Manufacture of the PI3K β-Sparing Inhibitor Taselisib. Part 2: Development of a Highly Efficient and Regioselective Late-Stage Process

Article abstract of DOI:10.1021/acs.oprd.9b00050

A highly efficient and regioselective manufacturing route for the phosphoinositide 3-kinase β-sparing inhibitor taselisib was developed. Highlights of the synthesis include: (1) magnesium-mediated formation of a challenging cyclic amidine; (2) regioselect

Full text of DOI:10.1021/acs.oprd.9b00050

Subscriber access provided by Washington University | Libraries
Full Paper
Manufacture of the PI3K #-Sparing Inhibitor Taselisib. Part 2:
Development of a Highly Efficient and Regioselective Late-Stage Process
Frédéric St-Jean, Travis Remarchuk, Remy Angelaud, Diane E. Carrera, Danial Beaudry, Sushant
Malhotra, Andrew McClory, Archana Kumar, Gerd Ohlenbusch, Andreas Schuster, and Francis Gosselin
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.9b00050 • Publication Date (Web): 06 Mar 2019
Downloaded from http://pubs.acs.org on March 7, 2019
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 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Manufacture of the PI3K -Sparing Inhibitor
Taselisib. Part 2: Development of a Highly Efficient
and Regioselective Late-Stage Process
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
Frédéric St-Jean,*,† Travis Remarchuk,† Rémy Angelaud,*,† Diane E. Carrera,† Danial Beaudry,†
Sushant Malhotra,† Andrew McClory,† Archana Kumar,† Gerd Ohlenbusch,+ Andreas M.
Schuster,+ and Francis Gosselin†
†Department of Small Molecule Process Chemistry, Genentech Inc., 1 DNA Way, South San
Francisco, CA 94080, USA
+Small Molecules Technical Development PTDC-C, F. Hoffmann-La Roche Ltd.,
Grenzacherstrasse 124, 4070 Basel, Switzerland
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 36
1
2
3
4
TOC
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
ACS Paragon Plus Environment
2

Page 3 of 36
Organic Process Research & Development
1
2
3
4
ABSTRACT
5
6
A highly efficient and regioselective manufacturing route for the PI3K -sparing inhibitor
7
8
9
taselisib was developed. Highlights of the synthesis include: 1) magnesium-mediated formation
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
of
a
challenging cyclic amidine; 2) regioselective imidazole construction via
alkylation/condensation with bromopyruvic acid; and 3) triazole formation with N’-isopropyl
acetamidrazone (IPAA) to generate the key bromobenzoxazepine core intermediate. Subsequent
highly efficient one-pot
palladium-catalyzed Miyaura borylation/Suzuki
cross-
coupling/saponification, followed by a CDI-mediated coupling with ammonia, led to the
pentacyclic taselisib. This new synthetic approach provides a more efficient route to taselisib
with a significant decrease in PMI compared to the previous early-development routes to the
benzoxazepine core. Finally, implementation of a controlled crystallization provided the active
pharmaceutical ingredient (API) with the desired polymorphic form.
Keywords: Suzuki-Miyaura cross-coupling, Miyaura borylation, benzoxazepine, late-stage
development, 1,2,4-triazole, amidine.
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 36
1
2
3
4
INTRODUCTION
5
6
7
8
9
Taselisib (GDC-0032)1 is an orally bioavailable, potent and selective small molecule inhibitor
of class I phosphoinositide 3-kinase (PI3K). It is active against all isoforms of the kinase (α, β, δ,
and γ), but exhibits 30 fold less potent inhibition against the β form; which is postulated to play a
role in insulin signaling and glucose homeostasis.2 Recently, it has been demonstrated that
taselisib has increased activity against PIK3CA-mutant breast cancer.3 As part of a phase I to
phase III accelerated drug development program, multi-kilogram amounts of taselisib active
pharmaceutical ingredient (API) were required to support human clinical studies. The late-stage
synthetic strategy involved a key carbon-carbon bond formation between bromobenzoxazepine 1
and the boronic ester derived from bromopyrazole 2 (Figure 1). Much of the development effort
was spent on preparing the highly functionalized tetracyclic compound 1, for which three
different manufacturing approaches were explored.
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 Negishi route 1 had many critical synthetic drawbacks, which led to the development of
chloroketone [3+2] route 2 (Figure 1), as discussed in our preceding Part 1 paper.4 This second
route was viable for early-development API manufacturing, but still suffered from some
synthetic issues that needed to be addressed, including: formation of a regioisomer at the
imidazole nitrogen alkylation step; formation of a difficult-to-purge N-vinyl impurity during the
construction of the benzoxazepine ring; and use of an unstable -chloroketone intermediate
under
cryogenic
conditions
for
the
imidazole
ring
synthesis.
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
Figure 1. Early vs. late-stage development retrosynthetic approaches to taselisib
These limiting aspects were mitigated in the development of the cyclic amidine route 3 (Figure
2), which starts from the same commodity material (4-bromo-2-fluorobenzonitrile) as route 2.
The key difference is that the benzoxazepine ring is now formed first, to access intermediate 3,
which then undergoes alkylation/condensation with bromopyruvic acid to provide the imidazole
ring of intermediate 4. This can then undergo an amide coupling/condensation with isopropyl
acetamidrazone (IPAA) 5 to access the desired benzoxazepine 1, the key intermediate en route to
taselisib. In this approach, all heterocycle formations are now highly regioselective and no
regioisomers have been detected, which considerably increased the overall purification
efficiency and yield (vide infra). As a result, route 3 greatly improved environmental metrics,
such as process mass intensity (PMI)5, compared to early development routes 1 and 2. Herein,
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 36
1
2
3
4
the development of this sustainable manufacturing process for taselisib is discussed.
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
Figure 2. Revised retrosynthetic analysis of 1 (Route 3)
RESULTS AND DISCUSSION
Development towards the manufacture of 1
To demonstrate the strategy illustrated in Figure 2, SNAr displacement of the aryl fluoride on
4-bromo-2-fluorobenzonitrile was first established with N-Boc-protected ethanolamine, using
t-BuOK (1.2 equiv) at 0 °C and gave rise to the desired product 6 in 68% yield (Scheme 1).
Subsequent deprotection with HCl in EtOH provided the corresponding amine hydrochloride salt
7•HCl. Furthermore, under the same conditions, the reaction was successfully run with
unprotected ethanolamine to afford 7 directly in 91% yield. It is worth noting that under these
conditions, the undesired N-addition product was not observed.
Scheme 1. SNAr reaction to access 7•HCl
After successful proof-of-concept for this first SNAr step had been demonstrated, there were
several issues that needed to be addressed prior to scaling up this transformation. The reaction
was found to be very exothermic and required long addition times of t-BuOK (up to 30 minutes
ACS Paragon Plus Environment
6

