀd6, 500 MHz) δ
(ppm) 10.26 (1H, s), 8.12 (2H, dd, J=8.2, 1.3 Hz), 7.74 (1H, tt, J=7.5, 1.3 Hz), 7.60 (2H, dd, J=8.2, 7.5
Hz), 7.32 (1H, dd, J=2.2, 1.4 Hz), 7.28 (1H, dd, J=2.2, 1.4 Hz), 6.98 (1H, dd, J=2.2, 2.2 Hz), 3.84 (3H,
s). 13C NMR (DMSOꢀd6, 126 MHz) δ (ppm) 165.5, 164.4, 158.6, 151.6, 134.1, 131.6, 129.9 (2C), 129.0
(2C), 128.8, 114.0, 113.6, 113.2, 52.3. HRMS elemental calculated for C15H11O5 (M–H) −: 271.0606;
found: 271.0615.
Preparation of phenol 2. (R)ꢀ1ꢀMethoxyꢀ2ꢀpropanol (7) (102 kg, 1.13 kmol) was added to a solution
of phenol 27 (245 kg, 900 mol) and triphenylphosphine (295 kg, 1.12 kmol) in toluene (2087 kg), and
washed through with toluene (20 kg). Diisopropyl azodicarboxylate (228 kg, 1.13 kmol) was added
over at least 1 h whilst maintaining the batch temperature at 2 °C. After stirring for 30 min at this
ACS Paragon Plus Environment

Page 25 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
temperature, and warming to 20 °C, the batch was seeded with the 1:1 mol/mol complex of
triphenylphosphine oxide and diisopropyl hydrazinedicarboxylate (0.6 kg, 0.2% wt/wt). After stirring
for 2 h, the slurry was filtered and the cake washed with toluene (210 kg). A chromatographicallyꢀ
purified sample of the filtrate gave ether 29 as a colorless oil. IR 1721, 1589, 1446, 1300, 1242, 1136,
1101, 1079, 1060, 1022, 996, 765, 705 cmꢀ1. 1H NMR (DMSOꢀd6, 500 MHz) δ (ppm) 8.13 (2H, dd,
J=8.3, 1.3 Hz), 7.71 (1H, tt, J=7.1, 1.3 Hz), 7.58 (2H, dd, J=8.3, 7.1 Hz), 7.40 (1H, app d, J=2.3 Hz),
7.40 (1H, app d, J=2.2 Hz), 7.22 (1H, dd, J=2.3, 2.2 Hz), 4.68 (1H, qdd, J=6.3, 6.1, 4.0 Hz), 3.84 (3H,
s), 3.49 (1H, dd, J=10.6, 6.1 Hz), 3.45 (1H, dd, J=10.6, 4.0 Hz), 3.29 (3H, s), 1.24 (3H, d, J=6.3 Hz).
13C NMR (DMSOꢀd6, 126 MHz) δ (ppm) 165.4, 164.4, 158.7, 151.7, 134.1, 131.7, 129.9 (2C), 128.9
(2C), 128.7, 114.8, 114.4, 113.8, 74.9, 73.3, 58.5, 52.4, 16.3. HRMS elemental calculated for C19H19O6
(M–H)−: 343.1182; found: 343.1202.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The filtrate was treated with methanolic sodium methylate (130 kg of 30.0% wt/wt, 722 mol) and stirred
at 25 °C for 60 min. After cooling to 2 °C, the solution was extracted with dilute potassium hydroxide
solution (1547 kg of 0.25 M, 387 mol, split over two extractions). A chromatographicallyꢀpurified
sample of the extracts gave phenol 21 as a colorless oil. IR (cmꢀ1) 3353, 1721, 1593, 1236, 1148, 1096,
1025, 1001, 765. 1H NMR (DMSOꢀd6, 500 MHz) δ (ppm) 9.82 (1H, s), 6.99 (1H, dd, J=2.2, 1.5 Hz),
6.94 (1H, dd, J=2.4, 1.5 Hz), 6.61 (1H, dd, J=2.4, 2.2 Hz), 4.55 (1H, qdd, J=6.3, 6.0, 4.1 Hz), 3.80 (3H,
s), 3.45 (1H, dd, J=10.5, 6.0 Hz), 3.38 (1H, dd, J=10.5, 4.1 Hz), 3.26 (3H, s), 1.18 (3H, d, J=6.3 Hz).
