From Milligram to Kilogram Manufacture of AZD4573: Making It Possible by Application of Enzyme-, Iridium-, and Palladium-Catalyzed Key Transformations

Article abstract of DOI:10.1021/acs.oprd.1c00058

With the first generation medicinal chemistry synthesis as a starting point, we describe herein process development of AZD4573, an oncology drug candidate. In addition to improved yields and removal of chromatographic steps, we have addressed other factors such as availability of starting materials as well as safety of the chemistry involved. With several steps involving volatile, reactive, and non-UV active materials, reaction optimization was facilitated by implementing off-line 1H NMR analysis of crude mixtures. Key transformations targeted for process development included a Wolff-Kishner reduction, an iridium-catalyzed borylation, and enzymatic resolution of a racemic amino-ester.

Full text of DOI:10.1021/acs.oprd.1c00058

pubs.acs.org/OPRD
Article
From Milligram to Kilogram Manufacture of AZD4573: Making It
Possible by Application of Enzyme‑, Iridium‑, and Palladium-
Catalyzed Key Transformations
Staffan Karlsson,* Helen Benson, Calum Cook, Gordon Currie, Jerome Dubiez, Hans Emtenäs,
Janet Hawkins, Rebecca Meadows, Peter D. Smith, and Jeffrey Varnes
Cite This: https://doi.org/10.1021/acs.oprd.1c00058
Read Online
ACCESS
Metrics & More
Article Recommendations
*
sı Supporting Information
ABSTRACT: With the first generation medicinal chemistry synthesis as a starting point, we describe herein process development of
AZD4573, an oncology drug candidate. In addition to improved yields and removal of chromatographic steps, we have addressed
other factors such as availability of starting materials as well as safety of the chemistry involved. With several steps involving volatile,
1
reactive, and non-UV active materials, reaction optimization was facilitated by implementing off-line H NMR analysis of crude
mixtures. Key transformations targeted for process development included a Wolff−Kishner reduction, an iridium-catalyzed
borylation, and enzymatic resolution of a racemic amino-ester.
KEYWORDS: iridium borylation, enzymatic resolution, acetylation, Suzuki, Wolff−Kishner, ammonolysis
4
INTRODUCTION
Wolff−Kishner reduction provided the known compound 4,
and after treatment with either NBS or NIS, building block 5 was
obtained. This was followed by palladium-catalyzed borylation
to give 6, which was purified by chromatography.
■
Cyclin-dependent-kinases (CDKs) are members of the Ser/Thr
family and are classified into two groups: those involved in cell
cycle control and those involved in transcription regulation/
RNA processing. Loss of cell cycle control through deregulation
of CDK activity is a key aspect of tumorigenesis. Cyclin-
dependent-kinase 9 (CDK9) is one of the most studied
transcriptional CDKs, the inhibition of which induces apoptosis
in a variety of cell lines and antitumor efficacy in vivo through
transcriptional downregulation of short-lived proteins such as
Building block 8 was either purchased or could be prepared
from 2-fluoropyridine 7 through ammonolysis in the presence of
concentrated aqueous ammonium hydroxide under microwave
conditions to afford the corresponding 2-aminopyridine 8 in
excellent yield after ion-exchange chromatography. Building
block 10 was obtained in a three-step process from commercially
1
,2
5
MCL-1. We recently reported the discovery of CDK9
available racemic amino acid 9. Acid 10 was coupled with 7
inhibitor AZD4573, which is currently in Phase 1 for
hematological malignancies.
using Ghosez’s reagent to afford 11 in moderate yield. Suzuki
coupling of 6 with 11 using second Generation XPhos
precatalyst was followed by chromatography to provide 12.
Deprotection under acidic conditions, purification of the
resulting amine by ion-exchange chromatography, and acetyla-
tion with acetic anhydride using either pyridine or triethylamine
as base then gave AZD4573 (after chromatography) in
moderate yield.
Although the route in Scheme 1 was efficient in terms of rapid
synthesis of compound libraries for evaluation of candidate
drugs, for application toward kilogram scale supply of AZD4573
we needed to address concerns regarding safety, solubility, and
purification. We therefore investigated route modifications and
alternatives based on the following assessments:
3
Retrosynthetically, AZD4573 can be obtained from the three
key building blocks A−C via a Suzuki reaction and an amide
coupling, respectively (Figure 1). For large scale applications,
while the convergent strategy used by the medicinal chemists to
achieve the synthesis of AZD4573 was appealing, the routes to
produce the various building blocks were not straightforward for
scale-up. Therefore, despite the need for rapid delivery of API
for both preclinical and clinical studies, we devoted time to
identify alternative routes and optimal conditions for assembly
of building blocks A−C.
The detailed synthesis of AZD4573 used by medicinal
chemistry (referred to as the first-generation synthesis) is
depicted in Scheme 1. Boronate building block 6 was prepared in
six steps starting with alkylation of pyrazole with commercially
available ethyl ester 1 in DMA using cesium carbonate as base.
After purification by chromatography, hydrolysis with sodium
hydroxide and filtration/trituration with aqueous acid (pH 3)
afforded acid 2. Following thorough drying, 2 was treated with 2
equiv of n-butyllithium (not titrated) in 2-methyltetra-
hydrofuran to afford diaza-bicycle 3 in moderate yield. A
Special Issue: Excellence in Industrial Organic Syn-
thesis 2021
Received: February 16, 2021
©
XXXX American Chemical Society
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
A

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 1. Retrosynthetic disconnection of AZD4573.
Scheme 1. First Generation Synthesis of AZD4573 (Purification Methods Are Italicized)
(
ii) The ammonolysis of the fluoropyridine 7 constituted a
(
i) A Suzuki coupling very late in the synthetic route resulted
in unacceptable levels of residual palladium in final API. A
change in order of the steps might also have a favorable
outcome on the solubility of intermediates.
hazardous operation.
(
iii) Although the Wolff−Kishner reduction of 3 could be used
for large scale manufacture, a more efficient catalytic
method was desirable.
B
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 2. Comparison of the First Generation Synthesis of BPin Ester 6 (Method A and B) with the Second Generation Synthesis
(Method C)
1
Figure 2. H NMR recording in CDCl of the crude mixture of (i) lithium−halogen exchange of 5a/triisopropylborate formation 13 (purple) and (ii)
3
pinacole trans-esterification to 6 (green).
(
iv) Racemic amino acid 9 was only available in limited
(vii) The highly reactive Ghosez reagent used for the amide
coupling is not accessible in large quantities, and handling
is nontrivial.
(viii) We believed that the removal of the BOC group of 12
followed by acetylation added two unnecessary steps to
the sequence, and we wanted to evaluate an alternative
convergent strategy where the acetyl group could function
also as a protecting group.
quantities.
(
v) The Pd-catalyzed borylation of compound 5 was
inefficient and low yielding, mainly due to proto-
dehalogenation as a side reaction.
vi) Several steps included chromatography for purification of
intermediates. We aimed for a process where crystal-
lizations could be employed instead.
(
C
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
In the following sections, we describe our efforts to address
the aforementioned issues identified in the first generation
synthesis and the development of a route (referred to as the
second generation synthesis) to prepare multigram quantities of
AZD4573 to support preclinical studies. We also describe the
development of a third generation synthesis used to manufacture
kilogram quantities of AZD4573 for clinical studies.
These contaminants did not interfere in the subsequent Suzuki
coupling, and therefore we chose not to spend time optimizing
this reaction further.
Third Generation Synthesis of 6, Addressing the
Hazardous Wolff−Kishner Reduction. For future larger
scale manufacture to support clinical studies, we believed that
employing a low yielding batchwise Wolff−Kishner reduction of
3
, as used in the first and second generation synthesis, would not
RESULTS AND DISCUSSION
be a first choice method for this transformation. Therefore, in
parallel to the ongoing work to manufacture 4 using the existing
Wolff−Kishner batch-wise process, a continuous flow approach
■
Development of a Large Scale Route to Pyrazole
Fragment (A). Second Generation Synthesis of 6. For the
initial scale-up development of AZD4573 we decided to
continue with use of the Wolff−Kishner conditions for
reduction of 3 and focused our efforts on finding better
conditions for the borylation of 5, which, in the first generation
synthesis, had a moderate yield mainly due to the formation of
up to 50% of byproduct 4 as a result of hydrodehalogenation
8
was investigated. Also a catalytic hydrogenation approach was
attempted as an alternative reduction method (Scheme 3).