Page 7 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
for a 5 g scale reaction). This exotherm resulted from deprotonation of the ethanolamine and
could be successfully controlled by preforming the aminoalkoxide. and adding that mixture to a
solution of 4-bromo-2-fluorobenzonitrile. Thus, for scale-up THF was replaced with the greener
alternative 2-MeTHF, and a solution of 4-bromo-2-fluorobenzonitrile in 2.5 volumes (V) of 2-
MeTHF was instead added to a solution of the preformed aminoalkoxide in 7.6 V of 2-MeTHF at
0 °C. Using this order of addition, the exotherm was completely dose-controlled.
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
Subsequent work on the amidine cyclization step showed that the HCl salt of 7 was much more
reactive than the free base form (see Table 1), thus an efficient process for generating and
isolating 7•HCl was implemented. It was found that the salt could be formed by adding
101 mol% of HCl directly to the reaction mixture after warming the solution to room
temperature. Both aqueous and organic solutions of HCl (IPA, EtOH, dioxane) led to the
formation of 7•HCl which can be isolated by filtration. However, analysis of the material formed
by direct addition of HCl without aqueous work-up revealed that it had lower than expected
assay with a residue on ignition (ROI) of up to 15.5%.6 Furthermore, it was also observed that
the glassware was being etched by the formation of HF during acidification to form the HCl salt.
For these reasons, aqueous washes with water and 13wt% NaCl solution prior to acidification
were implemented to remove the residual KF salt. Finally, acidification was performed with HCl
in 1-propanol instead of ethanol to avoid the formation of the potential mutagenic impurity
EtCl.7
Next, we turned our attention to the development of the key intramolecular cyclization to form
the benzoxazepine ring system. We attempted to initiate this cyclization from 7•HCl with
catalytic protic acids and bases, however no product was observed until we switched to a Lewis
acidic system (Table 1). In the presence of 20 mol% Zn(ClO4) in dioxane at 80 °C, 24%
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 36
1
2
3
4
5
6
7
8
9
conversion to 3 was obtained alongside formation of dimer 8 after 24 h (Table 1, entry 1).
Prolonged heating resulted in an increase in dimer 8 formation. Interestingly, the use of CuCl
(100 mol%) resulted in complete suppression of dimer formation. However, when these
conditions were performed on gram scale, the heterogeneous reaction stalled and only 13 A%
HPLC of amidine was obtained (Table 1, entry 2). Switching to a solution of 130 mol% of
trimethylaluminum8 in toluene resulted in a homogeneous reaction mixture that was reproducible
on gram scale, giving 83 A% HPLC of 3 and only 1 A% HPLC of the dimer (Table 1, entry 3).
Table 1. Evaluation of Lewis acid mediated cyclization
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
7•HCl
3
8
Lewis
Acid/mol% (°C)
Temp Time
Entry Scale
Solvent
(A%)a
(A%) a (A%)a
(h)
1
2
3
4
5
0.1 g
1.0 g
2.0 g
3.0 g
260 g
dioxane Zn(ClO4)2/20 80
24
24
14
14
5
34
87
3
23
13
95
83
86
23
0
EtOH
toluene
tolueneb
toluene
CuCl /100
80
AlMe3/180 100
AlMe3/180 100
AlMe3/170 100
ND
ND
ND
9
5
b
aDetermined by HPLC (254 nm) analysis of the crude reaction mixture. 7 employed as the
corresponding free base.
ACS Paragon Plus Environment
8

Page 9 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
Further investigation of the intramolecular cyclization showed that the reaction stalled when
starting from 7 (free base) with ~9 A% of the starting material remaining, even when the
stoichiometry of AlMe3 was increased to 180 mol% (Table 1, entry 4).
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
While this cyclization procedure using 170 mol% AlMe3 in toluene at 100 °C was successfully
demonstrated on 260 g to provide 3 in 70% yield with >99 A% purity by HPLC after isolation,
the hazardous pyrophoric properties of AlMe3 led us to try and find an alternative reagent. After
screening a number of Lewis acids, such as ZnCl2, Al(Oi-Pr)3, MgBr2, i-PrMgCl, t-BuMgCl,
PhMgBr, ZnOTf, TiCl4 and Mg(OMe)2, only the latter proved to mediate the cyclization in high
yield.9 While the initial hit was promising, work-up of this reaction proved laborious as the
addition of water resulted in the formation of gels (presumably Mg(OH)2) rendering the
extraction of amidine 3 very difficult. This was reflected in the low isolated yield (56%), from a
reaction conversion of 88 A% HPLC (Table 2, entry 1). Using MeTHF as a co-solvent solved the
gelling issues and amidine 3 could then be isolated in >97% yield (Table 2, entry 2). However,
obtaining the desired MeTHF/MeOH composition while using Mg(OMe)2 proved to be
operationally difficult as the reagent came as a 1M solution in MeOH and had to undergo a
distillation. Thus the Mg(OMe)2 solution was substituted with solid Mg(OEt)2, which afforded
similar high conversions to the desired product 3. Upon scale-up, it was found that dissolving
Mg(OEt)2 and 7•HCl in MeOH (4V) before adding the 2-MeTHF (8V) was required to obtain
near complete conversion (Table 2, entry 2). No dimer 8 was observed using these conditions.
Table 2. Optimization of cyclization with Mg(OR)2
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 36
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
Scale
Solvent
7•HCl
3
Entrya
Mg(OR)2
Yieldc
56
(g) (base/reaction) (A%)b (A%)b
MeOH (15V)
b
b
1
Mg(OMe)2
0.5
5.0
6
2
88
98
MeOH (5V)
MeOH (4V)
2
Mg(OMe)2
97.5
2-MeTHF
(8V)
MeOH (4V)
3
Mg(OEt)2 50.0
2
98
97
2-MeTHF
(15V)
aGeneral conditions: 7•HCl (1.8 mmol, 100 mol%) and Mg(OMe)2 (1M in MeOH, 3.6 mmol,
200 mol%) or Mg(OEt)2 (3.6 mmol, 200 mol%) were dissolved in MeOH or MeOH/THF and the
b
mixture was stirred at 70 °C for 24h. Determined by HPLC analysis (254 nm) of the crude
reaction mixture. cIsolated yield after workup.
The optimization of the work up conditions for the Mg-promoted amidine formation focused
primarily on fully solubilizing the resulting magnesium salts to prevent emulsions after quench.
A reverse quench of the reaction with 10 volumes of a 13% aqueous NH4Cl solution gave a clean
phase separation with pH 10, which allowed extraction of up to 40% of the product10 from the
aqueous phase. In order to increase the efficiency of the extraction, it was necessary to adjust the
ACS Paragon Plus Environment
10

