The Development of a Dimroth Rearrangement Route to AZD8931

Article abstract of DOI:10.1021/acs.oprd.6b00412

Recently, the aminoquinazoline motif has been highly prevalent in anticancer pharmaceutical compounds. Synthetic methods are required to make this structure on a multikilo scale and in high purity. The initial route to aminoquinazoline AZD8931 suffered from the formation of late-stage impurities. To avoid these impurities, a new high-yielding Dimroth rearrangement approach to the aminoquinazoline core of AZD8931 was developed. Assessment of route options on a gram scale demonstrated that the Dimroth rearrangement is a viable approach. The processes were then evolved for large-scale production with learning from a kilo campaign and two plant-scale manufactures. Identification of key process impurities offers an insight into the mechanisms of the Dimroth rearrangement as well as the hydrogenation of a key intermediate. The final processes were operated on a 30 kg scale delivering the target AZD8931 in 41% overall yield.

Full text of DOI:10.1021/acs.oprd.6b00412

Subscriber access provided by Warwick University Library
Full Paper
The Development of a Dimroth Rearrangement Route to AZD8931
William Robert Fraser Goundry, Kay Boardman, Oliver Cunningham, Matthew Evans, Martin F.
Jones, Kirsty Millard, Raquel Rozada-Sanchez, Yvonne Sawyer, Paul S Siedlecki, and Brian Whitlock
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.6b00412 • Publication Date (Web): 01 Feb 2017
Downloaded from http://pubs.acs.org on February 3, 2017
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 free 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 accessible to all
readers and 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.
Organic Process Research & Development 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 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
The Development of a Dimroth Rearrangement
Route to AZD8931
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
William R. F. Goundry,*ǂ‡ Kay Boardman,ⱷ Oliver Cunningham,ⱷ Matthew Evans,ⱷ Martin F.
Jones,ⱷ‡ Kirsty Millard,ⱷ Raquel RozadaꢀSanchez,ⱷ Yvonne Sawyer,ǂ Paul Siedlecki,ⱷ Brian
Whitlockⱷ
ǂ
The Department of Pharmaceutical Sciences, AstraZeneca, Silk Road Business Park,
Macclesfield, Cheshire, United Kingdom, SK10 2NA.
ⱷThe Department of Pharmaceutical Technology and Development, AstraZeneca, Silk Road
Business Park, Macclesfield, Cheshire, United Kingdom, SK10 2NA
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 36
1
2
3
4
Table of contents graphic
5
6
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
AcOH
2-MeTHF
80%
Overall yield for the route 41%, on a 30 kg scale
ACS Paragon Plus Environment
2

Page 3 of 36
Organic Process Research & Development
1
2
3
4
KEYWORDS
5
6
7
AZD8931, Quinazoline; Dimroth Rearrangement; Hydrogenation.
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
ABSTRACT
Recently the aminoquinazoline motif has been highly prevalent in antiꢀcancer pharmaceutical
compounds. Synthetic methods are required to make this structure on a multiꢀkilo scale and in
high purity. The initial route to aminoquinazoline AZD8931 suffered from the formation of late
stage impurities. To avoid these impurities, a new high yielding Dimroth rearrangement
approach to the aminoquinazoline core of AZD8931 was developed. Assessment of route options
on a gram scale demonstrated that the Dimroth rearrangement is a viable approach. The
processes were then evolved for largeꢀscale production with learning from a kilo campaign and
two plant scale manufactures. Identification of key process impurities offers an insight into the
mechanisms of the Dimroth rearrangement as well as the hydrogenation of a key intermediate.
The final processes were operated on a 30 kg scale delivering the target AZD8931 in 41%
overall yield.
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 36
1
2
3
4
5
6
INTRODUCTION
7
8
9
Deregulation of the HER receptor family, comprising four related receptor tyrosine kinases
(EGFR, HER2, HER3, and HER4), promotes proliferation, invasion, and tumor cell survival.1ꢀ3
Such deregulation has been observed in many human cancers, including lung, head and neck,
and breast.4ꢀ5 Numerous small molecules have been investigated for inhibition of tyrosine
kinases with the aminoquinazoline motif coming to the fore as a privileged scaffold. Three of the
clinically available treatments, gefitinib (1),6 lapatinib (2),7 and erlotinib (3)8 contain this
arrangement as well as the candidate drug dacomitinib (4) (Figure 1).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
Figure 1. Structure of gefitinib (1), lapatinib (2), erlotinib (3), dacomitinib (4) and AZD8931 (5).
ACS Paragon Plus Environment
4