Scheme 3. Alternative Strategies To Reduce 3 to Compound
4
(
Method A, Scheme 2). Metalation of 5b (Method B, X = I,
Scheme 2) had been investigated by the medicinal chemistry
team, and while this had been demonstrated to be feasible
(
yields up to 90% achieved), the results were inconsistent and
varied according to scale.
As a first step we sought to identify optimized conditions for
both the formation and subsequent borylation of bromide 5a,
aiming to improve atom efficiency, yield, and waste profile for
the overall conversion of pyrazole 4 to BPin ester 6. On a small
scale, we found that bromination of 4 using bromine in slight
excess furnished compound 5a in good isolated yield (85%).
However, using NBS as reagent gave a much cleaner profile and
a faster rate of reaction. Next, inspired by recent publications
which described the use of triisopropylborate in lithium
exchange−borylation reactions, we planned to apply a similar
The chief concerns in the catalytic hydrogenation approach
were (a) incomplete reduction leading to formation of
secondary alcohol 4-OH and (b) reduction of the pyrazole ring.
We found that the quality of the starting material 3 was crucial
in order to have a positive hydrogenation outcome. Less pure
starting material frequently gave lower yields or no conversion,
and it was assumed that this was due to poisoning of the catalyst
by contaminants. Starting from high quality 3 (95 wt %/wt), use
of the highly active commercially available Johnson-Matthey 5%
Pd/C paste catalysts 5R39 and 5R87L provided immediate hits
for complete conversion of 3 to the partially reduced alcohol 4-
OH (Table 1, entries 1−8). Based on these promising results a
reactivity screen was initiated, using the JM 5% Pd/C 5R39
catalyst in the presence of strong acid additives (MsOH, TFA,
and H SO ) at elevated temperatures. It was then found that, at
6
strategy for our substrate 5a. The reaction is based on an in situ
formation of an isopropylborate intermediate which is trans-
esterified to the desired pinacol boronate using pinacole. Before
we were ready to attempt such a strategy we needed a good
analytical tool to be able to monitor all possible intermediates
and byproducts formed during the reaction course. The low UV-
absorbance, volatility, and reactivity of the intermediates made it
difficult to use traditional methods like HPLC and GC. Instead,
1
we chose to use off-line H NMR analysis for this study and were
2
4
pleased to see that well separated peaks were obtained for the
compounds/intermediates of interest (Figure 2).
temperatures of 80 °C and above, the desired conversion to 4
started to occur, with the best conversions obtained with high
catalyst loading (entry 11). Above 120 °C, a breakdown in the
reaction progression (potentially due to sintering of the catalyst)
and additional byproducts (including pyrazole reduction)
started to manifest (entry 12). Finally, concentration screening
at the optimal temperature of 120 °C revealed that increasing
Optimal results were obtained using cryogenic conditions
−70 °C) where triisopropylborate was present from the start of
(
reaction prior to addition of butyllithium. Using such conditions
resulted in significantly less proto-debromination which
generates 4. Also, an excess of ∼1.8 equiv of the
triisopropylborate was required to efficiently inhibit this side-
reaction. To reach full conversion to intermediate 13 it was also
the concentration of the reaction, from 33.3 rel vol H O to 10 rel
2
vol H O, achieved the target profile of >90% conversion to
2
7
necessary to use an excess of butyllithium (∼1.5 equiv). These
desired compound 4 with <5% byproducts (entries 14 and 15).
Overall, the screening demonstrated that catalytic hydro-
results were in line with those reported by Stewart et al. using an
6
analogous compound. Using these optimal conditions we
genation (3 barg H ) of 3 in water (10 rel vol) using 150 wt % of
2
performed the reaction on a > 100g scale with consistent results
and only trace amounts of byproduct 4 were detected. Next, an
in situ trans-esterification of the isopropylborate ester inter-
mediate 13 with pinacol was performed resulting in the desired
the 5% Pd/C catalyst 5R39 under strongly acidic conditions
(1.25−1.50 equiv of H SO ) at 120 °C furnished the desired
2
4
compound 4 with an acceptable crude purity profile.
We believe that, with further appropriate development and
optimization, this alternative approach of reduction would have
great potential as a viable large-scale route. However, the
strongly acidic conditions used, in combination with the high
temperature, would require pressurized reactors with a strong
tolerance to these harsh conditions. Reluctantly, due to the
specialized nature of such equipment and the ongoing time
1
pinacole ester 6. This reaction was also easily monitored by H
NMR of the crude mixture (Figure 2), and we found that several
hours at 20 °C were required to reach full conversion. A simple
workup consisting of aqueous washes followed by concentration
1
gave pinacole ester 6 in 64% purity (w/w) by H NMR (92%
yield). Major impurities were pinacol and some residual toluene.
D
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
a
Table 1. Screening of Conditions for Hydrogenation of 3 to 4
%
HPLC Area
4-OH
b
Entry
Cat. loading (%wt/wt)
Additive (equiv)
Concn (rel vol H O)
Temp (°C)
3
4
Other
2
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
50
50
50
50
50
50
50
50
50
100
150
50
150
150
150
MsOH (2.50)
MsOH (1.25)
TFA (2.50)
33.3
33.3
33.3
33.3
33.3
33.3
33.3
33.3
33.3
33.3
33.3
33.3
33.3
10
60
60
60
60
60
0.0
0.0
7.8
1.1
0.0
0.0
0.6
0.0
0.9
0.4
0.3
0.0
0.0
0.0
0.0
0.0
0.0
0.8
6.9
29.8
80.3
92.0
17.1
92.4
91.9
92.4
99.2
97.7
91.4
98.3
99.2
99.4
97.9
88.6
64.8
18.1
6.6
0.8
2.3
0.8
0.6
0.8
0.6
0.7
4.5
4.4
1.3
1.1
17.9
0.7
3.7
2.5
TFA (1.25)
H SO (2.50)
2
4
H SO (1.25)
60
80
2
4
H SO (1.25)
2
4
H SO (1.25)
100
120
120
120
140
120
120
120
2
4
H SO (1.25)
2
4
0
1
2
3
4
5
H SO (1.25)
2 4
H SO (1.25)
2
4
H SO (1.25)
65.0
0.0
0.0
0.0
6.9
4.4
4.9
2
4
H SO (1.50)
2
4
H SO (1.25)
2
4
H SO (1.50)
10
0.2
2
4
a
The high throughput screening was performed in a Biotage Endeavor (8 reactor array). Catalyst and substrate 3 (90 mg) were charged, followed
by water and acid additive. The reactors were purged with nitrogen 3 times and H 3 times followed by hydrogenation under 3 barg of hydrogen at
2
1
the temperature stated in Table 1 for 16 h. The mixtures were centrifuged and aliquoted and analyzed by GC/MS, HPLC and H NMR.
b
Combined impurity fractions, including reduction of pyrazole ring.
a
Table 2. High-Throughput Screening for Direct Ir-Catalyzed Borylation of 4
Conversion of 4 to 6 (% UPLC)
b
Entry
[Ir]-cat. (mol %)
ligand
mol %
solvent
temp (°C)
1 h
3 h
23 h
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
1
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
tmphen
5
5
5
5
5
hexane
THF
60
60
60
60
60
60
60
60
60
60
60
60
80
80
80
80
28
34
40
36
15
32
34
29
40
30
41
40
49
77
82
58
68
81
98
91
72
74
72
88
76
72
98
97
88
100
100
99
100
100
99
100
72
100
100
100
100
100
100
98
100
100
100
100
iPrOAc
CPME
EtOAc
MeTHF
hexane
THF
MTBE
MeTHF
CPME
iPrOAc
hexane
THF
5
10
10
10
10
10
10
4
4
4
4
0
1
2
3
4
5
6
1
1
1
CPME
MeTHF
a
[
Ir(cod)OMe] (1−5 mol %), ligands (4−10 mol %), B pin (1.5 equiv), and internal standard were charged to reaction vials in an inerted
2
2
2
glovebox (<5 ppm of O and <1 ppm of H O). Solvent (0.85 mL) and 4 (0.2 mmol as a 1.86 M solution in hexane) were then added, and the vials
were sealed, stirred, and heated for 23 h. tmphen = 3,4,7,8-tetramethylphenanthroline.