Page 11 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
pH to 11 by addition of 2N aqueous NaOH.11 Although this process was successfully
demonstrated on 50 g of 7•HCl and provided cyclic amidine free base 3 in 97% yield and 98 A%
purity by HPLC (table 2, entry 3), the workup/isolation was lengthy and an alternative procedure
was investigated using acidic conditions instead. By simply adding 5N HCl in 1-propanol to the
crude reaction mixture, the corresponding amidine hydrochloride salt 3•HCl could be isolated as
a solid directly. However, to increase recovery and reduce Mg salt content in the product, MeOH
had to be removed first by distillation with continuous 2-MeTHF addition. With this new
efficient isolation, 3•HCl could be obtained as a white solid in 89% yield and 99.3 A% HPLC
purity (vide infra).
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
Next, the imidazole ring formation to 4 was investigated. Earlier development work had shown
that the imidazole ring could be successfully installed on a related acyclic amidine through a
tandem alkylation/condensation sequence with bromopyruvic acid (BPA, 130 mol%) and
1,1,3,3-tetramethylguanidine (TMG, 400 mol%) in NMP.12 Complete conversion to the desired
imidazole 4 was achieved by applying these same conditions to 3•HCl in NMP (5V) at 50 °C
(Scheme 2).
Scheme 2. Imidazole 4 formation with bromopyruvic acid from amidine 3•HCl
Slow acidification and addition of water resulted in direct crystallization of the product from
the reaction mixture as a single regioisomer.13 However, while the yield using this technique was
high (91% yield), the material obtained was a dark purple colored solid. Adding BPA as a
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 36
1
2
3
4
5
6
7
8
9
solution in 2-MeTHF (2V) and implementing a carbon treatment prior to isolation with aqueous
HCl resulted a lighter colored solid. For scale up, the amount of TMG was increased to
475 mol% from 400 mol% to ensure robustness of the reaction. Imidazole 4 was isolated by slow
addition of 3N aqueous HCl over 2 h followed by filtration and was obtained in 91% yield with
high purity (99.8 A% HPLC) and an acceptable color profile.
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
Next, the synthesis of 1,2,4-triazole 1 was inspired by the reported work from Castanedo and
coworkers,14 who used a three-component approach with a carboxylic acid, a primary amidine
and a monosubstituted hydrazine via intermediates I and II (Figure 3 (A)). To improve
efficiency, an alternative strategy, where intermediate II could be directly intercepted, was
implemented with the direct coupling of acid 4 with an acetamidrazone (Figure 3 (B)). This
method, which had not previously been reported, used intermediate IPAA 5, which had already
been synthesized on kg scale to build the triazole ring in both route 1 and 2 (Figure 1) to
taselisib.1
Figure 3. Three vs two-component approaches to 1,2,4-triazoles
In the published three-component methodology, HATU was used as coupling reagent. For the
two-component approach (Figure 3 (B)) safer and greener 1,1’-carbonyldiimidazole (CDI) was
ACS Paragon Plus Environment
12

Page 13 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
instead used for optimization. Two separate screens were performed to first assess the acyl
imidazole 10 formation with CDI and second, to identify the optimal base to convert 10 with
IPAA•HCl (5) to the corresponding imidazole 1 (Scheme 3). Conversion of starting material acid
4 (100 mol%) to acyl imidazole intermediate 10 was monitored using 140 mol% of CDI in 10
volumes of solvent (THF, 2-MeTHF, DMF, DMA, NMP, NEP, t-amyl alcohol and isoamyl
alcohol) for 30 min at rt. Polar aprotic solvents were found to give the highest conversions with
NMP providing not only the highest conversion (>98 A%), but also the cleanest HPLC profile
and was thus selected for further optimization.
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. One-pot triazole formation to 1
The second screen was performed to identify the optimal base for the amide coupling between
activated intermediate 10 and 5 (Table 3). Potassium carbonate (entry 10, Table 3) was found to
give the highest conversion by HPLC to the desired triazole 1, which was simply isolated by
crystallization after water addition, in >85% yield and >99 A% purity.15
Table 3. Base screen for conversion of acyl imidazole intermediate 10 to triazole 1
Triazole 1
Entry a
1
Base
(A%)b
DIPEA
85
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 36
1
2
3
4
5
6
7
8
9
DBU
TMG
85
60
25
85
70
50
75
90
95
2
3
DMAP
4
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
NMM
5
pyridine
proton sponge
collidine
DABCO
K2CO3
6
7
8
9
10
aGeneral procedure: acid 4 (16 mmol, 100 mol%) was dissolved in NMP (10 vol), then CDI
(22 mmol, 140 mol%) was added and stirred at rt for 30 min. Base (32 mmol, 200 mol%) was
then charged, followed by addition of 5 (32 mmol, 200 mol%), and the mixture was heated to 75
C for 16 h. bDetermined by HPLC analysis (254 nm) of the crude reaction mixture after 16 h.
To ensure consistent performance during this new one-pot triazole formation, a robust
synthesis of IPAA•HCl (5) was needed.16 It was discovered that, by simply adding isopropyl
hydrazine•HCl to CH3CN at 70 C in the presence of 150 mol% of HCl gas, 5 could be directly
isolated by filtration of the reaction mixture (Scheme 4) in 65% yield and 100 A% HPLC
purity.17,18 It is worth noting that this efficient bond-forming reaction has a very high atom
economy.19 Furthermore, because IPAA•HCl was obtained in high quality in this new process,
its loading could be lowered down to 140 mol% from 200 mol% in the triazole ring formation
step with no impact on the isolated yield of 1 (Scheme 5).
ACS Paragon Plus Environment
14

