Palbociclib Commercial Manufacturing Process Development. Part I: Control of Regioselectivity in a Grignard-Mediated SNAr Coupling

Article abstract of DOI:10.1021/acs.oprd.6b00070

This is the first in a series of three papers describing commercial manufacturing process development for palbociclib (1). This manuscript focuses on the S_NAr coupling between aminopyridine 3 and chloropyrimidine 7. The regioselectivity of the S_NAr coupling was studied from a synthetic and mechanistic perspective. Grignard bases were identified as the preferred class of bases for this reaction, allowing for a simplified process and reduced usage factor for aminopyridine 3. The development of this S_NAr reaction into a scalable commercial manufacturing process is also described.

Full text of DOI:10.1021/acs.oprd.6b00070

Subscriber access provided by Hong Kong University of Science and Technology Library
Full Paper
Palbociclib Commercial Manufacturing Process Development. Part
I: Control of Regioselectivity in a Grignard-Mediated SNAr Coupling
Shengquan Duan, David Place, Hahdi H. Perfect, Nathan D. Ide, Mark Thomas Maloney,
Karen Sutherland, Kristin E. Price Wiglesworth, Ke Wang, Mark A. Olivier, Fangming Kong,
Kyle R Leeman, Jon Blunt, John Draper, Marie McAuliffe, Maria M O'Sullivan, and Denis Lynch
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.6b00070 • Publication Date (Web): 12 May 2016
Downloaded from http://pubs.acs.org on May 13, 2016
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
Palbociclib Commercial Manufacturing Process
Development. Part I: Control of Regioselectivity in
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
a Grignard-Mediated S Ar Coupling
N
a
a
a
a†
a
Shengquan Duan, * David Place, Hahdi H. Perfect, Nathan D. Ide ,Mark Maloney, Karen
a
a
a
a
a
Sutherland, Kristin E. Price Wiglesworth, Ke Wang, Mark Olivier, Fangming Kong, Kyle
a
b
b
c
c
c
Leeman, Jon Blunt, John Draper, Marie McAuliffe, Maria O’Sullivan, and Denis Lynch
a
Chemical Research and Development and Analytical Research and Development, Pfizer
Worldwide Research and Development, Eastern Point Road, Groton, CT 06340
b
Chemical Research and Development, Pfizer Process Development Facility, Ramsgate Road,
Sandwich, Kent, CT13 9NJ, UK
c
Pfizer Global Supply, Ringaskiddy, Cork County, Ireland
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table of Contents Graphic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment
2

Page 3 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
KEYWORDS: Palbociclib, S Ar Coupling, Grignard, Regioselectivity
N
ABSTRACT: This is the first in a series of three papers describing commercial manufacturing
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
process development for palbociclib (1). This manuscript focuses on the S Ar coupling between
N
aminopyridine 3 and chloropyrimidine 7. The regioselectivity of the S Ar coupling was studied
N
from a synthetic and mechanistic perspective. Grignard bases were identified as the preferred
class of bases for this reaction, allowing for a simplified process and reduced usage factor for
aminopyridine 3. The development of this S Ar reaction into a scalable commercial
N
manufacturing process is also described.
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
INTRODUCTION
Cyclin-dependent kinases 4 and 6 (CDK4/6) are key regulators of the cell cycle, involved in
1
a-d
cellular progression from growth phase (G1) into the phase associated with DNA replication.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Increased CDK 4/6 activity is frequently observed in estrogen receptor-positive (ER+) breast
2a-d
cancer (BC).
Palbociclib (1) (Figure 1) is a highly selective, reversible inhibitor of CDK
4
/6, intended to block tumor cell proliferation. Palbociclib received accelerated approval from
the United States Food and Drug Administration (FDA) in February 2015. It is used in
combination with letrozole for the treatment of postmenopausal women with estrogen receptor-
positive, human epidermal growth factor receptor 2-negative (ER+/HER2-) advanced breast
cancer as initial endocrine-based therapy. Palbociclib is marketed under the brand name
®
IBRANCE .
Figure 1. Structure of Palbociclib (1)
The evolution of the manufacturing process for palbociclib is shown in Schemes 1, 2, and 3. The
3
discovery chemistry synthesis of 6 (1-HCl salt) is shown in Scheme 1. While this route
allowed for the preparation of preclinical supplies, it posed two major issues for further scale-up.
ACS Paragon Plus Environment
4