2
2
b
constraints to manufacture API for the initial clinical studies, we
did not pursue further this new hydrogenation strategy, nor the
continuous Wolff−Kishner approach mentioned above. Instead
it was decided to revert to the previously employed Wolff−
Kishner batch conditions to support the project with the
building block 4.
exchange/borylation sequence in the second generation syn-
thesis was efficient in terms of yield and ease of workup, for
further scale-up we were concerned about the cryogenic
conditions required. Furthermore, a direct access to pinacole
ester 6 where the bromination step could be excluded was
appealing. Inspired by successful iridium catalyzed borylations
of various arenes in the literature, which are applicable also on a
Investigating an Iridium Catalyzed Borylation Ap-
proach To Obtain 6. Although the lithium−halogen
9
large scale, this led us to attempt a similar strategy (Table 2).
E
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
From our initial condition screening (see Supporting
Information, Table 1), we could clearly see that direct borylation
process, and we wanted to avoid the dependence upon
microwave conditions.
Second Generation Synthesis of 8. To start with, we
screened various thermal conditions for the ammonolysis in
sealed microwave vials by varying the organic cosolvent and also
the loading of ammonia (Table 3).
A cleaner profile and a better conversion was obtained when
performing the reactions using aqueous ammonia and a
cosolvent such as DMSO or NMP (entries 4−8). Ammonia in
alcoholic solvents such as ethanol gave rise to byproducts, and
these reactions were also slow (entries 1 and 2). One major
drawback when using a combination of organic solvent and
aqueous ammonia was that, upon mixing, an exothermic
reaction occurred and severe foaming was also observed which
was expected to be troublesome to control on a large scale. Also,
the heat generated during mixing resulted in loss of ammonia.
To circumvent these problems, attempts were made to perform
the reaction under continuous processing (Figure 3).
of 4 using 5 mol % of [Ir(cod)OMe] as the catalyst looked
2
promising. Other than with highly polar solvents such as DMF
and DMSO, the desired borylated product 6 was produced with
conversions ranging from moderate to excellent at the 1 h
sample point. Selecting tmphen as the preferred ligand a more
extensive screen was performed to identify optimal solvent and
conditions (Table 2). It was found that ethers and other
nonprotic solvents gave full conversion to 6 within 23 h, and that
by increasing the reaction temperature from 60 to 80 °C, a low
catalyst loading of 1 mol % was sufficient to reach full conversion
within this reasonable time frame. Using the conditions of entry
1
4 with no additional optimization, we found that the reaction
gave consistent and good results and these conditions were
successfully applied on multikilogram scale.
Development of a Large Scale Route to Amino-
pyridine Fragment (B). Upon planning for future large scale
campaigns we sought suppliers of both the fluoro compound 7
and the aminopyridine 8 (Scheme 4).
Thus, starting material 7 was dissolved in NMP and mixed in a
simple T-piece with aqueous ammonia followed by reaction at
1
00−140 °C in a coiled PFA tubing reactor. Unfortunately, very
Scheme 4. Ammonolysis of Compound 7
low conversions were obtained and, given the time required to
reach full conversion, this process was deemed inefficient. Under
the strict time constraints given to deliver API for initial
preclinical studies, we chose to use a batch approach to
manufacture 8 which we considered safe at this scale despite the
aforementioned issues.
On 100 g scale in a steel vessel at 100 °C and using aqueous
ammonia (13 equiv) in NMP as solvent, fluoropyridine 7
underwent ammonolysis to give, after extraction and concen-
1
tration, aminopyridine 8 in 93% assayed yield by H NMR as a
highly discolored solid. Although we initially were concerned
about the generation of HF and its corrosive/etching properties,
we did not visually observe any damage to the reaction vessel,
probably due to the large excess of ammonia used. To avoid
the need for chromatography for purification of 8, a quick screen
of suitable solvents for crystallization was performed. Here we
found that the use of toluene gave the pure colorless crystalline
compound 8 in 65% assayed yield. For further scale-up, we
planned to do a more rigorous safety evaluation and find better
conditions for the ammonolysis reaction.
Third Generation Synthesis of 8. Accommodation of the
ammonolysis reaction (Scheme 5) at the kilogram scale required
careful consideration of chemical process hazard data with the
equipment capability in mind. In this case, a jacketed pressure
vessel constructed of C22 hastelloy having a working pressure of
up to 10 barg was used. A controlled addition of the 5-chloro-2-
fluoro-4-iodopyridine in NMP solution to the 28 wt % ammonia
It was found that aminopyridine 8 was not readily accessible in
large quantities from commercial sources. The fluoropyridine on
the other hand was easy to access from various suppliers.
Therefore, we purchased fluoropyridine 7 and devoted our time
in optimizing the ammonolysis of this compound. Intrinsic to
ammonolysis of fluoropyridines are several potential hazards
such as (i) exposure to toxic and volatile ammonia (ii)
pressurized vessels at a high temperature, (iii) incompatibility
of ammonia with organic solvents and (iv) formation of etching
hydrogen fluoride that might damage reactor equipment. In the
first generation synthesis, the ammonolysis was performed at
10
100 °C in a MW vial using a large excess of ammonia in an
organic cosolvent such as NMP. This was followed by
purification of the product 8 by ion-exchange chromatography.
We believed that a similar strategy could be used also for large
scale applications but where most of the aforementioned issues
were addressed. We also aimed for a chromatography-free
a
Table 3. Ammonolysis of Fluoropyridine 7
b
entry
solvent
cosolvent
NH (equiv)
temp (°C)
time
conv
comment
3
1
2
3
4
5
6
7
8
EtOH
EtOH
water
water
water
water
water
water
−
−
−
5
12
30
15
15
15
10
13
100
80
80
80
80
80
70
100
20 h
72 h
20 h
20 h
20 h
20 h
20 h
20 h
46%
60%
70%
100%
97%
50%
byproducts
−
byproducts
clean
clean
clean
clean
clean
NMP (55%)
DMSO (55%)
THF (55%)
NMP (55%)
NMP (40%)
100%
100%
a
b
The reactions were performed in sealed microvials (0.1 g substrate/1 mL solvent) and heated in an oil bath. Conversion of 7 determined by
HPLC.
F
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 3. Attempted continuous ammonolysis of 7.