13C NMR (DMSOꢀd6, 126 MHz) δ (ppm) 166.2, 159.0, 158.9, 131.6, 108.9, 107.8, 107.1, 75.1, 72.8,
58.5, 52.1, 16.5. HRMS elemental calculated for C12H15O5 (M–H)−: 239.0919; found: 239.0926.
The extracts were combined and diluted with potassium hydroxide liquor (100 kg of 50% wt/wt, 891
mol). The solution was stirred at 24 °C for 60 min. The pH was adjusted to 9.5–10.0 using hydrochloric
ACS Paragon Plus Environment

Organic Process Research & Development
Page 26 of 35
1
2
3
4
5
6
7
8
9
acid solution (approximately 57 kg of 33% wt/wt, 516 mol), before it was washed at 23 °C with tertꢀ
butyl methyl ether (927 kg, split over 2 washes). The pH was then adjusted to 1.5 at 23 °C using
concentrated hydrochloric acid (approximately 160 kg of 33% wt/wt, 1.45 kmol) before extractions were
performed with tertꢀbutyl methyl ether (927 kg, split over two extractions). The organic extracts were
combined and stirred for 30 min with activated charcoal (17 kg, 25W CECA). The slurry was filtered
and the charcoal washed with tertꢀbutyl methyl ether (89 kg). The batch volume was reduced by 670 kg
using vacuum distillation at 300 mBarA, before it was diluted with toluene (632 kg). The solution was
washed with water (735 kg) at 50 °C before being diluted with toluene (253 kg) and heptane (300 kg).
Solvent (315 kg) was removed by vacuum distillation at 350 mBarA. The batch was cooled to 38 °C
and seeded with acid 2 (0.2 kg). It was then cooled to 2 °C over at least 1 h and left to desupersaturate
for a further 30 min. The slurry was filtered and the cake washed with toluene (211 kg). The cake was
vacuum dried below 50 °C and 100 mBarA. This afforded acid 2 as a white solid (155 kg, 76%). Mp 42
°C. IR (cmꢀ1) 2933, 1687, 1592, 1448, 1296, 1203, 1147, 1082, 1022, 846, 770, 730, 704, 674. 1H
NMR (500 MHz, DMSOꢀd6) δ (ppm) 6.94 (1H, dd, J=2.2, 1.4 Hz), 6.91 (1H, dd, J=2.4, 1.4 Hz), 6.55
(1H, dd, J=2.4, 2.2 Hz), 4.54 (1H, qdd, J=6.3, 6.0, 4.1 Hz), 3.46 (1H, dd, J=10.5, 6.0 Hz), 3.44 (1H, br
s), 3.44 (1H, br s), 3.41 (1H, dd, J=10.5, 4.1 Hz), 3.27 (3H, s), 1.19 (3H, d, J=6.3 Hz). 13C NMR (126
MHz, DMSOꢀd6) δ (ppm) 167.2, 158.7, 158.6, 132.8, 108.9, 107.4, 107.0, 74.9, 72.6, 58.5, 16.5. HRMS
elemental calculated for C11H13O5 (M–H)−: 225.0763; found: 225.0759.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Preparation of ether 12. A slurry of acid 2 (150.0 kg at 98.5% wt/wt, 653.1 mol), chloropyrazine 3
(136.4 kg at 98.4% wt/wt, 679.2 mol) and potassium carbonate (226 kg, 1.63 kmol) in dimethyl
sulfoxide (886 L) and water (148 L) was stirred with a jacket temperature of 57 °C overnight (16 h).