Page 5 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
First, the coupling of 2 and 3 was problematic, with a yield at only 38%. Second, the Stille
coupling required an expensive, toxic vinyl tin reagent.
Scheme 1. Discovery Route to Palbociclib
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
Nevertheless, the discovery chemistry route is highly convergent, and the last two steps had good
chemical conversions. Thus, it became the basis for the initial process chemistry route (Scheme
4
2). The highlights of this route include: (1) coupling of 3 and 7 using a strong base (LHMDS),
which greatly improved the yield; (2) replacement of the Stille coupling with a Heck coupling,
which eliminated the need for the vinyl tin reagent; (3) isolation of the API as an isethionate salt,
with isethionic acid used to facilitate the hydrolysis of the enol ether and removal of the Boc
protecting group. Although this early process chemistry route was amenable to scale-up, there
were still a few aspects that were not ideal for commercial manufacturing. First, more than two
equivalents of 3 were required for good yield in the S Ar coupling. Second, lack of
N
5
regioselectivity in the Heck reaction resulted in challenges for impurity control. Finally, the
physical properties of the isethionate salt were not suitable for drug product manufacturing,
which resulted in a final form change to the API free base.
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 2. Clinical Manufacturing Route to Palbociclib
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
During commercial route development, a variety of new synthetic routes were evaluated.
These approaches include modification of step order of the enabling synthesis, synthesis of the
7
penultimate by palladium-catalyzed amination , construction of the pyridinone ring through
8
9
10
Friedländer annulation , Larock annulation reaction , and Knorr cyclization . However, none
of these approaches were cost effective compared to the enabling route, so the decision was
made to further develop the enabling route for commercial use. A thorough process
development program was carried out to improve the scalability, efficiency, and quality of the
enabling process. The final proposed commercial manufacturing process for palbociclib is
illustrated in Scheme 3.
Scheme 3. Commercial Manufacturing Process for Palbocicilib
ACS Paragon Plus Environment
6

Page 7 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
Herein we publish a series of three papers that document our journey to develop the commercial
manufacturing process for palbociclib. This paper (Part I) focuses on the development of the
1
1
Grignard-mediated S Ar coupling between compound 3 and compound 7. Part II describes
N
1
2
the installation of the enol ether side-chain by Heck coupling. Part III describes the final
deprotection step and the API particle engineering efforts.
RESULTS AND DISCUSSION
The nucleophilic aromatic substitution (S Ar) reaction between the amino pyridine compound 3
N
and the pyrimidine compound (2, 10, 7) is shown in Scheme 4. This reaction looks fairly
straightforward, but it involves a relatively complicated mechanism. Extensive development
work was completed in order to identify suitable conditions for this coupling, and to optimize the
conditions for process robustness.
Scheme 4. S Ar Coupling Reaction
N
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Selection of the Pyrimidine Coupling Partner
Different leaving groups (X, Scheme 4) were evaluated for the pyrimidine component. For the
preparation of the first clinical supplies, the sulfoxide compound 2 was used. A variety of
1
3
reaction conditions had been examined, and the best yield (38%) was obtained when the two
compounds were reacted in refluxing toluene. Change of solvents and reaction temperature did
not improve the yield. Fortunately, the product 4 precipitated directly from the reaction mixture
1
4a-c
with acceptable purity. Change of the leaving group to sulfone
(compound 10) improved the
reactivity, but only marginally improved the yield to 50%. Besides the low yield encountered in
the coupling with 2/10, the syntheses of 2/10 were lengthy, requiring 10 and 11 linear steps
1
5
respectively. To overcome the difficulty to access these compounds, an analogous compound
7
was identified. Synthesis of 7 is much more efficient, as it is accessible in four steps from
1
6
commercially available dichlorobromopyrimidine 11 (Scheme 5). Thus, 7 was selected as
starting material for palbociclib synthesis and the subsequent development has been focused on
coupling between 3 and 7.
Scheme 5. Synthesis of 7
ACS Paragon Plus Environment
8