Page 5 of 36
Organic Process Research & Development
1
2
3
4
5
6
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
1
2
4
3
5
AstraZeneca has further explored the aminoquinazoline scaffold and identified AZD8931 (5) as a
novel oral antiꢀproliferative agent for the treatment of tumors; its mode of action being reversible
tyrosine kinase inhibition (Erb b2/EGFR) (Figure 1).10 The synthetic medicinal chemistry work
used to discover AZD8931 branched out from the quinazoline 6, previously used by AstraZeneca
for the manufacture of gefitinib (1) (Scheme 1). Using this material was an attractive option for
the discovery team as it was easy to vary the functionality at the 4 and 6 positions, to explore a
structure activity relationship.
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 36
1
2
3
4
5
6
7
8
9
FIRST GENERATION ROUTE
Early kilogram scale campaigns to deliver material for toxicological studies used the medicinal
chemistry route (Scheme 1). Chlorination of quinazoline 6 then reaction with
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
3ꢀchloroꢀ2ꢀfluoroaniline (3ꢀCFA) (7) gave aminoquinazoline 8. Deprotection of the acetate
group revealed phenol 9, which was reacted with mesylate 10 to deliver quinazoline 11.
Mesylate 10 was prepared from the commercially available Boc protected alcohol 12.11 Removal
of the Boc group under standard acidic conditions revealed the secondary amine 13, which was
alkylated with 2ꢀchloroꢀNꢀmethylꢀacetamide (14) affording the target 5. The final difumarate salt
form of 5 was then crystallized from a mixture of ethyl acetate and isopropanol, the details of
which lie beyond the scope of this paper and will be discussed in a future communication.
Scheme 1. The 1st generation route to AZD8931
ACS Paragon Plus Environment
6

Page 7 of 36
Organic Process Research & Development
1
2
3
4
5
POCl3,
NaOH
6
DIPEA
7
8
Toluene
MeOH(aq)
9
76%
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
9
8
6
7
MsCl, TEA
Toluene
EtOH(aq)
79%
91%
12
10
HCl
IMS
14
TEA
MeCN
88%
90%
5
13
11
Following tactical deliveries of 6–8 kg, the route was scaled up to deliver 220 kg of AZD8931 at
which time the deficiencies of this route became apparent. The starting materials 7 and 14
contained low levels of the impurities 3ꢀchloroꢀ2,4ꢀdifluoroaniline and chloroacetic acid
respectively, both of which were incorporated into the final structure giving impurities 15 and 16
(Figure 2). Impurities 17 and 18, resulted from the overꢀalkylation of AZD8931 in the final
bondꢀforming step. Attempts to avoid over alkylation by undercharging 14 (0.95 eq) failed, as
over 10% of 13 remained at the end of reaction, and not all of this could be removed during the
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 36
1
2
3
4
5
6
7
8
9
workꢀup and isolation. Although removal of the impurities 15–18 from the free base 5 required
repeated crystallisations, the final difumarate salt formation proved sufficiently effective in
delivering material of the required quality. As part of our general development strategy, 5 was
subjected to a thorough route evaluation to see if the problems inherent with the route could be
avoided.
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 2. The main impurities from the first generation route: AZD8931 difluoro (15);
AZD8931 acid (16); overꢀalkylation products 17 and 18.
16
15
18
17
ACS Paragon Plus Environment
8

Page 9 of 36
Organic Process Research & Development
1
2
3
4
NEW ROUTE INVESTIGATION
5
6
7
8
Several routes were considered to avoid the late introduction of the methylamide side chain and
the associated overꢀalkylation. Avoiding the use of the Boc protecting group on mesylate 10 by
directly using the tertiary amines of general structure 19, failed due to the instability of the
piperidine ring, postulated to be due to decomposition via elimination (Scheme 2). Equally,
changing the order of the steps in the original route, for example introducing the piperidine and
acetamide side chain prior to the aniline, was also unsuccessful.
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
Scheme 2. Proposed decomposition of piperidines of general structure 19 by elimination; X=Cl,
Br, I or OMs
19
Building on existing company knowledge from the gefitinib project11 and external learning from
the literature,9, 12ꢀ15 a Dimroth Rearrangement approach was investigated and proven on gram
scale. Starting from the commercially available benzonitrile 20, the phenol was alkylated with
the previously prepared mesylate 10 (Scheme 3). Deprotection of the nitrogen followed by
alkylation with chloroacetamide 14 gave acetamide 21. Nitration, followed by dithionite
reduction gave aniline 22, the substrate for the Dimroth rearrangement. Potential short cuts in the
synthesis to the key aniline (nitration with concomitant Boc deprotection to give 23 or avoidance
of the Boc group by using 19) were not fruitful. Reaction of aniline 22 with N,Nꢀ
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 36
1
2
3
4
5
6
7
8
9
dimethylformamide dimethyl acetal (DMFꢀDMA) formed amidine 24, which upon heating with
7 in the presence of acetic acid underwent the Dimroth rearrangement to give AZD8931 (5) in
65% yield (Scheme 4). Having demonstrated this route was viable, attention turned to improving
the 15% overall yield, whilst delivering a kilogram scale manufacture to provide material for
toxicological studies and formulation work.
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 3. Approaches to the Dimroth precursor 22.
ACS Paragon Plus Environment
10