Page 15 of 36
Organic Process Research & Development
1
2
3
4
Scheme 4. Synthesis of IPAA•HCl (5)
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
The 4-step reaction sequence starting from commercially available 4-bromo-2-
fluorobenzonitrile to 1 was successfully demonstrated on pilot plant scale (Scheme 5). This
process generated >200 kg of 1 in 56% yield over 4 steps and >99.9 wt% purity.
Scheme 5. Manufacture of 1 on pilot plant scale.
Development towards the endgame manufacturing of taselisib
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 36
1
2
3
4
5
6
7
8
9
The final steps to generate taselisib from 1 consisted of three reactions. First, a one-pot three-step
Miyaura borylation/Suzuki/saponification reaction sequence between bromobenzoxazepine 1 and
bromopyrazole 2 to access carboxylic acid 13 (Scheme 6). Second, an amidation step to convert
13 to crude API, and finally, a recrystallization to provide taselisib with the required
polymorphic form (Scheme 7).
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 6. One-pot Miyaura borylation, Suzuki coupling and saponification to acid 13
Various Pd catalysts were screened for the optimization of the Miyaura borylation20 reaction in
the one-pot sequence (Table 4). Using conditions similar to those reported in the literature,21 only
the Pd(XPhos) system using Buchwald Pd(XPhos) G1 and G2 pre-catalysts22 (entry 8 and 9) led
to complete conversion to the desired boronic ester 11. After further optimization, the reaction
was run with with 125 mol% of bis(pinacolato)diboron (B2Pin2)23 and 125 mol% of AcOK in 10
volumes of EtOH at 75 C with only 0.3 mol% of pre-catalyst Pd(XPhos) G2 and 0.6 mol% of
extra XPhos ligand and afforded 100% conversion of 2 to the corresponding borylated ester 11.
Table 4. Catalyst screen for the Miyaura borylation of 2 to boronic ester 11
Bromopyrazole 2
Borylated Pyrazole 11
Entry a
1
Catalyst
Conversion (A%)b
(A%)b
Pd(t-Bu3P)2
95
45
ACS Paragon Plus Environment
16

Page 17 of 36
Organic Process Research & Development
1
2
3
4
5
6
Pd(Cy3P)2Cl2
25
85
15
65
2
3
c
Pd(dtbpf)2Cl2
7
8
9
Pd(di-(t-butyl)(4-fluoromethyl-
75
75
4
phenyl)phosphine)2Cl2
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
d
Pd(dppf)Cl2
65
95
35
80
5
6
7
8
9
e
Pd(Amphos)2Cl2
95
80
PdCl2(P(t-Bu)2Ph)2
Buchwald G1 XPhosf
Buchwald G2 XPhosf
100
100
100
100
aGeneral conditions: Bromopyrazole 2 (3.8 mmol, 100 mol%), 1 mol% catalyst, B2Pin2 (4.8
b
mmol, 125 mol%), KOAc (4.8 mmol, 125 mol%) in EtOH (20 V), 75 C, 16 h. Determined by
c
HPLC analysis (254 nm) of the reaction mixture after 16h. dppf = 1,1′-Ferrocenediyl-
bis(diphenylphosphine); ddtbpf = 1,1′-Bis(di-tert-butylphosphino)ferrocene; eAmphos = (4-(N,N-
f
Dimethylamino)phenyl)di-tert-butyl phosphine; XPhos = 2-Dicyclohexylphosphino-2′,4′,6′-
triisopropylbiphenyl.
The solution containing 11 was then directly subjected to the Suzuki-Miyaura conditions by
adding 1 (90 mol %) and aqueous LiOH (1M, 200 mol%). The cross-coupling reaction was then
carried at the same temperature, without the need for an additional catalyst charge. Full
conversion to ester intermediate 12, and upon addition of more LiOH (3.5M, 300 mol%),
complete conversion to acid 13 (>99 A%) was reached after 16 h. Subsequent workup and
acidification with aqueous HCl (12N, 500 mol%), provided acid 13 as a solid after filtration in
88% yield and with high purity (>98 A% HPLC). The isolation of acid 13 initially was
complicated by foaming of the material upon crystallization by acid addition (most likely due to
hydrophobicity of 13). The acid to precipitate the product was changed from HCl to H2SO4 due
ACS Paragon Plus Environment
17

Organic Process Research & Development
Page 18 of 36
1
2
3
4
5
6
7
8
9
to corrosivity issues. The foaming issue was solved by addition of 1-propanol to the aqueous
solution of 13-lithium salt before reacidification. Another added benefit from using 1-propanol
was a greater purging power of the residual Pd present in 13 (<40 ppm vs. ~ 400 ppm Pd without
1-propanol).
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. Amide formation and crystallization to taselisib
For the final amidation reaction, carboxylic acid 13 was pre-activated with CDI (150 mol%) for
3 h in THF at 22 °C. The activated acid was then added to a THF solution containing 7N NH3 in
MeOH (370 mol%) at 20 °C and stirred for 1 h to provide taselisib in high conversion (>99 A%).
Crude taselisib was then isolated via concentration and water addition (7V) to afford a white
solid in 91% yield and 99.7 A% HPLC purity on 12.5 kg scale. Optimal crystallization
conditions to obtain the desired polymorphic Form B were investigated. First, a screen was
performed to identify a single solvent or solvent combination that would provide the API in high
yield and purity with the desired crystal Form B (Table 5). In this study, a lower purity lot of
taselisib (97.2 A% purity) was slurried in the selected solvent mixture, heated until a
homogeneous solution was obtained, and was then cooled to rt to allow for the crystallization of
solids that were submitted to X-ray powder diffraction analysis (XRPD). Most solvent systems
ACS Paragon Plus Environment
18