Page 9 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
Alternative Approaches for Coupling of 3 and 7
To address the issues encountered in the enabling synthesis, many alternative approaches for the
coupling between 3 and 7 were investigated. These conditions include: palladium catalyzed
coupling, S Ar coupling under neutral conditions, acidic conditions, and weak basic conditions,
N
and fluoride catalyzed coupling. While all these conditions failed to meet our requirements,
studies on these reactions provided insight on the reaction mechanism and helped us identify the
optimal reaction conditions using strong base.
1. Palladium Catalyzed C-N Coupling
Palladium catalyzed C-N cross coupling had been investigated to couple 3 with 7 following
1
7a-b
literature procedures.
A few catalysts (Pd(OAc) , Pd (dba) , and POPd), ligands (BINAP,
2 2 3
DavePhos, and JohnPhos), bases (NaOt-Bu, K PO , and Cs CO ), and solvents (toluene, DME)
3
4
2
3
were screened (Figure 2.). While most of these reactions gave very low yield of the desired
product, the palladium phosphinous acid catalyst (POPd) provided the best yield (40%). Further
optimization of reaction conditions focusing on POPd catalyst failed to improve this yield. This
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
coupling method was deemed to be impracticable considering the cost associated with the
catalyst and the modest yield.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Structure of POPd, DavePhos and JohnPhos
2
. S Ar Coupling of 3 and 7 under Neutral and Acidic Conditions
N
In the course of our investigation of alternate conditions for this S Ar coupling, we also
N
evaluated the reactivity under neutral conditions. Coupling of 3 with 2/10 required heating
because of the weak nucleophilicity of the amino pyridine. Compared to the sulfone and
sulfoxide compounds, the chloride compound 7 is less reactive, which makes this transformation
even more challenging. In fact, no desired product was observed when 3 was combined with 7
under neutral conditions, even in refluxing toluene. The ambident reactivity of 3 turned out to be
1
8
the biggest challenge for this coupling. Aminopyridine 3 is an ambident nucleophile that can
react at either the 2-amino group or the pyridine ring nitrogen. In fact, the undesired
regioisomer 16 was dominant under neutral conditions, suggesting that the pyridine nitrogen was
more nucleophilic (Scheme 6).
Scheme 6. Regioselectivity Issue in Coupling of 3 and 7
ACS Paragon Plus Environment
1
0

Page 11 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
9
Acid catalysis was first explored to re-direct the selectivity . It was envisioned that the amino-
nitrogen would be available for nucleophilic substitution after the more basic pyridine nitrogen
was blocked by forming a pyridinum salt with acid (the calculated pKa of the conjugated acid for
the pyridine nitrogen and amino nitrogen was 8.2 and 3.1 respectively). To that end, a few
hydrochloride salts of 3, with varied amount of counter-ion (1.0 equiv. 2.0 equiv. and 3.0 equiv.),
were prepared, and subjected to the coupling with 7. It turned out that this approach did not alter
the coupling pathway at all. While the reaction proceeded very slowly, the pyridinum salt 16
was obtained without exception in all of the experiments.
3. S Ar Coupling of 3 and 7 under Weak Basic Conditions
N
The problematic ambident reactivity was also observed when weak bases were used to assist the
reaction. Table 1 listed the results of a few coupling experiments under weakly basic conditions.
As can be seen from this table, none of the organic bases resulted in desired product, although
some of them did react with 7. When pyridine, Hunig’s base, and N-methylmorphline were used
ACS Paragon Plus Environment
1
1