Page 11 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
10
9
K2CO3/DMF
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
80%
20
25
23
HCl(aq)
75%
19
14
K2CO3
MeCN
85%
21
Acetic acid
26
85%
Sodium dithionite
water
65%
22
28
Scheme 4. Initial Dimroth rearrangement route.
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 36
1
2
3
4
5
6
7
8
9
DMF-DMA
Toluene
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
85%
24
22
AcOH
Toluene
7
65%
5
FIRST KILO DIMROTH MANUFACTURE
For operational efficiency, the etherification and Boc deprotection from phenol 20 to piperidine
26 were developed as a telescope process. Phenol 20 was reacted with mesylate 10 in nꢀbutanol,
which allowed a water wash to remove impurities (Scheme 5). The volume was reduced by
distillation and reaction with aqueous HCl resulted in deprotection of 25 and crystallization of
nitrile (26) as the HCl salt in 71% yield and >99% purity. Three batches were manufactured, one
at 20 L scale (0.69 kg) and two at 100 L scale (3.4 kg). During the etherification step, the
reaction was an extremely thick slurry, implying the chemistry could not be performed at a larger
scale.
Scheme 5. First Kilogram manufacture, conditions to the Dimroth precursor 22.
ACS Paragon Plus Environment
12

Page 13 of 36
Organic Process Research & Development
1
2
3
4
5
6
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
10
K2CO3
HCl(aq)
n-butanol
26
71%
20
, EtOH
14
Triethanolamine
25
81%
Acetic acid
Sodium
dithionite
76%
HCl(aq)
2-MeTHF
78%
21
28
22
As part of our wider strategy to investigate replacing alkylating agents with the borrowing
hydrogen approach (hydrogen autoꢀtransfer), we attempted the reaction of piperidine 26 with
glycolic amide 27 (Scheme 6).16 After chromatography to remove several byproducts, a 45%
yield of 21 was achieved. Although this was a promising demonstration of the borrowing
hydrogen approach, due to the impurities generated it was not advanced further.
Scheme 6. Reaction of 26 and 27 under borrowing hydrogen conditions.
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 36
1
2
3
4
5
6
7
8
[IrCP*Cl2]2 (2.5 mol%)
NaHCO3 (5 mol%)
Toluene, 125°C, 21h
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
45%
27
26
21
Development of the alkylation began with a combined base and solvent screen, covering a broad
range of conditions. Investigation of six preliminary hits identified reaction conditions using
triethanolamine as base in 4.0 rel vol (relative volumes, liters of solvent relative to kg of
substrate) of ethanol. Following reaction completion, addition of water as an antiꢀsolvent directly
crystallised 21 from the solution. Two batches, one 20 L scale (1.4 kg) and one 100 L scale (5.8
kg) were manufactured. The chemistry scaled up as expected; complete dissolution occurred
prior to crystallisation, which afforded a more controlled crystallisation. This stage proved robust
to impurities, for example, when 2ꢀchloroꢀNꢀmethylꢀacetamide (14), containing 0.4% w/w
chloroacetic acid was used, none of the related acid impurity was present in the isolated product.
Safe nitration conditions using concentrated sulfuric and nitric acid were developed, leading to a
high yield of 28, with <0.1% of nitro regioisomers and dinitro compounds observed. Control of
the regioisomeric nitration impurities was key, since they could carru through the synthesis to
give regioisomers of AZD8931 (Figure 3). The workup developed involved quenching the
strongly acidic nitration mixture with aqueous ammonia, extraction into propionitrile, and
ACS Paragon Plus Environment
14

Page 15 of 36
Organic Process Research & Development
1
2
3
4
subsequent crystallisation. The level of the nitroꢀregioisomers was <0.1% wt/wt in isolated 28
5
6
and these impurities and their derivatives were subsequently fully purged before the final API.
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
Several issues occurred during the four batch manufacture ꢀ one 20 L scale (0.4 kg) and three
100 L scale (1.9 kg). For the reaction, the vessel had a low fill volume (minimum process
volume 5 rel vol) in order to accommodate the dilute work up (maximum process volume 37 rel
vol). In the 100 L vessel, the agitator splashed the reaction mixture up onto the walls; as the
jacket was at 5°C, the acetic acid crystallised onto the sides of the vessel, resulting in the reaction
mixture becoming more concentrated. In the final crystallisation, the product oiled and plated the
walls of the vessel, resulting in a problematic discharge and filtration process.
Figure 3. Regioisomeric impurities observed in the nitration reaction.
29
30
31
For this delivery, the reduction of the nitro group to the aniline was achieved by sodium
dithionite. To avoid reaction hazards associated with an all at once addition of solid sodium
dithionite, an aqueous solution of sodium dithionite was added over 90 minutes to a slurry of
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 36
1
2
3
4
5
6
7
8
9
Nitro 28 in 2ꢀmethyltetrahydrofuran (2ꢀMeTHF). After initial reduction, aqueous HCl was added
over 1.5 h to facilitate the conversion of intermediate nitroso to aniline 22. The mixture was then
quenched with ammonia, the product extracted into 2ꢀMeTHF and isolated by crystallisation.
Both of the batches manufactured, 20 L (0.7 kg) and 100L (2.7 kg) suffered from crystallisation
on the vessel walls, with yields of 69% and 78% respectively.
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 initial route work had focused on showing the Dimroth rearrangement was viable, building
the amidine functionality directly off aniline 22. To make the route more convergent we
investigated forming the required amidine on aniline 7; two amidines were successfully
synthesized, formamidine 32 and aniline dimer 33 (Figure 4). Both when reacted with aniline 22
afforded quinazoline 5 in good yield although 33 gave fewer impurities. Compound 33 can be
isolated as a white crystalline solid but in contrast, formamidine 32 was a viscous oil at room
temperature. Attention focused on using the amidine with both the cleaner reaction profile and
the benefit of the crystalline intermediate.
Figure 4. Coupling partners 32 and 33 for the Dimroth reaction; intermediate 34.
34
32
33
ACS Paragon Plus Environment
16