Water (1240 L) was added and the batch maintained with the same jacket temperature for a further 60
min before being cooled to 21 °C.35 Concentrated hydrochloric acid (203 L of 11.60 M, 2.35 kmol) was
added over 60 min. Dilute hydrochloric acid (approx. 261 L of 1 M, 261 kmol, CARE: offgassing of
ACS Paragon Plus Environment

Page 27 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
carbon dioxide) was then added so that the final measured pH was in the range of 3.0−4.5. The batch
was then warmed and isopropyl acetate (1480 L) charged so that the batch temperature did not drop
below 40 °C. The batch was allowed to equilibrate with a 75 °C jacket and the phases cut. The aqueous
phase was extracted once again at 75 °C with more isopropyl acetate (1480 L) before being discarded.
The two isopropyl acetate extracts were combined and washed at 63 °C with water (739 L), seeded (738
g of acid 12). The jacket temperature was linearly ramped to 60 °C over 6 h. The batch was then
distilled to 900−1200 L using a −450 mbarG vacuum. The jacket temperature was then ramped to 21 °C
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
at −12 °C/h. After stirring out at this temperature for a further 6 h, the slurry was filtered and the damp
cake washed with isopropyl acetate (591 L). Vacuum oven drying (–250 mBarG) at 50 °C afforded acid
12 as a white solid (227.2 kg, 89.8%). Mp 157 °C. IR (cmꢀ1) 1701, 1609, 1565, 1438, 1274, 1112,
1028, 771, 704. 1H NMR (DMSOꢀd6, 500 MHz) δ (ppm) 8.65 (1H, d, J=1.3 Hz), 8.50 (1H, d, J=1.3
Hz), 7.36 (1H, dd, J=2.3, 1.3 Hz), 7.31 (1H, dd, J=2.1, 1.3 Hz), 7.14 (1H, dd, J=2.3, 2.1 Hz), 4.66 (1H,
qdd, J=6.3, 6.1, 3.9 Hz), 4.53 (2H, t, J=7.7 Hz), 4.07 (2H, t, J=7.8 Hz), 3.49 (1H, dd, J=10.6, 6.1 Hz),
3.44 (1H, br s), 3.44 (1H, dd, J=10.6, 3.9 Hz), 3.27 (3H, s), 2.27 (2H, tt, J=7.8, 7.7 Hz), 1.21 (3H, d,
J=6.3 Hz). 13C NMR (DMSOꢀd6, 126 MHz) δ (ppm) 166.4, 162.7, 159.9, 158.9, 153.6, 142.1, 141.4,
133.6, 133.4, 114.4, 113.5, 113.4, 74.9, 73.2, 58.5, 54.2, 48.8, 16.3, 16.2. HRMS elemental calculated
for C19H22N3O6 (M+H+): 388.1509; found: 388.1511.
Preparation of Form VI of AZD1656 (1). A mixture of acid 12 (100 kg, 258 mol) and pyridine (61.3
kg, 774 mol) in dry acetonitrile (500 L) was prepared. Thionyl chloride (36.9 kg, 310 mol) was added
rapidly (CARE: exotherm, offꢀgassing of sulfur dioxide) and washed through with dry acetonitrile (100
L), whilst maintaining the batch temperature below 20 °C. The yellow solution so formed was added
without undue delays to a preꢀprepared solution of aminopyrazine 4 (31.0 kg, 284 mol) in dry
acetonitrile (500 L),36 whilst maintaining the batch temperatures of both solutions below 20 °C. After
washing through with dry acetonitrile (100 L) and a further 1 h at 20 °C, the batch was washed (CARE:
ACS Paragon Plus Environment

Organic Process Research & Development
Page 28 of 35
1
2
3
4
5
6
7
8
9
offꢀgassing) twice with sodium bicarbonate solution (150 L total of 7% wt/wt, split over 2 washes). The
organic phase (solution A) was retained, whilst the aqueous washes were backꢀextracted at 20 °C with
MIBK (250 L). The backꢀextract and solution A were combined and distilled to 450 L at 250 mBarA.