Organic Process Research & Development
Page 12 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
entry 1), the reaction proceeded slowly to give the pyridinum side product 16. This outcome is
similar to when the reaction was run under neutral or acidic conditions. For the more
2
0
nucleophilic bases, such as DMAP, imidazole, and DABCO (entries 2, 3, 4), the bases reacted
with 7 but the resulting compounds did not serve as a reactive intermediate. In the reactions with
DMAP and imidazole, the intermediates were stable and no further reaction with 3 was observed.
In the case of the DABCO reaction, the liberated chloride from the substitution reaction attacked
and opened the DABCO ring. None of these results with weak organic bases were considered to
warrant any additional investigation. The inorganic bases gave similar results (Entry 5). When
carbonates (K and Cs) were used, the pyridinum product 16 was again obtained.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 1. Results of S Ar Coupling under Weak Basic Conditions
N
Desired Product
Entry
Base
Major Side Product
Yield
1
Pyridine, Hunig’s base, N-
<5%
methylmorphline
2
3
DMAP
<5%
<5%
Imidazole
4
DABCO
<5%
ACS Paragon Plus Environment
1
2

Page 13 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
5
K CO , Cs CO
3
<5%
2
3
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
4
. S Ar Coupling of 3 and 7 under Fluoride Catalyzed Conditions
N
Fluoride-catalyzed amination was also investigated based on literature reports suggesting that
21
fluoride catalysis might be effective for this type of reaction. The accepted mechanism for this
type of reaction is that the chloride electrophile is converted to the corresponding fluoride
analog, followed by reaction with the nucleophile. However, when compound 3 and 7 were
heated with CsF and KF in presence of crown ethers (18-C-6 or 15-C-5 respectively), a
complicated mixture was obtained in both reactions with only small quantity of desired product.
22
The conditions identified by Senanayake et al in a similar S Ar coupling were also
N
investigated. In that study, the monomethylpolyglyme solvent (TGME) played a critical role to
+
activate the CsF by coordinating with Cs and hydrogen bonding to the naked fluoride ion.
However, this reaction failed to give any desired product in our case. The major product isolated
from this reaction was identified as the displacement product in which the chloride was displaced
by the TGME solvent (Compound 20, Figure 3). This suggested that the amino pyridine
compound was less nucleophilic than the alcohol under fluoride catalysis conditions.
ACS Paragon Plus Environment
1
3

Organic Process Research & Development
Page 14 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Product Formed under Fluoride Catalyzed Reaction
SNAr Coupling under Strong Basic Conditions
The HOMO calculation of compound 3 at neutral state indicated that the nucleophilicity of the
pyridine nitrogen and the amino nitrogen was comparable (Figure 4). This explained why the
reaction with 3 showed ambident reactivity. Formation of the undesired regioisomer (16) in
most cases further suggested that the nucleophilic reaction with the pyridine nitrogen was
kinetically favored. When the similar calculation was done on the deprotonated 3, it showed that
the nucleophilicity can be differentiated between these two nitrogens, with the exocyclic nitrogen
bearing much higher electron density. Considering that this would lead to the desired
regioselectivity, we turned our attention to coupling with strong bases.
Figure 4. HOMO Calculation of 3 and Deprotonated 3
ACS Paragon Plus Environment
1
4

Page 15 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
1
. S Ar Coupling: LiHMDS Base
N
LiHMDS was found during early process development work to produce the desired coupling
product in good yield. When 3 was treated with LiHMDS, the lithiated amide reacted readily
with 7, at ambient temperature, to give the desired product. However, a new problem was
introduced with the use of strong base because the product is significantly more acidic than the
starting material. When less than 2 equivalents of LHMDS were used, the reaction stalled at
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
<
50% conversion (Table 2, entry 1). Simply adding a second equivalent of LiHMDS (Table 2,
entry 2) resulted in higher conversion (starting material consumption) but not higher yield since
the starting material and the product were not stable to excess LiHMDS (vide infra). The
reaction worked the best when 2+ equivalents of aminopyridine 3 and 2+ equivalents of LHMDS
were used at the same time (Table 2, entry 3).
Table 2. S Ar Coupling with LiHMDS as Base
N
Stoichiometry of
Entry
Stoichiometry of 3
Yield
LiHMDS
1
2
3
1.5
1.6
1.6
2.2
27.6%
53.7%
85%
2.5
2.5
While the conditions in entry 3 were effective and were used to produce material for clinical
supplies, the requirement for 2+ equivalents of aminopyridine 3, an advanced intermediate, was
not ideal. Additional investigations into stoichoimetry and addition modes did not reveal any
additional improvements. Investigations also showed that NaHMDS and KHMDS were inferior
to LiHMDS, providing a more complex purity profile. In order to reduce the usage factor of the
advanced intermediate 3, mechanistic work was carried out to understand why the second
ACS Paragon Plus Environment
1
5