Page 17 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Investigating the formation of 33, initial reactions of 7 in 2ꢀMeTHF with ethyl orthoformate
stalled at the ethyl imidoformate intermediate 34, requiring distillation to remove ethanol and
drive the equilibrium to product (Figure 4.). A solvent screen was performed to select a solvent
with a good separation from ethanol in the distillation. In cyclohexane the reaction unexpectedly
progressed without requiring distillation. Development of the process gave a simple allꢀin
reaction from which the product crystallises as the reaction progresses (Scheme 7). Manufacture
of both 20 L scale (2.5 kg) batches proceeded in excellent yield (95%) and purity (99% w/w).
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. The synthesis of 3ꢀCFA dimer 33.
Cyclohexane
Acetic acid
2
3 EtOH
95%
33
7
Selecting between the linear Dimroth approach via intermediate 24 and the convergent route
utilizing amidine 33 was a finely balanced decision. The linear route offered an extra isolation
point and both step yields were ~90%, however at least 2 equiv. of aniline 7 were required to
achieve good purity. The convergent route using amidine 33, which gave the slightly cleaner
reaction profile in the Dimroth reaction, was chosen for scale up.
For the Dimroth rearrangement, a sequential acid and solvent screen gave reaction conditions of
anisole with acetic acid as coꢀsolvent (Scheme 8). Development by design of experiment
ACS Paragon Plus Environment
17

Organic Process Research & Development
Page 18 of 36
1
2
3
4
5
6
ensured conditions for manufacture that gave a good spread of representative impurities for the
toxicological batch.
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
For the workꢀup, following the addition of ethanol, and neutralisation with aqueous base, 5 could
be isolated by a seeded cooling crystallisation. Early development reactions were quenched with
sodium hydroxide, but low levels of sodium acetate contaminated the crystallised product. For
the manufacture, the quench was switched to ammonium hydroxide, since the resulting
ammonium acetate was predicted to stay in solution. Two 20 L scale (1.1 kg) batches were
manufactured in 80% yield, with only the second batch displaying a high level of acetate
(Scheme 8). At scale, the product form was poor, leading to extended filtration times and
retention of filter liquors. Discharge from the filter for kilo lab tray drying was difficult as the
solid was damp and sticky. However, the Dimroth route was successfully delivered in the kilo
lab with an overall yield of ~30%.
Scheme 8. Kilo manufacture Dimroth rearrangement conditions.
AcOH
Anisole
80%
22
5
33
7
ACS Paragon Plus Environment
18