Page 19 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
afforded the API in very high purity (>99.6 A% HPLC) with the desired Form B,24 except for
MeOH which provided the wrong methanoate Form A (entry 1 and 2, Table 5) and benzyl
alcohol that gave low recovery (entry 5, Table 5). Isoamyl alcohol (entry 3, Table 5) was
selected for development since it provided the API with the desired crystal Form B in 96% yield
and 99.6 A% HPLC purity.
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
Table 5. Solvent screen for the crystallization of taselisib
Purity
Entry a
Solvent (V)
Anti-Solvent (V)
XRPD
Yield
(A%)e
MeOH (30)b
MeOH (26)d
iPrOAc (5)
none
Form A & B 78%
99.6
1
2
3
4
5
6
7
Form A
Form B
Form B
Form B
Form B
Form B
80%
96%
60%
36%
42%
25%
99.9
99.6
99.9
97.5
99.9
Isoamyl alcohol (15)c
Ethyl lactate (8)c
Benzyl alcohol (6)c
NMP (5)b
none
none
iPrOAc (10)
iPrOAc (10)
iPrOAc (10)
AcOH (5)b
99.9
b
aAll experiments were performed on 5 g scale with 97.2 A% purity taselisib. Two-solvent
crystallizations (entries 5-7) were heated until a homogeneous solution was obtained then, 10V
of iPrOAc (5V for entry 1) were added and the solution was then, seeded, cooled to rt and the
solids were filtered and dried at 80 C for 16 h. cSingle solvent crystallizations (entries 2-4) used
the same procedure as the two-solvent crystallizations, but were not seeded. dDrying a sample of
the methanoate solvate (Form A) between 100–120 °C under vacuum for >16 h resulted in a
form change to Form B without impact on the purity of taselisib as determined by HPLC.
eDetermined by HPLC analysis of the isolated solid.
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 36
1
2
3
4
5
6
7
8
9
Further investigation into crystallization parameters was done to ensure a robust procedure
could be developed. In those experiments, crude taselisib (5g, 97.2 A% purity) was suspended in
isoamyl alcohol (22 V), the mixture was heated to 100 C until a homogeneous solution is
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
obtained, cooled down using different cooling ramps (0.05 C/min, 0.1 C/min and 1.0 C/min),
and the solids were filtered and dried. In those experiments, a few important trends were
observed: 1) the optimal seeding point to promote secondary nucleation was found to be between
75–90 °C; 2) no large crystals25 were generated using a fast cooling ramp (1.0 ºC/min), but using
a slower cooling ramp (0.05 to 0.3 ºC/min), desired crystal growth was promoted and bigger
crystals26 were observed; 3) the isoamyl alcohol solvate of taselisib was formed when it was
isolated at low temperatures (<5 °C), but converted to the desired Form B upon drying at ≥80 °C
for 16h. Using this seeding point and slower cooling ramp, highly pure taselisib was obtained in
93% yield and 99.9 A% purity by HPLC, with <0.15 wt% residual isoamyl alcohol by GC on
multi-hundred gram scale.
In order to achieve consistent batch color for the API on large scale-production, a carbon
treatment was implemented prior to the final crystallization. Since taselisib had a low solubility
in the crystallization solvent isoamyl alcohol, the material was dissolved in a mixture of water
and n-propanol for the charcoal and polishing filtration. After filtration the solvent was swapped
to isoamyl alcohol by distillation. However, removal of water and n-propanol had to be
controlled to avoid the formation of undesired solvates. Sampling after the solvent exchange was
technically difficult, given the high potency of taselisib27 and the process conditions. Therefore,
an in-line process analytical technology (PAT) near infrared (NIR) spectroscopy method was
implemented as an in-process control (IPC) test prior to crystallization to avoid any operator
exposure from sampling of the supersaturated hot solution with a highly potent compound. The
ACS Paragon Plus Environment
20

Page 21 of 36
Organic Process Research & Development
1
2
3
4
5
6
NIR PAT measurement also provided the API concentration before addition of the seed crystals
to ensure high recovery yield.
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
Scheme 8. Manufacture of taselisib on pilot plant scale
In conclusion, a great number of improvements were achieved towards a more sustainable
manufacturing process for taselisib (Scheme 8). When comparing the original Negishi route 1
and the cyclic amidine route 3 to access taselisib, a >5-fold reduction in PMI was achieved
(Scheme 9). The longest linear synthesis was reduced from 10 to 7 steps, making the new route
more convergent. Also, a robust controlled crystallization was developed to ensure that taselisib
was produced with the desired polymorphic Form B, in high yield and purity. Using this new
process, taselisib was manufactured on ~16 kg scale per batch in 35% overall yield and with
>99.9 A% purity by HPLC.
Scheme 9. PMI comparison between the early-stage (Negishi and [3+2]) and the late-stage
(cyclic amidine) manufacture routes to taselisib
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 36
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
EXPERIMENTAL SECTION
General Information. Unless otherwise noted, all reactions were run under a nitrogen
atmosphere, and solvents and reagents were used without further purification. Nuclear magnetic
resonance (NMR) measurements were carried out on a Bruker Avance 600 spectrometer at 25C.
1
The 600 MHz machine is equipped with a 5 mm CryoProbe. Data for H NMR chemical shifts
() are reported in parts per million (ppm) downfield from tetramethylsilane [(CH3)4Si] (TMS;
δ  0 ppm). Since TMS is not present in the solution, chemical shifts are experimentally
referenced to the center line of the residual proton signal in deuterated NMR solvent. Signals are
reported as follows: chemical shift (ppm), multiplicity (s, singlet; bs, broad singlet; d, doublet; t,
triplet; q, quartet; h, heptet; dd, doublet of doublet; m, multiplet), integration, coupling constant
(Hz). Data for 13C NMR are reported in terms of chemical shift (ppm) with 13C signal of the
deuterated solvent as reference. Melting points were measured by differential scanning
calorimetry
(Mettler-Toledo
differential
scanning
calorimeter
DSC2).
4bromo2fluorobenzonitrile
(CAS#105942-08-3),
isopropylhydrazine
hydrochloride
(CAS#16726-41-3), bromopyruvic acid (CAS#1113-59-3) and compound 2 (CAS#1040377-17-
ACS Paragon Plus Environment
22