Organic Process Research & Development
Page 16 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
equivalent of 3 was necessary. Two on-line analytical tools were utilized to collect the required
information. First, in-situ ReactIR monitoring of the reaction progress provided some insight
into the reaction mechanism (Figure 5). Based on the IR data, deprotonation of aminopyridine 3
was instantaneous upon addition of LiHMDS. Subsequent addition of 7 resulted in the formation
of a small amount of desired product 4 and a larger amount of an intermediate that slowly
converted to desired product 4. Using a Mettler Toledo Easy Sampler® probe, a tool that can
take multiple samples in short period, we were able to capture samples containing this
intermediate. Structure elucidation, performed on samples that had undergone an aqueous
quench, confirmed that the isolated intermediate was the pyridinum salt 16. It is believed that
the actual intermediate under strong basic conditions bears the imino form 21, which converts to
the protonated form 16 upon quenching.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. ReactIR Monitoring of the LiHMDS Reaction
Scheme 7. Reaction Intermediate before and after Aqueous Quench
ACS Paragon Plus Environment
1
6

Page 17 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
The discovery of this intermediate suggested that while the electron density on the amino
nitrogen increased after deprotonation by strong base, the reaction between the pyridinyl
nitrogen and 7 was still kinetically favored. This also suggested that the coupling between 3 and
7
was not a simple S Ar displacement, but involved a more complicated mechanism.
N
We initially consider a bimolecular mechanism that involved reaction of 21 with a second
equivalent of deprotonated 3 (Scheme 8). In this mechanism, the imino-intermediate 21 would
first be formed in the reaction then a second equivalent of lithiated 3 would react with the imino
intermediate in another S Ar reaction, liberating the excess 3 and generating the desired product.
N
Scheme 8. Initially Proposed Reaction Mechanism between 3 and 7
ACS Paragon Plus Environment
1
7

Organic Process Research & Development
Page 18 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
This mechanism would explain why the second equivalent of 3 was required in the coupling
reaction. However, when the reaction conversion data was modeled using kinetics modeling
nd
software DynoChem®, it was found that the proposed 2 order reaction mechanism did not fit
st
the data. Instead, the data fitting supported a model of 1 order reaction for the conversion of 21
to 4 (Figure 6).
ACS Paragon Plus Environment
1
8

Page 19 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
2
1
00
50
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
Compound 4
Intermediate 21
Compound 7
100
Compound 3
Compound 4 (fit)
Intermediate 21 (fit)
Compound 7 (fit)
Compound 3 (fit)
5
0
0
0
20
40
60
80
100
120
time, min
Figure 6. DynoChem Modeling of LiHMDS Reaction
st
To further confirm the 1 order reaction mechanism, a series of experiments at -20 °C, 0 °C, and
TM
2
0 °C were performed and the in-situ IR data were exported to the iC Kinetics software for
kinetic modeling. The kinetic simulation confirmed that the decay of 21 was indeed first order,
nd
and the 2 equiv. of lithiated 3 did not participate in the rate determining step (Figure 7).
ACS Paragon Plus Environment
1
9

Organic Process Research & Development
Page 20 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Kinetics Study at Different Temperatures
The above studies suggested that the conversion of the imino intermediate 21 to the desired
product was through an intramolecular rearrangement process. Multiple mechanistic pathways
2
3a-c
24
are possible, but based on the work of Chapman
and Granik , we suspect that the reaction
mechanism is a [1,3] aryl shift via a four-membered ring transition state (Scheme 9).
Scheme 9. Rearrangement Mechanism
ACS Paragon Plus Environment
2
0