Page 19 of 36
Organic Process Research & Development
1
2
3
4
PLANT MANUFACTURES
5
6
7
8
Significant improvements were still required before delivering the material on pilot plant scale.
The key points for attention were: operational issues with thick reaction mixtures; the reactions
crystallising on vessel surfaces; a lengthy workꢀup for the nitration; improving the yield of the
nitro reduction.
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
Diluting the thick reaction mixture during the etherification of 20 did not solve the problem, as it
was coupled with incomplete conversion and longer reaction times. Instead, the problem was
addressed by changing the solvent system to a mixture of water, PEG(400) and
methylisobutylketone (MIBK). This solvent mixture gave a mobile reaction mixture, with faster
reaction times and complete conversion of the starting material 20. An aqueous separation
followed by solvent swap into IPA then allowed the BOC deprotection of 25 and isolation of the
hydrochloride 26. The process was performed on a 20 kg batch scale with 89% yield. The
alkylation to give 21 was unchanged from the kilo lab and scaled effectively as expected.
Development work focused on the remaining steps. A narrow window of operability was
identified for the nitration reaction. Above 1.2 eq of nitric acid resulted in a thermally unstable
nitration mixture, but below 1.0 eq gave an unacceptably slow conversion. Two design of
experiments established that 1.05 eq of nitric acid, with 8.0 eq of sulfuric acid, and 3 rel vol of
acetic acid at 45°C, gave complete reaction in 3 h. Upon completion, the reaction was diluted
with pyridine, dissolved at an elevated temperature, seeded and then cooled to 0˚C to give
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 36
1
2
3
4
5
6
crystalline 28. These improved conditions gave less than 0.05% (area/area by HPLC) of both
nitro regioisomers (29 and 30), with a product purity of 99.4% in 67% isolated yield.
7
8
9
With the goal of simplifying the process and reducing the disparity between minimum and
maximum reactor volumes, a hydrogenation process replaced the dithionite reduction. A mixed
catalyst system of 1% platinum with 2% vanadium was used. It is postulated that vanadium
causes the rapid disproportionation of the intermediate hydroxylamine 35a to nitroso 36 and
product 22;17 this reduces the potential formation of amide impurity 37 by rearrangement
(Scheme 9). Amide impurity formation is conjectured to occur through ring closure of tautomer
35b followed by reduction of the resulting benzoxazolimine 38.
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 9. Postulated mechanism of nitro reduction with mixed platinum and vanadium catalysis
showing formation of amide 37. There are two potential pathways for 35a to give 22.
Disproportionation - occurs rapidly with V promoter
H2
H2
H2
-H2O
-H2O
22
36
35a
28
-H2O
H2
H2
H2
Ar-N=N-Ar
Ar-NH-NH-Ar
Ar-N=N(O)-Ar
-H2O
R =
35b
H2
37
38
ACS Paragon Plus Environment
20

Page 21 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
The reaction was trialed on a 1.5 L scale with 5% of the amide impurity 37 produced. When
transferred to a 30 kg scale on plant, the level of 37 increased to ~8%. This was attributed to
reduced gasꢀliquid mass transfer, increasing the reaction time and affording the intermediate 35b
more time to form the impurity. It was postulated that the problem was exacerbated by increasing
the agitation speed in the vessel. Under the reaction conditions, the nitro starting material is
poorly soluble and the reaction is a viscous slurry; as the reaction progresses, the starting
material gradually dissolves. During the manufacture, dissolution of nitro 28 had been even
slower than expected at the planned agitation speed of 225 rpm, and so the speed was increased
to the maximum of 300 rpm. It was thought this may have resulted in a temporary spike in the
concentration of 35a and hence a period of hydrogen starvation. A second batch with the
agitation constant at 300rpm gave 5% of the impurity 37.
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
Prior to the next plant manufacture, the reduction of the nitro group was reinvestigated from first
principles. In the lab, reducing the hydrogen pressure from 3.5 bar to 0.5 bar only resulted in an
increase in impurity 37 from 1.5% to 2.0%; increased impurity formation was not just down to
poor hydrogen uptake. Experiments using different particle sizes of 28 inferred that faster
dissolution of the starting material resulted in increased levels of the amide impurity 37.
Modelling of the process using computational fluid dynamics,18 confirmed by experimentation,
found that reducing the vessel fill level to maximize hydrogen uptake minimized the side
reactions. The subsequent plant manufacture, produced 61 kg of 22 in 77% yield, with the amide
impurity 37 <1.5%.
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 36
1
2
3
4
5
6
7
8
9
For plant manufacture, the formation of 3ꢀCFA dimer 33 was intensified, reducing the solvent to
5.0 rel vol. Two 40 kg batches were successfully manufactured in 92% yield with purity >97%.
For the following plant campaign, a single 85 kg batch was manufactured. Recirculation of the
mother liquors ensured full transfer of the dense product crystals from the reactor to the filter,
pushing up the yield to 95%.
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 solvent for the Dimroth rearrangement was switched from anisole to 2ꢀMeTHF; this gave an
equivalent reaction profile but faster, cleaner separations from the aqueous layer during workꢀup.
Following a sodium hydroxide quench of the acetic acid and discard of the aqueous phase,
crystallisation of 5 was attempted directly from 2ꢀMeTHF. When the level of residual water was
>4% the product form was poor and a hydrate was suspected (this may have been the cause of
product stickiness in the previous campaign). Drying the solution by distillation to a water level
of <1% improved the form, but the material crystallised in low purity. A small amount of IPA
was added to increase the solubility of impurities and serendipitously an IPA solvate (subꢀ
stoichiometric solvate ~0.93 eq IPA), crystallised in high yield and purity. Optimization of the
crystallisation around these conditions used a partial solvent swap into IPA to dry the reaction.
The losses to liquors were highly dependent on achieving a low water level at the end of the
distillation (Table 1). Aniline 7 was fully rejected by these crystallisation conditions.
Table 1. Crystallisations of 5 from 10 rel vol of 85% IPA 15% 2ꢀMeTHF demonstrating the
effect of water content.
Added water
(Rel Vol)
Purity of 7
Solvate formed
Losses
to
Liquors (%)‡
(HPLC %)
ACS Paragon Plus Environment
22