5
Scheme 5. Ammonolysis of 7 Using a Two-Stage
Temperature Ramping
yield. Therefore, we wanted to investigate other routes to this
compound, and as such, two strategies were evaluated (Scheme
6
):
(
1) Enzyme catalyzed enantioselective hydrolysis of racemic
esters (Strategy 1)
(
2) Enzyme catalyzed enantioselective acetylation of racemic
amino-ester rac-14 (Strategy 2)
Literature reports describe the efficient resolution of BOC
protected amino-esters by employing various lipases to give
11
enantiopure protected amino acids such as C. We began to
investigate such an approach and in parallel to that work we also
spent time on the acetylation/resolution of rac-14 (strategy 2)
which, if successful, would provide the additional advantage of
removing the need for the additional Boc-deprotection/
acetylation sequence used in the first generation synthesis of
AZD4573. There are several examples of enantioselective
acylations of racemic primary amines in the literature, and in
particular the immobilized Candida Antarctica lipases such as
solution at ambient temperature with efficient agitation ensured
good control of the mixing exotherm. However, prior assess-
ment of the “all in” reaction exotherm had indicated that, for the
heat of reaction of 156 kJ/mol 5-chloro-2-fluoro-4-iodopyridine
and operating at a reaction temperature of 90 °C with the vessel
vent closed, an adiabatic temperature rise to ca. 130 °C could
occur, resulting in an associated pressure of 13.8 barg and hence
a vessel overpressure scenario. The reaction conversion, process
temperature, and pressure relationship was explored further,
based on taking a two-stage temperature approach to the
12
Novozym435 have been successfully employed. We were
pleased to find that in our case, using Novozym435 as the lipase
and either EtOAc or iPrOAc as the solvent/acetyl donor, an
efficient resolution of a racemic mixture of 14 was obtained
furnishing, after subsequent crystallization and hydrolysis,
reaction. The mixture of 7 in 28% NH OH (aq.) and NMP
4
13
would be held at ∼50 °C until a conversion of at least 60% was
reached prior to increasing the temperature to ∼85 °C to
complete the reaction (Scheme 5). Modeling of this two-stage
approach indicated that the temperature rise caused by the
reaction exotherm in a loss of vessel cooling event would be
insufficient to cause vessel overpressure. Thus, an inherent safety
approach for the accommodation was achieved. Figure 4a shows
the predicted pressure drop for the reaction hold at 55 °C and
then the actual trend from the plant data historian for the hold.
The former is with respect to reaction conversion, and the actual
data are recorded with time, as it was not possible to monitor
reaction conversion during the hold. Figure 4b shows pressure
data with time from the plant data historian for the second
reaction temperature step to 85 °C and hold.
Development of a Large Scale Route to Amino Acid
Fragment (C). Second/Third Generation Synthesis of 15.
While the homochiral building block C (R = BOC) was
commercially available, we decided that the limited accessibility
made commercial sourcing an undesirable strategy to cover the
supply needs for both preclinical and initial clinical trials.
Although the compound can easily be prepared in enantiopure
form by following published procedures, these are based on
repeated crystallization with homochiral amines in low overall
amidoacid 15 in >99% ee. With some further optimization
and development of this approach we have established a viable
route suitable for multikilogram manufacture of 15 in 32%
overall yield from rac-14. With this new building block in hand, a
subsequent deprotection/acetylation strategy after amide
coupling, as used in the first generation synthesis of
AZD4573, was not needed.
AZD4573 Route Strategy: Assembly of Building Blocks
A−C. To assemble the three building blocks A−C, two options
are available. Either B can be coupled with C to give BC followed
by coupling with A to give AZD4573 or a coupling of A with B to
give AB followed by coupling with C would give the same
compound (Figure 5). Both of these strategies were evaluated
and the factors that were taken into consideration were (i)
physical properties and ease of handling of BC vs AB, (ii)
palladium content in final API AZD4573, and (iii) cost and
accessibility of A vs C.
We found that intermediate BC had very unfavorable
properties compared with AB and resulted in a troublesome
extractive workup due to low solubility. This also led to a low
yield in the subsequent Suzuki coupling. A Suzuki coupling as
the final step also was unfavorable in terms of risk of
contamination of final API with unacceptable levels of
G
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 4. (a) Predicted pressure drop vs conversion for the ammonolysis reaction hold at 55 °C and actual pressure drop vs time for the reaction hold at
5 °C. (b) Actual pressure vs time for the second reaction hold at 85 °C.
5
palladium. This would require substantial amounts of scavenger
to reduce levels of Pd to an acceptable level. These disadvantages
should of course be balanced against the relative cost of building
block A vs C. We considered A to be of higher value than C
because of the recent development of the cost-effective enzyme
catalyzed acetylation/resolution strategy which could be used
for multikilogram manufacture of C. However, because of the
aforementioned issues with the BC intermediate and the
nonfavorable strategy to implement a palladium catalyzed
reaction as the last step in the synthesis of AZD4573, we
chose to use an approach where we first coupled A with B via a
Suzuki reaction followed by an amide coupling with C to give
AZD4573. In the next sections we describe our efforts in
developing good large scale methods for these two trans-
formations.
Optimization of Suzuki Reaction. Although a good yield
in the Suzuki coupling was obtained in the first-generation
synthesis, for cost reasons, the excess of A required (110−150
mol %) to reach full conversion of B was of major concern. Also,
we wanted to replace the chromatography employed in the first
H
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 6. Two Potential Strategies To Obtain Homochiral Building Block C
Figure 5. Strategies to assemble the three building blocks A, B, and C.
1
Figure 6. H NMR recorded of the crude Suzuki mixture in CDCl .
3
generation synthesis with crystallization. Furthermore, we
wanted to avoid the use of the solvent 1,4-dioxane, which easily
forms peroxides, and also identify a less expensive base relative
to K HPO used in the first generation synthesis. To find
the kinetics as well as potential side reactions is of great value.
Therefore, we searched for an analytical technique where this
could be achieved. Due to the volatile nature and low UV-
absorbance of some of the intermediates, we chose to use off-line
2
4
1
optimal conditions, a good understanding of the reaction and
H NMR monitoring as an analytical tool for this reaction
I
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 7. Optimized Conditions for Suzuki Coupling
Scheme 8. Optimized Conditions for the Amide Coupling
Scheme 9. Conditions for Multikilogram Manufacture of AZD4573
1
optimization. Fortunately, a H NMR spectrum recorded in
significantly slower than the rate of the desired reaction.
Therefore, we suspected that a large excess of the boronate 6
probably was not required. We performed a few control
experiments to verify this hypothesis and found that as
anticipated, only a slight excess of boronate 6 was required to
reach full conversion of iodide 8. Using optimal conditions, we
chose to use 1.1 equiv of the boronate 6 which resulted in a good
assayed yield of product 16 (Scheme 7). After a quick screen of
solvents and bases for the reaction, we found that aqueous
CH CN in combination with K CO gave equal or better results
CDCl of the crude mixture showed well resolved peaks for the
two reactants 6 and 8 as well as for the expected byproduct 4
resulting from a protodeborylation of the pinacole ester 6
3
(Figure 6).
Surprisingly, we did not observe any byproduct resulting from
14
reductive dehalogenation of the iodo-starting material 8. On
the other hand, we noted that, after prolonged heating, the
starting boronate 6 underwent proto-deborylation resulting in
byproduct 4. Fortunately, this side-reaction seemed to be
3
2
3
J
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
compared with dioxane/K HPO used in the first generation
synthesis and was chosen as the preferred method for large scale
silica gel. Subsequently, after a solvent switch from the THF
solution of 6 into MeOH and water, the desired product 6
crystallized and was isolated by filtration. In order to obtain an
acceptable purity, we found it necessary to redissolve the
product in warm heptane followed by filtration which removed
additional insoluble impurities. After cooling the filtrate, the
suspension formed was then filtered to afford 5.6 kg of boronate
6 (70% yield) as an off-white solid of high purity (99% purity by
HPLC; 99 wt %/wt assay by NMR). The workup and isolation
procedures employed here to obtain 6 were quite time-
consuming and required a large amount of organic solvent (in
total about 30 relative volumes of MeOH/heptane were used),
so improving this workup will be a focus for process design
ahead of future manufacturing campaigns. However, in
comparison with the previously employed multistep bromina-
tion/lithium−halogen exchange/boronate quench sequence
described above, requiring both nonenvironmentally benign
solvents such as CH Cl and cryogenic conditions, an overall
2
4
manufacture of 16. The catalyst Pd(Cl) dppf was selected due to
2
accessibility and low cost.