Page 21 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
While the newly proposed mechanism fits better with the kinetics, it could not explain why the
second equivalent of 3 was required to promote the yield (and not simply a second equivalent of
LiHMDS). Careful study on the reaction profile identified one major impurity in the LiHMDS
reaction mixture. This impurity has a molecular weight of 1174. Isolation and structure
elucidation identified this impurity as a dimer of two product molecules (Compound 22, Figure
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
8
). The structure of this impurity suggested that the methyl proton was deprotonated under
strong basic conditions and then underwent a possible 1, 6-addition to another product molecule.
Exposure of product 4 to LiHMDS indeed reproduced the formation of this dimer. Similar
degradation pathways were observed upon exposure of the chloropyrimidine 7 to LiHMDS.
Figure 8. Structure of Dimer Impurity 22
The above observation explained the function of the second equivalent of 3. It served as a buffer
to consume the second equivalent of LiHMDS. The lithiated 3 was basic enough to deprotonate
the acidic proton in the product, which drove the reaction to completion, but it was not strong
ACS Paragon Plus Environment
2
1

Organic Process Research & Development
Page 22 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
enough to abstract the proton in the methyl group, which was the root cause of the dimer
formation. Without the second equivalent of LiHMDS, the reaction would stall. Without the
second equivalent of 3, the excess LiHMDS would react with the product and/or 7, which
resulted in impurity formation and low yield.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2. S Ar Coupling: Grignard Base
N
On the basis of the mechanistic understanding, we decided to investigate a few other strong
bases, attempting to find a strong base that does not promote the dimer (22) formation.
While LiH and NaH failed to deprotonate 3, tertiary alkoxide bases gave some interesting results.
Based solely on pKa value, it was not expected that the alkoxide base would be able to
deprotonate 3. However, a complete consumption of 7 was indeed observed when NaOt-Bu,
KOt-Bu, and NaOt-Am (2 equiv. of base with 1 equiv. of 3) were tested. The yields of these
reactions were disappointedly low at 30-40% though. The major impurity was due to reaction
of water with 7 to give hydroxypyrimidine (23, Figure 9). When rigorous drying was applied on
the starting materials and solvents, and the reaction was handled completely in the glove box,
this impurity was minimized and up to 87% isolated yield was obtained with 2 equivalents of
NaOt-Bu and 1.3 equivalents of 3. However, such rigorous exclusion of water was obviously not
practical at manufacturing scale. Note that impurity 23 was not observed in the LiHMDS
reaction. It is believed that the major reaction pathway in the alkoxide reaction was slow, which
provided the opportunity for the adventitious water to compete.
ACS Paragon Plus Environment
2
2

Page 23 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Figure 9. Structure of Hydroxypyrimidine 23
Another type of base that showed promise was the Grignard base. Reaction with i-PrMgCl (2
equiv. base and 2 equiv. 3), gave comparable yield as the LiHMDS reaction. The reaction
profile was cleaner with no dimer impurity detected. A test reaction, with the use of only 1.1
equiv. of 3 and addition of 2 equiv. of i-PrMgCl in two portions, led to the first successful
reaction in which the usage factor of 3 was reduced to nearly 1. The initial 81% isolated yield
was enhanced to 94% when optimal stoichiometry was identified: 1.3 equiv. of 3 and 2.2 equiv.
of i-PrMgCl.
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
We initially suspected that the improved reaction profile might be due to increased stability of
the product 4 in the presence of the Grignard base vs. LiHMDS, but the follow-up experiments
showed that this was not the case. While exposure of product to i-PrMgCl did not generate the
dimer impurity, it resulted in the rapid formation of des-bromo impurity 24 (Figure 10) via
metal-halogen exchange reaction. Interestingly, this side reaction between 4 and i-PrMgCl was
kinetically faster than the side reaction between 4 and LiHMDS.
Figure 10. Structure of Des-Br Impurity 24
ACS Paragon Plus Environment
2
3