Page 23 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
0.1
0.3
0.5
98.9
99.1
99.0
98.9
IPA
IPA
IPA
None
2.3
4.3
6.1
5.5
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
‡ Losses to Liquors, the amount of product remaining in the filtrate
For the 30 kg sized plant batches, the process performed worse than expected with a yield of
74.6%, down from the expected 83%. The increased level of amide impurity 37 in the campaign
was linked to an increase in two impurities, the hydrolysed intermediate 39 and
aminoquinazoline 40, which may have accounted for the reduction in yield. Probing the
formation of 39 and 40 through stressing reactions and adding ammonium chloride (a potential
contaminant from the nitration workꢀup) showed a correlation with the level of 37. It is proposed
the formation of these impurities results from amide 37 reacting with amidine 33 to give
quinazolinone 39, releasing ammonia in the process. Ammonia can then react with additional
amidine 33 and subsequently with nitrile 22 to give amino quinazoline 40 (Scheme 10).
Scheme 10. The proposed mechanism of formation for hydrolysed intermediate 39 and
aminoquinazoline 40.
ACS Paragon Plus Environment
23

Organic Process Research & Development
Page 24 of 36
1
2
3
4
5
6
7
8
9
Acetic
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
6
39
37
7
33
7
22
40
Reworking of the batches by recrystallizing from 10 rel vol of IPA, reduced the levels of the
impurities 39 and 40 to an acceptable level; yield for the rework was 93%. Overall, for the first
plant manufacture the yield was 30% including the rework of the crude material.
The Dimroth process remained unchanged for a second plant manufacture. For the first 40 kg
batch, the seed crystals dissolved at 75 ˚C, although the batch subsequently selfꢀseeded. For the
second 40 kg batch, seeding at the lower temperature of 70˚C, allowed the seed to persist and the
batch to crystallise in a more controlled manner. Now with cleaner input material, due to the
ACS Paragon Plus Environment
24

Page 25 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
improved hydrogenation conditions (37 reduced from 5% to 1.5%), the crystallisation losses
were lower, with a higher average yield of 86.5%. All impurities were rejected to an acceptable
level with no need for a rework and the main goal of avoiding impurities 15-18 was achieved by
switching to the Dimroth route. For the second plant manufacture, the overall yield was an
improved 41%.
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
CONCLUSIONS
A scalable approach to the novel active agent AZD8931 has been successfully developed and
operated on a multiꢀkilo scale. The yield of the route was comparable to that of the original
approach. The original problematic impurities were both avoided and purged, whilst new
impurities encountered were successfully controlled. The mechanisms of the nitro reduction and
Dimroth rearrangement reactions were explored highlighting impurity formation pathways.
EXPERIMENTAL METHODS
General. All reactions were carried out in dry reaction vessels under an atmosphere of dry
nitrogen unless otherwise specified. All reagents and solvents were used as received without
further purification unless otherwise specified. tertꢀbutylꢀ4ꢀmethanesulfonyloxypiperidineꢀ1ꢀ
carboxylate (10) was prepared as per the literature.11
Synthesis of 4-methoxy-3-(4-piperidyloxy)benzonitrile hydrochloride (26). 3ꢀhydroxyꢀ4ꢀ
methoxyꢀbenzonitrile (20) (20.0 kg, 134 mol, 1.00 eq), PEG400 (10.7 kg) and MIBK (32.1 kg)
were charged to a 500 L glassꢀlined reactor which had been purged with nitrogen. After addition,
the mixture was heated to 60~80˚C. A solution of potassium carbonate (22.2 kg, 161 mol, 1.20
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 36
1
2
3
4
5
6
7
8
9
eq) dissolved in purified water (20.1 kg) was added dropwise into the mixture at a rate of 40~60
kg/h while keeping the temperature at 60~90˚C. Then the mixture was heated to 95~100˚C. tertꢀ
butylꢀ4ꢀmethanesulfonyloxypiperidineꢀ1ꢀcarboxylate (10) (56.2 kg, 281 mol, 2.10 eq) and MIBK
(96.0 kg) were mixed in another 500 L glassꢀlined reactor which had been purged with nitrogen,
and then heated to 50~70˚C to a clear solution. The solution of
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
tertꢀbutylꢀ4ꢀmethanesulfonyloxypiperidineꢀ1ꢀcarboxylate (10) was added into the main reaction
mixture at a rate of 60~90 kg/h, keeping the reaction temperature at 95~100˚C. After addition,
the mixture was refluxed at 97~103˚C for reaction. The reaction was considered complete when
conversion of 3ꢀhydroxyꢀ4ꢀmethoxyꢀbenzonitrile (20) was ≥98%. After reaction completion, the
mixture was cooled to 85~90˚C. Purified water (80.0 kg) was added into the mixture at a rate of
50~100 kg/h whilst maintaining the temperature at 65~90 ˚C. The mixture was stirred for 0.5h
and held for 0.5h at 60~66˚C before separation. The organic phase was washed with purified
water (60.0 kg) at 60~66˚C. It was then stirred for 0.5 h, held for 0.5 h and then the aqueous
phase was separated off. After washing, the pH of aqueous phase was ≤10. The organic phase
was cooled to ≤50˚C, and then concentrated at ≤50˚C under reduced pressure (P≤ꢀ0.08 MPa)
until 70~80 L remained and the residual water of mixture was ≤0.15% by KF. The residue was
diluted with isopropanol (173.5 kg) and then heated to 75~85˚C. Maintaining the temperature at
75~85˚C, HCl (21% in IPA, 36.8 kg) was added into the mixture at a rate of 30~40 kg/h. Then 4ꢀ
methoxyꢀ3ꢀ(4ꢀpiperidyloxy)benzonitrile hydrochloride 26 seed (0.1 kg, 0.43 mol, 0.003 eq) was
added and stirred at 75~85˚C for 0.5h. HCl (22.4% in IPA, 18.2 kg) was added into the mixture
at a rate of 30~40 kg/h whilst maintaining the temperature at 75~85˚C. The mixture was stirred
at 75~85˚C until the conversion of tertꢀbutyl 4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀphenoxy)piperidineꢀ1ꢀ
carboxylate (25) was ≥98%. Then the mixture was cooled to ꢀ3ꢀ3˚C at a rate of 5~10˚C /h and
ACS Paragon Plus Environment
26