Page 23 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
0) were commercially available and used as is. A% indicates area % purity by HPLC unless
otherwise noted throughout the experimental section and supporting information. See supporting
information for analytical methods, spectral and chromatographic data.
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
Synthesis of 2-(5-bromo-2-cyanophenoxy)ethan-1-aminium chloride (7•HCl): Ethanolamine
(41.6 kg, 681 mol, 1.1 equiv) was dissolved in 2-MeTHF (100 L) followed by addition of t-
BuOK (84.5 kg, 753 mol, 1.2 equiv) at 5 °C. The mixture was stirred for 40 minutes, and a
solution of 4bromo2fluorobenzonitrile (130.0 kg, 650 mol, 1.0 equiv) in 2-MeTHF (320 L)
was added. The mixture was stirred for 3 h until an IPC sample showed complete conversion
(target 4bromo2fluorobenzonitrile < 0.1 A%). Water (400 L) was added at 25 °C to quench
the reaction. The layers were separated and the lower aqueous phase was discarded. A solution
of 3N HCl in n-PrOH (272 kg) was added to the organic phase and the suspension was heated to
70 °C and stirred for 3 h at this temperature. The suspension was cooled to 20 °C and the product
was isolated by filtration and washed with 2-MeTHF (3 x 65 L). The product was dried at 45 °C
under reduced pressure until constant weight was attained. 7•HCl was isolated in 91% yield (164
kg) as a white powder (99.4 wt%, 99.4 A%). mp (DSC): 246.1 °C. HRMS (EI) m/z Calcd for
C9H9BrN2O: 239.9898; found: 239.9903. 1H NMR (600 MHz, DMSO-d6) δ ppm 8.29 (br s, 3H),
7.73 (d, J = 8.3 Hz, 1H), 7.60 (d, J = 1.7 Hz, 1H), 7.37 (dd, J = 8.3, 1.7 Hz, 1H), 4.43 (t, J = 5.2
13
Hz, 2H), 3.25–3.21 (m, 2H); C NMR (151 MHz, DMSO-d6) δ = 159.8, 135.2, 128.6, 124.8,
116.7, 115.7, 100.4, 66.1, 37.8.
Synthesis of 8-bromo-2,3-dihydrobenzo[f][1,4]oxazepin-5-aminium chloride (3•HCl):
7•HCl (74.5 kg, 268 mol, 1.0 equiv) was charged into a reactor and suspended in MeOH (300
L). The solution was cooled to 0 °C and Mg(OEt)2 (65.3 kg, 571 mol, 2.1 equiv) was added
ACS Paragon Plus Environment
23

Organic Process Research & Development
Page 24 of 36
1
2
3
4
5
6
7
8
9
within 60-90 minutes while keeping the temperature < 10°C (exothermic reaction: +15 °C
temperature rise observed in laboratory!). The thin brownish slurry was stirred for 1 h at 15 °C
and 2-MeTHF (600 L) was added. Then, the mixture was heated to 65 °C for about 45 h until in-
process control showed complete consumption of starting material (target 7 < 2.0 A%). The
slurry was concentrated to 360 L by distillation under reduced pressure with continuous addition
of 2-MeTHF. An in-process control sample from the reaction mixture confirmed the removal of
MeOH (target MeOH < 4.0 wt%). 2-MeTHF (525 L) was added, the mixture was cooled to 15
°C and a 5-6 N HCl in n-PrOH (310 kg) was added. The slurry was stirred for 4 h at 15 °C,
filtered and the cake washed with 2-MeTHF (670 L). The product was dried at 55 °C under
reduced pressure until constant weight is attained. 3•HCl was isolated in 87% yield (64.4 kg) as
a white powder (99.4 wt%; 99.4 A%). mp (DSC): 255.7 °C. HRMS (EI) m/z Calcd for
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
C9H9BrN2O: 239.9898; found: 239.9905. H NMR (600 MHz, DMSO-d6)  ppm 10.36 (br s,
1H), 9.50 (br s, 1H), 9.43 (br s, 1H), 7.67–7.58 (m, 2H), 7.51 (d, J = 1.0 Hz, 1H), 4.46 (t, J = 5.3
13
Hz, 2H), 3.52 (q, J = 5.2 Hz, 2H). C NMR (151 MHz, DMSO-d6) δ = 165.3, 156.4, 132.4,
128.9, 128.0, 126.5, 120.8, 77.5, 41.1.
Synthesis of 9-bromo-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine-2-carboxylic acid
(4): 3•HCl (61.9 kg, 223 mol, 1.0 equiv) was added to NMP (315 L), the mixture was heated to
50 C, and N,N,N',N'-tetramethylguanidine (122 kg, 1059 mol, 4.8 equiv) was added.
Bromopyruvic acid (48.4 k g, 290 mol, 1.3 equiv) was then added as a solution in 2-MeTHF (74
L). The reaction mixture was stirred until conversion was confirmed (target < 5 A% intermediate
(a) 13 ). The reaction slurry was transferred into a second vessel containing preheated water (945
L) at 50 C. The first vessel was rinsed with NMP (315 L) and transferred to the second vessel.
The combined solution was filtered through a charcoal pad (Norit A Supra, 6 kg) at 50 C to
ACS Paragon Plus Environment
24