Next, a straightforward and robust workup procedure was
established. We chose to perform a direct extractive workup with
EtOAc and aqueous washes. In the manufacture of the
preclinical material, to minimize risk of contamination of API
with palladium to an unacceptable level, we chose to add a Si-
thiol scavenger at this stage to remove the major part of the
palladium content. We believed that two subsequent crystal-
lizations further downstream in the sequence would be enough
to reach acceptable levels of palladium. Crude product 16 after
treatment with scavenger followed by concentration was
obtained as a highly discolored solid in quantitative assayed
yield. After a screen of solvents for crystallization/slurrying of
the crude product, we eventually found that the discoloration
was efficiently removed after slurrying in methylisobutylketone
2
2
(
MIBK). On a 50−100 g scale, the desired product 16 was
improvement in process mass intensity (PMI), environmental
sustainability, and ease of operations has been accomplished by
developing the Ir-catalyzed borylation conditions for this
transformation.
obtained as a colorless solid in 83% isolated yield with 100%
assay (w/w) by H NMR. By further optimization of the workup
1
procedure above, we found that a direct crystallization of 16 was
possible after basification of an acidic aqueous solution of 16 and
no MIBK slurrying operation was necessary. Also, for the
following large scale campaign, we decided to exclude the use of
the Si-thiol scavenger from this reaction step and implement that
operation after the final reaction step instead.
Optimization of Amide Coupling. For reasons of limited
availability and nontrivial handling on a large scale, we searched
for alternatives to Ghosez coupling reagent used in the first
generation synthesis. We quickly found that propylphosphonic
anhydride (T3P) efficiently mediated the coupling of carboxylic
acid 15 with amine 16 and this reagent was also easy to handle,
being sourced as a 50% solution in EtOAc (Scheme 8).
As described further above in more detail, ammonolysis of the
2-fluoropyridine 7 to provide 8 could be performed safely and
with good yield on multikilogram scale. After an extractive
workup using MTBE as the organic solvent, a solvent swap to
toluene enabled crystallization and the desired compound 8 was
isolated by filtration as a colorless solid (4.13 kg; 85% yield; 100
wt %/wt assay by NMR).
The Suzuki coupling of 6 and 8 using a 2 mol % loading of the
Pd(Cl) dppf catalyst proceeded cleanly, and an extractive
2
workup using EtOAc as the organic solvent left the desired
compound in the organic layer. To facilitate the subsequent
crystallization and to remove most of the impurities originating
from the catalyst, the product was extracted into an acidic
aqueous layer followed by basification, before addition of
MeOH to affect crystallization. As stated above, the previously
employed Si-thiol scavenger was removed from this step, to be
introduced subsequently at the API stage. The desired product
On a 50 g scale using an almost equimolar ratio of carboxylic
acid 15 and amine 16 and using pyridine as the base, a 99% crude
assayed yield of AZD4573 was obtained after an extractive
workup. We found that slurrying the crude product in CH CN
3
finally gave the desired polymorph of AZD4573 in 90% yield
1
and with better than 98% assay (w/w) by H NMR. Chiral
1
6 was isolated as a colorless solid on a multikilogram scale by
HPLC analysis showed better than 99% ee and the
concentration of palladium was at an acceptable level of <10
ppm. For the following multikilogram campaign, we found that a
solvent swap from EtOAc to MIBK resulted in crystallization of
the desired polymorph of AZD4573 and a separate crystal-
filtration (89% yield; 100 wt %/wt assay by NMR). ICP-OES
analysis was performed and showed residual B, Ir, and Pd of
5
.40, 67.4, and 638 ppm, respectively.
The enantiomerically pure amido-ester 17 was obtained on
multikilogram scale after a resolution of the racemic cis-
configured 14 using a 5 weight% loading of Novozym 435 as
catalyst followed by crystallization (38% yield; >99% purity,
HPLC; >99% ee, HPLC). Compound 15 was then obtained by
hydrolysis of 17 using sodium hydroxide and subsequent
isolation by crystallization in 84% overall yield with excellent
purity (100% purity, HPLC; 100% ee, HPLC; 99 wt %/wt assay
by NMR).
lization step in CH CN was not necessary.
3
Multikilogram Manufacture of AZD4573. For the
multikilogram manufacture of AZD4573 to support the clinical
studies, the optimized conditions described in Scheme 9 were
used.
The iridium-catalyzed borylation of 4 proceeded non-
eventfully using 1 mol % of the catalyst, 4 mol % of the
phenanthroline ligand, and 1.5 equiv of the B (pin) boron
15
2
2
Finally, the coupling of building blocks 15 and 16 using the
optimized conditions described in Scheme 9 gave crude
AZD4573 after a simple extractive workup. To remove residual
palladium and iridium from the previous step, we chose to apply
an immobilized metal scavenger (3-mercaptopropyl ethylsulfide
silica). This was followed by final crystallization from MIBK to
give 3.96 kg of AZD4573 (86% yield; 98 wt %/wt assay by
NMR; 99.9% chemical purity by HPLC; >99.5% ee by HPLC).
Residual B, Ir, and Pd were at acceptable levels below 5 ppm
(ICP-OES).
source. These were the same conditions as identified from the
original high-throughput screening work, which had been found
to work well with no need for further optimization. In order to
achieve some level of consistency with the original glovebox
screening conditions (<5 ppm of O ), the oxygen level in the
2
headspace was maintained below 5000 ppm by implementing
several nitrogen-vacuum cycles. To remove most discoloration
and insoluble impurities that were found to inhibit the
crystallization of 6, the workup consisted of treatment with
activated carbon followed by filtration through a short bed of
K
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
CONCLUSION
pyrazole 4 (79.5 g, 90 wt %/wt, 525 mmol) and CH Cl (800
2 2
■
mL). With a jacket temperature of +10 °C, to the homogeneous
yellow solution was added portionwise 1-bromopyrrolidine-2,5-
dione (95.4 g, 533 mmol) during 15 min. The reaction
temperature was kept in the interval +20 to +23 °C during this
addition. A solution of 8% Na SO (aq., 250 mL) was added, and
Using the first generation mg-scale synthesis of AZD4573 as a
starting point, we have applied a focused combination of route
design and process development to rapidly generate a much
more scalable and efficient route, enabling multigram
manufacture of API for use in preclinical studies. This second
generation synthetic route was further modified, and the steps
were optimized to give a third generation synthesis which was
successfully used in the multikilogram scale manufacture of API
to support initial clinical studies. With the primary focus being to
develop a manufacturing route which is safe and easy to operate,
with better environmental sustainability, we have made
significant improvements along the way such as (i) reduced
overall step count (6 steps vs 9 steps for third generation vs first
generation); (ii) replacement of all chromatographic purifica-
tions with crystallizations; (iii) elimination of all chlorinated
solvents; (iv) clear demonstration that the hazardous Wolff−
Kishner reduction could potentially be replaced with a
palladium-catalyzed hydrogenation; and (v) extensive use of
catalytic transformations, including development of an
enzymatic resolution/acetylation of a racemic amino-ester to
afford access to a homochiral amido acid and replacement of a
cryogenic two-step borylation sequence with a direct iridium-
catalyzed borylation.
2
3
the biphasic mixture was stirred for 45 min. The organic layer
was washed with Na CO solution (aq. sat., 2 × 250 mL) and
2
3
last water (100 mL) followed by careful concentration at 30 °C
and 400 mbar. Residual water was azeotropically removed using
THF (3 × 100 mL). This afforded the title compound 5a as a
pale brown oil (138 g, 81 wt %/wt, 520 mmol, 99% yield) with
THF as the major impurity. Analytical data were in accordance
3
with those given in literature.
5,5-Dimethyl-3-(4,4,5,5-tetramethyl-1,3,2-dioxaboro-
lan-2-yl)-5,6-dihydro-4H-pyrrolo[1,2-b]pyrazole (6) via
Lithiation/Triisopropylborate Quench. Under an atmos-
phere of nitrogen and in a 3 L three-necked flask equipped with a
thermometer, to a −70 °C solution of 3-bromo-5,5-dimethyl-
5
5
,6-dihydro-4H-pyrrolo[1,2-b]pyrazole 5a (137 g, 81 wt %/wt,
16 mmol) and triisopropyl borate (175 g, 929 mmol) in THF
(
0.7 L) and toluene (0.7 L) was slowly (80 min) added a
solution of butyllithium (2.5 M in hexanes, 0.309 L, 772 mmol)
while maintaining the reaction temperature in the interval −65
1
°
C to −70 °C. H NMR analysis of the crude mixture showed
full conversion to the intermediate i-Pr-borate 13. At −70 °C, a
solution of 2,3-dimethylbutane-2,3-diol (91 g, 774 mmol)
dissolved in toluene (0.5 L) was added during 10 min. The
mixture was slowly allowed to attain 20 °C in the ice bath and
was then stirred for an additional 16 h. A pale brown
EXPERIMENTAL SECTION
■
All reactions were performed under an atmosphere of nitrogen.