Organic Process Research & Development
Page 24 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Kinetic study of the i-PrMgCl reaction was carried out to understand the difference from the
LiHMDS reaction. Interestingly, the imino intermediate 21 was not detected by ReactIR in the
Grignard reaction. Off-line analysis indicated very low level of this compound throughout the
process. There are two possible scenarios: (1) the reaction mechanism completely changed and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
the reaction went through a direct S Ar displacement or (2) the reaction mechanism was the
N
same but the conversion of intermediate 21 to desired product was so fast that the concentration
of the intermediate was always low and below the ReactIR detection limit. There is no solid
evidence to prove one way or another. Either way, it can be concluded that the kinetics in the i-
PrMgCl reaction were much faster than the LiHMDS reaction. As seen in Figure 11, the reaction
went to completion at the moment the Grignard dosing was completed. Because of the fast
kinetics of the main reaction, the second equivalent of the Grignard reagent was quickly
consumed by the newly generated product, since the product has a more acidic proton than
starting material. Although the metal-halogen exchange side reaction with Grignard was fast, it
became non-competitive due to lack of the presence of the Grignard in the reaction mixture.
Figure 11. Kinetics Study of Grignard Reaction at 0 °C
ACS Paragon Plus Environment
2
4

Page 25 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
It was later found that the addition mode in the Grignard reaction was less critical than the
LiHMDS reaction. In the LiHMDS reaction, the base had to be added first to deprotonate 3,
before compound 7 was charged, to avoid the side reaction. Otherwise it would result in
impurity formation. For the i-PrMgCl reaction, the base can be added before addition of 7, in
two portions before and after addition of 7, or added directly to a mixture of 3 and 7. The latter
charging mode was selected because it greatly streamlined the operation on scale. Again, the
fast kinetics provides the advantage for the Grignard reaction so the more convenient addition
mode is feasible.
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
Process Development
1
. Off-gassing Issue Associated with the Use of i-PrMgCl
With respect to selection of the Grignard reagent, we had initially selected i-PrMgCl because it is
readily available at large scale. Since the consumption of i-PrMgCl produces propane gas, we
needed to carefully consider the safety consequences. Our experiments showed that the
dissipation of propane was dictated by the solubility of propane in THF, so slow dosing of
Grignard reagent did not solve the issue. The propane gas accumulated in the reaction solvent
and was released during the heating of the reaction mixture. A stage-wise heating protocol was
defined to mitigate this off-gassing issue and this approach worked out well on pilot scale. It
was determined, however, that this propane off-gassing would become unsustainable as the scale
increased. Thus, a screening of different Grignard reagents was performed to identify a new
Grignard without the off-gassing issue. The results of this effort are summarized in Table 3.
While several Grignard reagents worked well, cyclohexylmagnesium chloride (CyMgCl) in THF
was selected after taking into consideration the cost and the commercial availability.
Table 3. Grignard Base Screening Results
ACS Paragon Plus Environment
2
5

Organic Process Research & Development
Page 26 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Crude
Yield
Product
Entry
Grignard
Product Assay
Purity
99.0%
99.5%
99.5%
99.2%
98.0%
99.7%
97.0%
92.8%
97.1%
1
2
3
4
5
6
7
8
9
CyclohexylMgCl (THF)
CyclohexylMgCl (2-MeTHF)
CyclopentylMgCl (2-MeTHF)
t-ButylMgCl (THF)
86%
78%
100.0%
100.2%
99.5%
100%
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
86%
70%
p-TolylMgCl (THF)
81%
93.3%
100.0%
84.8%
85.2%
89.5%
HexylMgCl (THF)
73%
IsopentylMgCl (Diethyl Ether)
CyclohexylMgCl (Toluene/THF)
CyclohexylMgBr (THF)
74%
>100%
>100%
2. Filtration Challenge
One of the major challenges with this process was the slow filtration of the product 4. In the
LiHMDS process, the reaction mixture was quenched with aqueous NH Cl and the product
4
precipitated promptly during the aqueous quench. The filtration of the triphasic mixture
(
aqueous/organic/solid) was extremely slow, taking several days at 15 kg scale. The isolated
solids were observed to have a fine needle morphology (Figure 12). Polymorph screening did
not identify any addition solid forms.
ACS Paragon Plus Environment
2
6