Page 25 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
remove colored impurities. The second vessel and charcoal pad were rinsed with NMP (130 L)
and combined with the main solution. Then, 3 N HCl (170 kg) was added at 50 C within 2 h
(target pH ~2.5) to initiate product crystallization, and the mixture was cooled to 10 °C. Crude 4
was isolated by filtration and washed with water (750 L). The cake was dried at 85 C under
reduced pressure to provide crude 4 in about 94% yield (65.4 kg) as a light pink powder (96.8
wt%; 97.3 A%).
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
Crude 4 (63.5 kg, 205 mol, 1.0 equiv) was slurried in a mixture of NMP (64 L) and toluene (660
L), heated to 110 °C and toluene (65 L) was distilled off to remove water. Then, more toluene
(1015 L) was added and the resulting slurry was cooled to 10 C and filtered. The cake was
washed with toluene (580 L) and dried at 85 C under reduced pressure to provide 4 in 90%
yield (57.2 g) as a light pink white powder (>99.9 wt%, >99.9 A%). mp (DSC): 263.3 °C.
1
HRMS (EI) m/z Calcd for C12H9BrN2O3: 307.9797; found: 307.9805. H NMR (600 MHz,
DMSO-d6) δ ppm 13.08–11.46 (m, 1H), 8.31 (d, J = 8.7 Hz, 1H), 7.98 (s, 1H), 7.32 (dd, J = 8.7,
13
2.1 Hz, 1H), 7.27 (d, J = 1.9 Hz, 1H), 4.54–4.44 (m, 4H). C NMR (151 MHz, DMSO-d6) δ =
163.5, 155.9, 143.7, 132.6, 131.4, 128.9, 125.3, 123.1, 122.3, 117.2, 68.7, 49.5.
Synthesis of (Z)-N'-isopropylacetohydrazonamide hydrochloride (5): Isopropylhydrazine
hydrochloride (35.0 kg, 317 mol, 1.0 equiv) was suspended in MeCN (420 L) in an autoclave.
Gaseous HCl (17.5 kg, 480 mol, 1.5 equiv) was introduced at 40 °C. The reaction was heated to
70 °C for 36 h. After reaction conversion (monitored as the N-iPr-hydrazone derivative with
benzaldehyde < 8.0 A%), the reaction mixture was diluted with MeCN (210 L) and 210 L were
distilled off under atmospheric pressure. The operation was repeated once more to remove excess
HCl. MeOH (43 L) and MeCN (210 L) were added and the mixture was cooled to 25 °C. The
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 36
1
2
3
4
5
6
7
8
9
solid by-products were removed by filtration and rinsed with MeCN (35 L). The filtrate was
further concentrated by distillation at 65 °C under reduced pressure to a final volume of 380 L
(target MeOH < 1.0-1.5 wt% by GC). The suspension was cooled to 20 °C within 1 h and
product 5 was isolated by filtration, washed with MeCN (70 L) and then dried at 65 °C under
reduced pressure until constant weight was attained. 5 was isolated in 65% yield (29.0 kg) as a
white powder (98.7 wt%, >99.9 A%). mp (TGA): 162.8 °C. HRMS (EI) m/z Calcd for C5H13N3:
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
115.1109; found: 115.1112. H NMR (600 MHz, D2O) δ ppm 3.11 (dtd, J = 12.6, 6.3, 6.3, 0.7
13
Hz, 1H), 2.18 (d, J = 0.8 Hz, 3H), 1.07–0.98 (m, 6H). C NMR (151 MHz, D2O) δ = 165.8,
50.2, 19.3, 19.3, 15.2.
Synthesis
of
9-bromo-2-(1-isopropyl-3-methyl-1H-1,2,4-triazol-5-yl)-5,6-
dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepine (1): 4 (21.9 kg, 71 mol, 1.0 equiv) was dissolved
in NMP (240 L) and the solution was heated to 40 C. 1,1-Carbonyldiimidazole (CDI, 16.4 kg,
1.4 equiv) was added and the mixture was stirred until activation of the starting material was
verified by an HPLC IPC sample (target 4 < 3 A%). Compound 5 (15.3 kg, 108 mol, 1.5 equiv)
and K2CO3 (12.5 kg, 90 mol, 1.3 equiv) were added and the reaction mixture was heated to 85
C. The suspension was stirred at this temperature until complete consumption of the activated
starting material by an HPLC IPC sample was observed (target 10 < 0.50 A%). Then, the
mixture was cooled to 55 C, water (295 L) was added within 4.5 h and the suspension was
cooled to 10 C. The slurry was filtered, washed with water (200 L) and dried at 90 C under
reduced pressure to give 1 in 83% yield (22.8 kg) as a white powder (<99.9 wt%; <99.9 A%).
1
mp (DSC): 168.7 °C. HRMS (EI) m/z Calcd for C17H18BrN5O: 387.0695; found: 387.0687 H
NMR (600 MHz, DMSO-d6) δ ppm 8.32 (d, J = 8.7 Hz, 1 H), 7.95–7.90 (m, 1H), 7.35 (dd, J =
8.7, 2.0 Hz, 1H), 7.31–7.27 (m, 1H), 5.87–5.72 (m, 1H), 4.52 (s, 3H), 4.57–4.49 (m, 1H), 2.25 (s,
ACS Paragon Plus Environment
26