Reagents were commercially available and used without further
1
13
purification. H and C NMR were recorded on a Bruker
Avance III spectrometer at 25 °C at the frequency stated in each
experiment. The chemical shifts (δ) are reported in parts per
million (ppm), with the residual solvent signal used as a
reference. Coupling constants (J) are reported as Hz. NMR
abbreviations are used as follows: br = broad, s = singlet, d =
doublet, t = triplet, q = quartet, m = multiplet. The strength/
assay of intermediates and final products are given as weight/
1
homogeneous solution was obtained. H NMR analysis of the
crude mixture showed full conversion of 13 to 6. The reaction
mixture was transferred to a 5 L reactor containing a 10 °C
solution of NH Cl (aq., sat, 2.5 L). The biphasic mixture was
4
stirred for 15 min at 20 °C. The organic layer was washed with
water (2 × 500 mL) followed by concentration under reduced
pressure at 35 °C to a pale yellow wax-solid 6 (193.6 g, 64 wt
%/wt, 473 mmol, 92% yield) with toluene and pinacole as major
impurities. Analytical data were in accordance with those given
1
weight (w/w) and were determined by H NMR measurements
using benzyl benzoate, maleic acid, 1,3,5-trimethoxybenzene, or
,2,4,5-tetrachloro-3-nitrobenzene (TCNB) as internal stand-
3
1
in the literature.
ards. In the iridium catalyzed borylation screen, the following
instrument and conditions were used for determination of
conversion of 4 to 6: UPLC/MS using a Phenomenex Kinetex
5,5-Dimethyl-3-(4,4,5,5-tetramethyl-1,3,2-dioxaboro-
lan-2-yl)-5,6-dihydro-4H-pyrrolo[1,2-b]pyrazole (6) via
Iridium Catalyzed Borylation. 5,5-Dimethyl-4,6-dihydro-
pyrrolo[1,2-b]pyrazole (4, 4.1 kg, 30.1 mol) and dry THF
2
.6 μm C18 100A 75 mm × 3 mm column + guard column, 220
nm, 40 °C. A = 0.1% (w/v) NH Ac/H O, B = 0.1% (w/v)
(36.5 kg, ≤ 0.05% H O wt/wt) were charged into a 80 L glass
4
2
2
NH Ac/MeCN, gradient 5−95% B over 4 min, 95% B 0.5 min,
reactor. Bis (pinacolato) diboron (11.5 kg, 45.3 mol) and
3,4,7,8-tetramethyl-1,10-phenanthroline (0.29 kg, 1.2 mol) were
added, and the solution was then stirred for 1 h at 15−25 °C.
The mixture was evacuated to P ≤ −0.08 MPa and then filled
with nitrogen to normal pressure. This operation was repeated 5
times until the oxygen content was ≤5000 ppm. Bis [cycloocta-
diene(methoxy)iridium] (0.20 kg, 0.30 mol) was added, and the
mixture was evacuated to P ≤ −0.08 MPa and then filled with
nitrogen to normal pressure. This operation was repeated 5
times until the oxygen content was ≤5000 ppm. The mixture
was heated to 70−80 °C and then stirred for 12 h. The reaction
mixture was cooled to 20 °C. Active charcoal (0.41 kg) was
added in two portions, and the mixture was then heated to 45 °C
and stirred for 3 h. The mixture was filtered using a 10 L vacuum
filter flask which was preloaded with Silica gel (2 cm bed, 0.5 kg,
200−300 mesh). The filter cake was rinsed twice with THF (7.3
kg +7.3 kg). The filtrates were combined and concentrated at
≤45 °C under reduced pressure (P ≤ −0.08 MPa). While the
4
1
.2 mL/min, 3 μL injection volume. The UHPLC/MS analyses
were performed using a Waters Acquity UPLC system
combined with a Waters SQD Mass Spectrometer. The
UHPLC was equipped with both a BEH 1.7 μm C18 column
2
.1 mm × 50 mm using a 46 mM ammonium carbonate/NH3
buffer at pH 10 as eluent or an HSS 1.8 μm C18 column 2.1 mm
50 mm using an 11 mM ammonium formate buffer at pH 3 as
×
eluent. The mass spectrometer was operated with electrospray
ionization (ESI) in both positive and negative mode. The
enantiomeric excess of the corresponding isopropylester of 15
was determined by analytical SFC [Lux C2 column (4.6 mm ×
1
50 mm), 15% 2-propanol in CO , 120 bar, as eluent]. HRMS
2
measurements were performed on a QTOF 6530 instrument
Agilent), mass precision ±5 ppm.
-Bromo-5,5-dimethyl-5,6-dihydro-4H-pyrrolo[1,2-b]-
(
3
pyrazole (5a). In a 1 L reactor under an atmosphere of nitrogen
were charged 5,5-dimethyl-5,6-dihydro-4H-pyrrolo[1,2-b]-
L
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
temperature was maintained at ≤45 °C, methanol (6.3 kg) was
added, and the mixture was stirred for 3 h followed by
concentration at ≤45 °C under reduced pressure (P ≤ −0.08
MPa). Methanol (6.5 kg) was again added, and the procedure
above was repeated twice. Then the mixture was cooled to 20
was washed twice with a solution of NaCl (1.02 kg) in water (9.1
kg) and then sequentially extracted with an aqueous solution of
HCl (2M, 15 L, 10.5 L, 10.5 L). The combined three aqueous
layers were transferred back to the reactor via an in-line filter (5
μm), and MeOH (10 L) was added followed by careful pH
adjustment to 11 using a 45% aqueous solution of KOH (16.36
kg, 131.3 mol) while maintaining the temperature below 30 °C.
The mixture was then stirred at 20 °C for 1 h followed by
filtration and washing of the filter cake with water (6 L). The
cake was then dried at 50 °C under vacuum to give the title
compound 16 as a colorless solid (2.76 kg, 100 wt %/wt, 10.5
mol, 88% yield). ICP-OES analysis showed residual B, Ir, and Pd
°
C. Water (40.9 kg) was added, and the suspension formed was
stirred for 3 h followed by filtration. Methanol (3.7 kg) and the
filter cake were added into the reactor at 20 °C and then stirred
for 1 h. Water (41.1 kg) was added, and the mixture was stirred
for 3 h followed by filtration. The filter cake was charged to the
reactor, and n-heptane (70 kg) was added followed by heating to
5
0 °C and stirring for 3 h. The mixture was filtered, and the
filtrate was concentrated at 55 °C under reduced pressure P ≤
0.08 MPa until 14 L of the content remained. The mixture was
1
of 5.40, 67.4, and 638 ppm, respectively. H NMR (500 MHz,
−
CDCl ) δ 1.32 (6H, s); 2.85 (2H, s); 3.94 (2H, s); 4.41 (2H, s,
3
1
3
cooled to 0 °C and was stirred for 4 h followed by filtration and
washing of the filter cake with ice-cold n-heptane (1.4 kg × 2)
and last water (4.1 kg + 4.0 kg). The cake was dried at 30 °C
br); 6.44 (s, 1H); 7.85 (s, 1H), 8.06 (s, 1H). C NMR (126
MHz, CDCl ) δ 28.1, 40.5, 43.5, 61.1, 107.8, 112.0, 119.3, 141.3,
3
+
142.8, 144.1, 148.1, 157.0. HRMS [M + H] m/z calcd for
under vacuum to give the title compound 6 as an off-white solid
C H ClN 263.1058, found 263.1066.