Page 27 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
then stirred at 0~5˚C for at least 3h. The mixture was filtered and washed twice with IPA (47.0
kg, 48.0 kg) which was preꢀcooled to 0~5˚C. Each time it was soaked for at least 15 min and
then filtered. The cake was dried at 35~40˚C until the content of IPA was ≤1%, to give 4ꢀ
methoxyꢀ3ꢀ(4ꢀpiperidyloxy)benzonitrile hydrochloride (26) as a white powder (31.9 kg, 137 mol,
88.5% yield); 1H NMR (400 MHz, DMSOꢀd6) δ ppm 1.87 (m, 2 H), 2.11 (ddd, J=10.2, 6.7, 3.3
Hz, 2 H), 3.04 (ddd, J=12.7, 9.0, 3.4 Hz, 2 H), 3.22 (m, 2 H), 3.86 (s, 3 H), 4.66 (m, 1 H), 7.17
(d, J=8.5 Hz, 1 H), 7.47 (dd, J=8.4, 1.9 Hz, 1 H), 7.58 (d, J=1.9 Hz, 1 H), 9.28 (br. s, 2 H); m/Z
ES+ 232.8 [MH]+; HRMS found [MH]+ = 233.1286, C13H16N2O2 requires [MH]+ = 233.1212;
Assay (QNMR) 95.2 wt %/wt.
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
Synthesis of 2-[4-(5-cyano-2-methoxy-phenoxy)-1-piperidyl]-N-methyl-acetamide (21).
Anhydrous ethanol (50.5 kg) was charged into a 500 L glassꢀlined reactor which had been
purged with nitrogen, followed by the addition of 4ꢀmethoxyꢀ3ꢀ(4ꢀpiperidyloxy)benzonitrile
hydrochloride (26) (31.9 kg, 137 mol, 1.00 eq), triethanolamine (64.2 kg, 430 mol, 3.14 eq) and
anhydrous ethanol (25.0 kg). Then a solution 2ꢀchloroꢀNꢀmethylacetamide (14) (17.6 kg, 163
mol, 1.19 eq) in anhydrous ethanol (25.0 kg) was slowly added into the mixture. After addition,
the mixture was heated to 80~81˚C and held at this temperature for 12 h. The reaction was
considered complete when the tertꢀbutyl 4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀphenoxy)piperidineꢀ1ꢀ
carboxylate (26) conversion was ≥96%. After reaction completion, the mixture was cooled to
70~73 ˚C. Purified water (257.4 kg) was added into the mixture while maintaining the
temperature at 65~73 ˚C; the mixture was then cooled to 58~60 ˚C. 2ꢀ[4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀ
phenoxy)ꢀ1ꢀpiperidyl]ꢀNꢀmethylꢀacetamide (21) seed (0.1 kg, 9.33 mol, 0.07 eq) was added and
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 36
1
2
3
4
5
6
7
8
9
stirred at 58~60 ˚C for at least 0.5 h. The mixture was cooled to 49~51 ˚C at a rate of ≤20 ˚C /h
and stirred for further 30~60 min, heated to 59~61 ˚C and stirred for further 30~60 min; this
temperature cycle was repeated four times. Then the mixture was cooled to 0~3˚C with a rate of
≤20˚C /h and stirred for further 1~2 h. The mixture was filtered on a nutsche filter. The 500 L
glassꢀlined reactor was rinsed with mixed solvent of purified water (48.0 kg) and anhydrous
ethanol (37.5 kg). The wash liquor was transferred to soak the filter cake for at least 15 min and
filtered. The filter cake was soaked with purified water (96.0 kg) for at least 15 min and then
filtered. The filter cake was dried at 40~45˚C until KF≤1.5% and gave 2ꢀ[4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀ
phenoxy)ꢀ1ꢀpiperidyl]ꢀNꢀmethylꢀacetamide (21) as white solid (29.2 kg, 96.2 mol, 81.1% yield);
1H NMR (400 MHz, DMSOꢀd6) δ ppm 1.70 (m, 2 H), 1.93 (m, 2 H), 2.32 (br t, J=9.3 Hz, 2 H),
2.64 (m, 5 H), 2.92 (s, 2 H), 3.84 (s, 3 H), 4.43 (m, 1 H), 7.13 (d, J=8.4 Hz, 1 H), 7.41 (dd,
J=8.4, 1.9 Hz, 1 H), 7.48 (d, J=1.9 Hz, 1 H), 7.71 (m, 1 H); m/Z ES+ 304.1 [MH]+; HRMS found
[MH]+ = 304.1660, C16H21N3O3 requires [MH]+ = 304.1583; Assay (QNMR) 99.0 wt %/wt.
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
Synthesis of 2-[4-(5-cyano-2-methoxy-4-nitro-phenoxy)-1-piperidyl]-N-methyl-acetamide
(28).
Acetic acid (85.0 kg) was charged to a 500 L nitrogen inerted glassꢀlined reactor, the stirrer was
started, and 2ꢀ[4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀphenoxy)ꢀ1ꢀpiperidyl]ꢀNꢀmethylꢀacetamide (21) (29.2 kg,
96.2 mol, 1.00 eq) was added. After addition, the mixture was cooled to 10±1˚C, then
concentrated sulfuric acid (77.1 kg) was added into the mixture while maintaining the
temperature ≤35˚C. Then 60.3% concentrated nitric acid (10.7 kg) was added via a carboy into
ACS Paragon Plus Environment
28