Page 27 of 36
Organic Process Research & Development
1
2
3
4
13
3H), 1.45 (d, J = 6.6 Hz, 6H). C NMR (151 MHz, DMSO-d6) δ = 158.3, 155.9, 147.1, 143.1,
5
6
131.1, 130.5, 125.4, 124.2, 123.1, 122.1, 117.2, 68.7, 49.9, 49.5, 22.3, 22.3, 13.8.
7
8
9
Synthesis of ethyl 2-methyl-2-(4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1H-pyrazol-
1-yl)propanoate (11): KOAc (7.4 kg, 75 mol, 1.2 equiv) and B2Pin2 (19.2 kg, 76 mol, 1.2 equiv)
were charged to reactor, followed by EtOH (130 L). 2 (16.4 kg, 63 mol, 1.0 equiv) was then
added, followed by an EtOH rinse (25 L) and the solution was degassed by lowering pressure 3
times to <100 mbar, followed by N2 refill. XPhos-Pd G2 catalyst (150 g, 0.2 mol, 0.003 equiv)
and XPhos ligand (180 g, 0.4 mol, 0.006 equiv) were added and the mixture was degassed once
again. The reaction mixture was heated to 77 C and stirred until reaction conversion was
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
indicated by an HPLC IPC sample (typically 16 h, target 2  0.35 A%). The resulting solution
was used directly in the next step. A sample of 11 was isolated by preparative HPLC for
characterization. mp (DSC): 80.5 °C. HRMS (EI) m/z: [M+H]+ Calcd for C15H26BN2O4:
1
309.1986; found: 309.1982. H NMR (600 MHz, CDCl3) δ ppm 7.89 (d, J = 0.6 Hz, 1H), 7.84
(s, 1H), 4.17 (q, J = 7.1 Hz, 2H), 1.85 (s, 6H), 1.33 (s, 12H), 1.20 (t, J = 7.2 Hz, 3H). 13C NMR
(151 MHz, CDCl3) δ = 172.2, 145.5, 134.1, 83.3, 83.3, 64.2, 61.8, 25.5, 25.5, 24.8, 24.8, 24.8,
24.8, 14.0.
Synthesis
of
ethyl
2-(4-(2-(1-isopropyl-3-methyl-1H-1,2,4-triazol-5-yl)-5,6-
dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepin-9-yl)-1H-pyrazol-1-yl)-2-methylpropanoate
12: 1M LiOH (19 L, 19 mol, 0.3 equiv) was added to the reactor containing 11, followed by
addition of 1 (22.0 kg, 57 mol, 0.9 equiv) previously dissolved in THF (95 L) and another
portion of 1M LiOH (107 L, 107 mol, 1.7 equiv). The reaction was stirred at 77 °C until reaction
conversion was indicated by an HPLC IPC sample (target 1  1.0 A%). The resulting solution
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 36
1
2
3
4
was used directly in the next step. A sample of 12 was isolated by preparative HPLC for
5
6
characterization. mp (DSC): 169.7 °C. HRMS (EI) m/z Calcd for C26H31N7O3: 489.2488; found:
7
8
1
489.2487. H NMR (600 MHz, CDCl3) δ ppm 8.51 (d, J = 8.4 Hz, 1H), 7.87 (dd, J = 4.0, 0.7
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
Hz, 2H), 7.63 (s, 1H), 7.31 (dd, J = 8.4, 1.8 Hz, 1H), 7.19 (d, J = 1.8 Hz, 1H), 5.93 (quin, J =
6.6 Hz, 1H), 4.56–4.49 (m, 2H), 4.48–4.42 (m, 2H), 4.21 (q, J = 7.1 Hz, 2H), 2.42 (s, 3H), 1.91
13
(s, 6H), 1.59 (d, J = 6.6 Hz, 6H), 1.24 (t, J = 7.2 Hz, 3H). C NMR (CDCl3, 151 MHz) δ =
172.2, 159.2, 156.0, 147.6, 144.6, 137.1, 134.7, 131.7, 130.7, 124.5, 122.4, 121.6, 120.2, 116.8,
116.1, 68.2, 64.5, 61.9, 50.6, 50.2, 25.5, 25.5, 22.4, 22.4, 14.0, 14.0.
Synthesis
of
2-(4-(2-(1-isopropyl-3-methyl-1H-1,2,4-triazol-5-yl)-5,6-
dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepin-9-yl)-1H-pyrazol-1-yl)-2-methylpropanoic acid
13: The previous solution was cooled to 55 C and 3.5 M LiOH (61 L, 214 mol, 3.4 equiv) was
added. The reaction was stirred at this temperature until reaction conversion was indicated by an
HPLC IPC sample (target 12  0.7 A%).
Work-up: the organic solvents were distilled off (residual volume 240 L) at 55 C under reduced
pressure and the precipitated lithium and borate salts were filtered off. The cake was washed
three times with warm water (100 L, 55C). The combined filtrates were extracted with i-PrAc
(3 x 75 L) at 55 C and the combined organic phases were discarded. n-PrOH (75 L) was added
to the aqueous phase, and the mixture was acidified by slow addition of 20% H2SO4 (15 kg) at
55 °C until the product crystallized (pH 5.2  5.8). The mixture wa stirred for at least 2 h at this
temperature and the pH was then further adjusted to 3.1 with 20% H2SO4 (40 kg) and the
resulting slurry was cooled to 20 C. The product was isolated by filtration, washed with water
(100 L) and dried at 85C under reduced pressure to provide 13 in 75% yield (22.6 kg) as a white
ACS Paragon Plus Environment
28

Page 29 of 36
Organic Process Research & Development
1
2
3
4
powder (96.1 wt%; >99.9% A%). mp (DSC): 234.8 °C. HRMS (EI) m/z Calcd for C24H27N7O3:
5
6
1
461.2175; found: 462.2186. H NMR (600 MHz, DMSO-d6)  ppm 12.99 (br d, J = 0.6 Hz, 1
7
8
9
H), 8.46 (d, J = 0.7 Hz, 1H), 8.36 (d, J = 8.4 Hz, 1H), 7.99 (d, J = 0.7 Hz, 1H), 7.89–7.86 (m,
1H), 7.45 (dd, J = 8.4, 1.8 Hz, 1H), 7.36 (d, J = 1.8 Hz, 1H), 5.87–5.78 (m, 1H), 4.52 (s, 2H),
4.52 (br d, J =11.7 Hz 2H), 2.26 (s, 3H), 1.78 (s, 6H), 1.47 (d, J = 6.5 Hz, 6H). 13C NMR
(CDCl3, 151 MHz) δ = 173.6, 158.2, 155.9, 147.4, 144.0, 136.5, 134.6, 130.3, 130.0, 126.1,
123.6, 120.4, 119.3, 116.2, 115.3, 68.3, 63.9, 49.9, 49.7, 25.1, 25.1, 22.3, 22.3, 13.8.
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 taselisib: Compound 13 (20.0 kg, 43 mol, 1.0 equiv) was dissolved in THF (220 L)
and the solution was heated to 65 °C. The THF was partially distilled off (residual volume 180
L), azeotropic distillation was completed with continuous addition of THF (275 L) to remove
water (target  0.07% water in distillate). The mixture was cooled to 25 C, CDI (9.8 kg, 60 mol,
1.4 equiv) was added, followed by a THF rinse (75 L). The slurry was stirred until a HPLC IPC
sample indicated complete conversion (target 13  0.30 A%). The resulting slurry was added to a
mixture of NH3 (7M in MeOH, 18.3 kg, 162 mol, 3.8 equiv) in THF (35 L) at 20 C, followed by
a THF rinse (35 L). The mixture was stirred for 1 h at 25 C and water (45 L) was added. The
mixture was concentrated by distillation under reduced pressure (residual volume ~100 L) and
then more water (140 L) was added within 1 h. The slurry was cooled to 20 C and crude
taselisib was isolated by filtration and washed with water (585 L). The cake was dried at 95 C
under reduced pressure to provide crude taselisib in 93% yield (18.8 kg) as a white powder (98.5
wt%; 99.7 A%).
Crude taselisib (18.0 kg, 39 mol, 1.0 equiv) was suspended in n-PrOH (145 L) and water (50 L)
and heated to 60 C to obtain a homogeneous solution. This solution was filtered via a charcoal
ACS Paragon Plus Environment
29