1
3
16
4
1
(
5.6 kg, 99 wt %/wt, 21.1 mol, 70% yield). H NMR (500 MHz,
(1S,3R)-3-Acetamidocyclohexanecarboxylic Acid (15).
CDCl ) δ 1.28 (s, 6H); 1.29 (s, 12H); 2.79 (s, 2H); 3.86 (s,
Resolution of Enantiomers rac-14. A reactor was charged with
compound rac-14 (16.8 kg, 90.7 mol) and i-PrOAc (132 L). The
reaction temperature was kept at 18 °C, and Candida Antarctica
lipase B (839 g, 5 wt %) was added. The mixture was stirred for
at least 13 h until a conversion of 48% of rac-14 had been
obtained. The mixture was centrifuged, and the cake was washed
with i-PrOAc (16.8 L × 2). The filtrate was collected and
transferred back to the reactor. A solution of oxalic acid
dihydrate (2.85 kg, 22.6 mol) in water (50.3 L) was added to the
reactor at 20 °C. The mixture was kept at that temperature and
stirred for 1 h followed by separation of phases. The aqueous
phase was extracted with i-PrOAc (50.3 L), and the combined
organic layers were washed with water (16.8 L) and brine (83.9
L, 35 wt %) followed by concentration under vacuum P ≤ −0.08
MPa at 40 °C until the concentration of product is 3.0 relative
volumes. Cyclopentylmethyl ether (CPME, 42.0 L) was added,
and the mixture was concentrated under vacuum P ≤ −0.08
MPa at 40 °C until 3 relative volumes of solvent remained. The
procedure was repeated three times until the content of i-PrOAc
was 3%. Cyclohexane (154.4 L) was added (CPME content =
25%). The mixture was heated to 60 °C and kept at that
temperature for 0.5 h. The clear homogeneous solution was
cooled to 50 °C in 2 h and kept at that temperature for 1 h. The
mixture was cooled to 40 °C in 2 h and kept at that temperature
for 1 h. The mixture was cooled to 30 °C in 2 h and kept at that
temperature for 1 h. The mixture was then finally cooled to 20
°C in 2 h and stirred for an additional 12 h. The suspension was
centrifuged, and the cake was washed with cyclohexane/CPME
= 5/1 (16.8 L × 2). The cake was dried in an oven under P ≤
−0.08 MPa at 40 °C for 6 h to give the desired amido-ester 17
(7.88 kg, 38.3% yield) with 99.7% chemical purity and ee 99.9%.
Analytical data were in accordance with those reported
3
13
2
2
1
H); 7.76 (s, 1H). C NMR (126 MHz, CDCl ) δ 25.0 (4 Cs),
3
8.5 (2 Cs), 39.3, 43.3, 61.0, 83.0 (2 Cs), 100.3 (br), 149.5,
+
53.0. HRMS [M + H] m/z calcd for C H BN O 263.1925,
14
24
2
2
found 263.1935.
5
-Chloro-4-iodo-pyridin-2-amine (8). An aqueous sol-
ution of ammonium hydroxide (28 wt %/wt, 12.6 kg, 101 mol)
was added to a well agitated nitrogen purged C22 hastelloy
pressure reactor followed by line rinsing with water (2 kg). At a
temperature of 13 °C, 5-chloro-2-fluoro-4-iodopyridine (4.915
kg, 19.1 mol) dissolved in NMP (8 kg) was slowly added during
1
0 min (slightly exothermic) followed by line rinsing with NMP
(
3.2 kg). The reaction temperature was maintained below 20 °C
during this addition. The reactor was sealed, and the internal
temperature was increased to 55 °C which resulted in an internal
pressure of 1.78 barg. The mixture was stirred for 18 h which
resulted in an 83% conversion of the starting fluoropyridine
(
internal pressure 1.41 barg). The reaction mixture was heated
to 85 °C (internal pressure 3.9 barg) and was stirred for an
additional 18 h followed by cooling to ambient temperature.
MTBE (20.3 kg) was added followed by the addition of water
(
×
(
10.8 kg). The aqueous layer was extracted with MTBE (13.7 L
2), and the combined organic layer was washed with water
9.5 L × 3). Toluene (10 L) was added, and the jacket
temperature was increased to 110−130 °C to distill off most of
the MTBE. Toluene (10 L) was again added, and distillation
continued until a total of 70 L of the volatiles had been distilled
off. The reaction mixture was then slowly cooled to 2 °C during
4
h and was stirred at that temperature for an additional 18 h
followed by filtration of the suspension and washing of the filter
cake with toluene (2 L × 2). The cake was then dried under
vacuum at 40 °C to give the title compound 8 as a colorless solid
13
(
4.13 kg, 100 wt %/wt, 16.2 mol, 85% yield). Analytical data
were in accordance with those of commercial authentic material.
-Chloro-4-(5,5-dimethyl-4,6-dihydropyrrolo[1,2-b]-
pyrazol-3-yl)pyridin-2-amine (16). Compound 8 (3.03 kg,
1.9 mol) was charged to a reactor followed by water (25.3 L),
previously.
Hydrolysis To Give 15. At 20 °C, to a reactor containing water
(39.7 L), NaOH (2.05 kg, 51.3 mol), and MeOH (28.0 L) was
added in portions the above amidoester (7.78 kg, 34.2 mol). The
mixture was stirred at 25 °C for 3 h followed by the addition of
water (54.5 L). The mixture was concentrated under vacuum P
≤ −0.08 MPa at ≤40 °C until the solution was not more than 9.0
relative volumes and the MeOH content = 7%. Sodium chloride
(14.8 kg, 253.3 mol) was added and the mixture was stirred for 1
h. The reactor content was filtered through a fluid filter
membrane followed by washing with water (0.5 relative
volumes). The mixture was cooled to 15 °C and a concentrated
5
1
compound 6 (3.58 kg, 13.4 mol), and last CH CN (25.3 L). The
3
vessel was degassed by using vacuum/nitrogen purge. Potassium
carbonate (4.11 kg, 29.7 mol) was added followed by [1,1′-
bis(diphenylphosphino)ferrocene]dichloropalladium(II) (174
g, 0.24 mol). The two-phase mixture was stirred vigorously at
5
4 °C for 18 h followed by cooling to 20 °C. EtOAc (22.8 L) was
added followed by separation of the phases. The organic phase
M
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
aqueous solution of HCl was added dropwise until pH ≤ 1. The
suspension formed was stirred for an additional 0.5 h followed
by centrifugation and washing of the cake with water (7.8 L × 2).
After drying of the solid under vacuum P ≤ −0.08 MPa at 40 °C
for 6 h, the desired compound 15 (5.39 kg, 29.1 mol) was
obtained in 100% chemical purity by HPLC (assay 99 wt %/wt
by 1H NMR), ee > 99.5%. Other analytical data were in
AUTHOR INFORMATION
Corresponding Author
■
Staffan Karlsson − Early Chemical Development,
Pharmaceutical Sciences, Biopharmaceuticals R&D,
AstraZeneca, Gothenburg, Sweden; orcid.org/0000-0002-
302-7157; Email: staffan.po.karlsson@astrazeneca.com
5
13
accordance with those reported previously.