Page 29 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
the mixture at 25~35˚C. The pipe and the carboy were washed with acetic acid (7.6 kg) and then
the wash liquor was transferred into the 500 L glassꢀlined reactor. The mixture was heated to
45˚C at a rate of ≤5˚C/h. The mixture was held at 42~48˚C for 4 h and the reaction was
considered complete when conversion of 2ꢀ[4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀphenoxy)ꢀ1ꢀpiperidyl]ꢀNꢀ
methylꢀacetamide (21) was ≥99%. After reaction completion, the mixture was cooled to
10~20˚C. Then the reaction mixture was quenched with mixture of purified water (118.0 kg) and
ice (116.9 kg) in a nitrogen inerted 3000 L glassꢀlined reactor; the temperature was maintained
≤35˚C during the quench. After quenching, ammonia (22.3% 59.1 kg) was added dropwise into
the mixture while maintaining the temperature at ≤25˚C. Then pyridine (258.0 kg) was added
slowly into the mixture while maintaining the temperature at ≤35˚C. The mixture was warmed to
80~83˚C and stirred for 30 min until the solid dissolved completely, then held at 78~83˚C before
separation. The organic phase was cooled to 72~73˚C, seeded with 2ꢀ[4ꢀ(5ꢀcyanoꢀ2ꢀmethoxyꢀ4ꢀ
nitroꢀphenoxy)ꢀ1ꢀpiperidyl]ꢀNꢀmethylꢀacetamide (28) (90.0g, 0.26 mol, 0.003 eq) and stirred at
72~73˚C for 1 h. The mixture was cooled to 0~5˚C, and purified water (146.2 kg) was added
dropwise into the mixture. The mixture was stirred for 1 h at 0ꢀ5˚C, then the mixture was
filtered. The reactor was washed with mixed solvent of purified water (58.4 kg) and anhydrous
ethanol (41.0 kg). The wash liquor was transferred to wash the filter cake by soaking for at least
15 min. The reactor was washed again with mixed solvent of purified water (58.5 kg) and
anhydrous ethanol (41.0 kg). The wash liquor was transferred to wash the filter cake by soaking
for at least 15 min. The filter cake was dried at ≤60˚C until KF was ≤2% giving 2ꢀ[4ꢀ(5ꢀcyanoꢀ2ꢀ
methoxyꢀ4ꢀnitroꢀphenoxy)ꢀ1ꢀpiperidyl]ꢀNꢀmethylꢀacetamide (28) as a light yellow solid (22.5
kg, 64.6 mol, 67.1% yield); 1H NMR (400 MHz, DMSOꢀd6, TFA) δ ppm 2.10 (br s, 1 H), 2.18
(br s, 1 H), 2.30 (m, 1H), 2.70 (d, J=4.5 Hz, 3 H), 3.21 (br s, 2 H), 3.40 (br s, 1 H), 3.58 (br d,
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
ACS Paragon Plus Environment
29