MIBK (1000 L) was charged and the batch redistilled to 400 L at 300 mBarA. MIBK (600 L) was
added and the solution washed at 45 °C with dilute sulfuric acid solution (1000 L of 1 M). The organic
phase (solution B) was retained, whilst the aqueous washes were backꢀextracted at 45 °C with MIBK
(250 L).37 The backꢀextract and solution B were combined and washed at 45 °C with sodium
bicarbonate solution (500 L, 7% wt/wt) then water (500 L). The organic phase was distilled to 400 L
before being diluted with more MIBK (420 L). Seeding at 45 °C with AZD1656 (1) (Form VI, 4.0 kg,
4.0% wt/wt) was followed by aging at this temperature for 4 h. The batch was cooled at 6 °C/h to 20 °C
where it was held for a further 4 h before heptane (1200 L) was added in 6 equal portions at 1 h
intervals. The batch was isolated by filtration and the cake washed with a mixture of heptane and MIBK
(500 L of 2:1 v/v) before being vacuum dried at 45 °C. This yielded AZD1656 (1) as a white solid (107
kg, 85% corrected for 98% wt/wt). Mp 108 °C. IR (cmꢀ1) 3249, 1632, 1549, 1435, 1344, 1307, 1291,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
1273, 1264, 1248, 1225, 1175, 1132, 1113, 1090, 1027, 843, 752, 699. H NMR (DMSOꢀd6, 500 MHz)
δ (ppm) 11.03 (1H, s), 9.26 (1H, d, J=1.4 Hz), 8.67 (1H, d, J=1.3 Hz), 8.55 (1H, d, J=1.3 Hz), 8.34 (1H,
d, J=1.4 Hz), 7.57 (1H, dd, J=2.2, 1.6 Hz), 7.47 (1H, dd, J=2.1, 1.6 Hz), 7.12 (1H, dd, J=2.2, 2.1 Hz),
4.78 (1H, qdd, J=6.3, 6.0, 3.9 Hz), 4.55 (2H, t, J=7.4 Hz), 4.08 (2H, t, J=7.8 Hz), 3.52 (1H, dd, J=10.7,
6.0 Hz), 3.47 (1H, dd, J=10.7, 3.9 Hz), 3.29 (1H, s), 3.26ꢀ3.31 (2H, m), 2.47 (3H, s), 2.28 (2H, tt, J=7.8,
7.4 Hz), 1.25 (3H, d, J=6.3 Hz). 13C NMR (DMSOꢀd6, 126 MHz) δ (ppm) 164.6, 162.7, 159.9, 158.8,
153.6, 148.7, 146.5, 142.2, 141.7, 141.4, 136.3, 135.9, 133.5, 113.3, 113.0, 111.9, 74.9, 73.0, 58.5, 54.1,
48.7, 20.4, 16.4, 16.1. HRMS elemental calculated for C24H27N6O5 (M+H+): 479.2043; found:
479.2051.
ACS Paragon Plus Environment

Page 29 of 35
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Recrystallization of AZD1656 (1). A supersaturated solution of AZD1656 (1) (100 kg, 1 eq.) and
MIBK (670 L), that had been preꢀscreened at 70 °C, was seeded at 45 °C with a micronized (surface
area = 3.3 m2/g) sample of AZD1656 (1) (Form VI, 1.75 kg, 1.75% wt/wt). After being aged for 3 h, the
batch was cooled at 0.1 °C/min to 0 °C,38 where it was isolated by filtration after being held for a further
3 h.39 The cake was sequentially washed with a chilled (0 °C) mixture of MIBK and heptane (200 L of
1:2 v/v) and then heptane (200 L), before being dried under vacuum at 60 °C. This yielded AZD1656
(1) (Form VI, 83.3 kg, 83.3%) as a white solid.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ASSOCIATED CONTENT
1
Supporting information available: H NMR spectra, 13C NMR spectra, infrared spectra and DSC traces
for ester 27, ether 29, phenol 21, acid 12 and AZD1656 (1).
AUTHOR INFORMATION
Corresponding Author
Eꢀmail: Alan.Steven@astrazeneca.com
ORCID Alan Steven: 0000ꢀ0002ꢀ0134ꢀ0918
Notes
The authors declare no competing financial interest.
Acknowledgement
We are grateful to Dottikon Exclusive Synthesis for performing the manufacture of acid 2.
REFERENCES
ACS Paragon Plus Environment