Page 27 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 12. Typical Crystal Morphology Observed for 4
In addition, this compound has a very limited solubility in a broad spectrum of solvents. In most
of solvents, the solubility is <1 mg/mL. Thus, the precipitation of the product was spontaneous
upon the quench with water and development of a controlled crystallization process was not
possible. In the enabling process, THF/toluene was used as reaction solvent and the filtration
was very challenging. Considering that the THF is one of the better solvents for the product
(albeit giving a solubility of 1.5 mg/mL at ambient temperature), it was decided to use pure THF
as reaction solvent to relieve the over-saturation to some extent. Temperature cycling was also
attempted to improve the aspect ratio and physical properties of the crystals, but this resulted in
slower filtration due to particle attrition during the temperature cycling. Fortunately, the reaction
mixture had acceptable stability at 60 °C, and reaction quench at 60 °C allowed for some control
of the precipitation. The solvent change to pure THF and increased quench temperature
improved the filtration rate four-fold, but the filtration was still slow.
The change of the base from LiHMDS to i-PrMgCl also required a change of the quenching
protocol, because NH Cl generated poorly soluble magnesium salts, and these species resulted in
4
problematic emulsions. Acetic acid, citric acid, hydrochloric acid, and tartaric acid were
ACS Paragon Plus Environment
2
7

Organic Process Research & Development
Page 28 of 36
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
evaluated for the quench and acetic acid was deemed to be optimal. A quench protocol with 0.75
M aqueous HOAc (8.5 vol) was developed and worked well at lab scale. However, when this
process was scaled up, a bad emulsion was again observed. Experiments aiming at breaking up
the emulsion using 10% aqueous NaCl or 10% aqueous Rochelle’s salt successfully suppressed
the emulsion, but resulted in precipitation of large quantity of inorganic salts in the product
stream.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Due to the emulsions, the product was contaminated with insoluble magnesium salts, which
resulted in difficult drying of the product, and water was determined to be detrimental to the
subsequent Heck reaction. Aqueous washes of the filter cake, both displacement washes and
slurry washes, were ineffective in reducing the inorganic content. The inorganic could only be
reduced by reslurrying the crude product in a THF/aqueous acetic acid mixture. In the process of
searching for a solution to this issue, a counter-intuitive approach was discovered. It was found
that when the reaction mixture was quenched with acetic acid diluted with THF, in the absence
of water, all of the inorganic materials remained in the mother liquor and the isolated product
had a nearly 100 % assay. While the filtration was not complicated by the co-precipitation of
the inorganics, this non-aqueous quench protocol also resulted in a much better filtration. It is
believed that the solvent viscosity played some role in the filtration improvements. The THF-
rich mother liquors are less viscous than the THF/water mixture used in the earlier processes.
3. Impurity Control Strategy
The isolation process effectively purges most of the process related impurities. Among all of the
impurities, the des- bromo impurity 24 is the only impurity that progresses through the
subsequent steps and has potential to impact drug substance quality. Two possible pathways
ACS Paragon Plus Environment
2
8

Page 29 of 36
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
contribute to the formation of the des-bromo impurity (Scheme 10), both due to metal halogen
exchange with Grignard bases.
Scheme 10. Formation of des-bromo impurity 24
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
Multivariate design of experiment (DoE) studies were used to evaluate the impact of process
parameters on the level of des-bromo impurity. Statistical analysis of reaction data showed that
the reaction temperature has a significant impact on des-bromo level, which increases linearly as
temperature increases. Significant interaction between the stoichiometry of 3 and
cyclohexylmagnesium chloride stoichiometry was also detected. As shown in the contour plot
below (Figure 13), elevated levels of des-bromo are observed at high reaction temperature, low
ACS Paragon Plus Environment
2
9