1S,3R)-3-Acetamido-N-[5-chloro-4-(5,5-dimethyl-4,6-
Authors
(
Helen Benson − Early Chemical Development, Pharmaceutical
Sciences, Biopharmaceuticals R&D, AstraZeneca, Macclesfield,
United Kingdom
dihydropyrrolo[1,2-b]pyrazol-3-yl)-2-pyridyl]cyclo-
hexanecarboxamide (AZD4573). Compound 16 (2.758 kg,
1
0.50 mol) and EtOAc (27 L) were charged to the reactor
Calum Cook − Early Chemical Development, Pharmaceutical
Sciences, Biopharmaceuticals R&D, AstraZeneca, Macclesfield,
United Kingdom
Gordon Currie − Early Chemical Development,
Pharmaceutical Sciences, Biopharmaceuticals R&D,
AstraZeneca, Macclesfield, United Kingdom
Jerome Dubiez − Early Chemical Development,
Pharmaceutical Sciences, Biopharmaceuticals R&D,
AstraZeneca, Macclesfield, United Kingdom
Hans Emtenäs − Early Chemical Development, Pharmaceutical
Sciences, Biopharmaceuticals R&D, AstraZeneca, Gothenburg,
Sweden
followed by the addition of 15 (2.041 kg, 10.80 mol) and
pyridine (3.39 L, 41.9 mol). 1-Propanephosphonic anhydride
(
T P, 50% in EtOAc, 10.01 kg, 17.0 mol) was added during 20
3
min maintaining the reaction temperature below +23 °C. The
mixture was stirred at 20 °C for 18 h followed by the addition of
water (19.5 L) during 10 min. The mixture was stirred for 25
min at 20 °C followed by separation of the phases. The aqueous
phase was extracted with EtOAc (13.7 L), and the combined
organic layers were washed with a solution of K CO (1.407 kg,
2
3
1
0.18 mol) in water (11.17 kg). The organic layer was washed
with water (13 L × 2) followed by the addition of 3-
mercaptopropyl ethyl sulfide silica (1.32 kg, 60−200 mesh).
The mixture was stirred at 20 °C for 18 h followed by filtration of
the suspension and rinsing with EtOAc (7 L × 2). The organic
layer was charged to the reactor via an in-line filter (1 μm), and
the volatiles were distilled off at 70 °C until a total volume of
Janet Hawkins − Medicinal Chemistry, Oncology R&D,
AstraZeneca, Cambridge, United Kingdom
Rebecca Meadows − Chemical Development, Pharmaceutical
Technology & Development, Operations, AstraZeneca,
Macclesfield, United Kingdom
Peter D. Smith − Early Chemical Development, Pharmaceutical
Sciences, Biopharmaceuticals R&D, AstraZeneca, Macclesfield,
United Kingdom; orcid.org/0000-0001-5312-9893
Jeffrey Varnes − Medicinal Chemistry, Oncology R&D,
AstraZeneca, Boston, Massachusetts 02451, United States
2
7.6 L was obtained. Methylisobytyl ketone (MIBK, 27.5 L) was
added followed by removal of 20 L of the volatiles through
distillation. More MIBK (27.5 L) was added followed by
distilling off 16 L of the volatiles. More MIBK (13.75 L) was
added followed by removal of 13.5 L of the volatiles through
distillation. This procedure was repeated one more time. The
temperature of the mixture was cooled to 105 °C, and seeds of
correct polymorph (3.6 g) of AZD4573 were added followed by
stirring for 1 h. The suspension was then cooled to 20 °C at 0.2
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.oprd.1c00058
Notes
The authors declare no competing financial interest.
°C/min and stirred for 1 h. A temperature cycling program was
applied: heating to 40 °C followed by stirring for 1 h and then
cool mixture to 20 °C again and stir for 1 h. This procedure was
repeated 3 times. The suspension was then stirred at 20 °C for
ACKNOWLEDGMENTS
■
We thank Steven McKown (Early Chemical Development,
Pharmaceutical Sciences, AstraZeneca), Trevor Churchill
Global Process Safety, Pharmaceutical Technology & Develop-
ment, AstraZeneca), Danny Mortimer (Johnson-Matthey
Catalysis and Chiral Technologies, Cambridge, U.K.), Chris
Atherton (Onyx Scientific Ltd, Sunderland, U.K.), Jianbo Yang
et al. (Asymchem Laboratories, Tianjin, China), and Liao
Xiaoming et al. (Pharmaron, Beijing, China) for their
contributions to this work.
1
8 h followed by cooling to 10 °C and stirring for an additional 1
(
h. The mixture was filtered, and the cake washed with MIBK (4.5
L × 3) followed by drying under vacuum at 70 °C to give the title
compound AZD4573 as colorless crystals (3.96 kg, 98 wt %/wt
by NMR, 99.9% purity by HPLC, 9.0 mol, 86% yield). ICP-OES
analysis showed levels of B, Ir, and Pd below 5 ppm. HRMS [M
+
+
H] m/z calcd for C H ClN O 430.2004, found 430.2009,
22
29
5
2
ee > 99.5% by HPLC. NMR data were in agreement with those
3
recently reported.
ABBREVIATIONS
■
TLC, thin layer chromatography
DMA, Dimethylacetamide
NBS, N-Bromosuccinimide
ASSOCIATED CONTENT
■
*
sı Supporting Information
The Supporting Information is available free of charge at
NIS, N-Iodosuccinimide
https://pubs.acs.org/doi/10.1021/acs.oprd.1c00058.
DIPEA, Diisopropylethylamine
Ghosez reagent, 1-Chloro-N,N-2-trimethyl-1-propenylamine
MsOH, Methanesulfonic acid
TFA, Trifluoroacetic acid
PFA, Perfluoroalkoxy
API, Active Pharmaceutical Ingredient
DSC, Differential Scanning Calometry
1H NMR, 13C NMR, HPLC chromatograms for
intermediates and AZD4573, screening results for Ir-
catalyzed borylation of 4. Analytical testing report for
compound 15. Experimental procedures for preparation
of compounds 2, 3, 4 (PDF)
N
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
PMI, Process Mass Intensity
ICP-OES, Inductively Coupled Plasma Optical Emission
Spectroscopy
Iridium-Catalyzed C−H Borylation of Heteroarenes: Scope, Regiose-
lectivity, Application to Late-Stage Functionalization, and Mechanism.
J. Am. Chem. Soc. 2014, 136, 4287−4299.
(
10) A recent publication describes the use of Ca(EtCO ) to
2 2
efficiently remove HF formed: Blacker, A. J.; Malagon, G. M.; Powell,
REFERENCES
L.; Reynolds, W.; Stones, R.; Chapman, M. R. Development of an S Ar
■
N
Reaction: A Practical and Scalable Strategy to Sequester and Remove
(
1) For example see: (a) Sonawane, Y. A.; Taylor, M. A.; Napoleon, J.
HF. Org. Process Res. Dev. 2018, 22, 1086−1091.
V.; Rana, S.; Contreras, J. I.; Natarajan, A. Cyclin Dependent Kinase 9
(11) Badland, M.; Bains, C. A.; Howard, R.; Laity, D.; Newman, S. D.
Inhibitors for Cancer Therapy. J. Med. Chem. 2016, 59, 8667−8684.
An improved synthesis of (1R,3S)-3-[(tert-butoxycarbonyl)amino]-
(b) Wang, S.; Griffiths, G.; Midgley, C. A.; Barnett, A. L.; Cooper, M.;
cyclohexanecarboxylic acid. Tetrahedron: Asymmetry 2010, 21, 864−
Grabarek, J.; Ingram, L.; Jackson, W.; Kontopidis, G.; McClue, S. J.;
McInnes, C.; McLachlan, J.; Meades, C.; Mezna, M.; Stuart, I.; Thomas,
M. P.; Zheleva, D. I.; Lane, D. P.; Jackson, R. C.; Glover, D. M.; Blake,
D. G.; Fischer, P. M. Discovery and characterization of 2-anilino-4-
8
(
66.
12) (a) González-Sabín, J.; Gotor, V.; Rebolledo, F. Enantioselective
acylation of rac-2-phenylcycloalkanamines catalyzed by lipases.
Tetrahedron: Asymmetry 2005, 16, 3070−3076. (b) González-Sabín,
J.; Gotor, V.; Rebolledo, F. Kinetic resolution of (±)-trans- and (±)-cis-
(
thiazol-5-yl)pyrimidine transcriptional CDK inhibitors as anticancer
agents. Chem. Biol. 2010, 17, 1111−1121.
2) (a) Booher, R. N.; Hatch, H.; Dolinski, B. M.; Nguyen, T.;
2
-phenylcyclopentanamine by CALB-catalyzed aminolysis of esters:the
key role of the leaving group. Tetrahedron: Asymmetry 2004, 15, 481−
88. (c) Gotor, V. Non-conventional hydrolase chemistry: amide and
carbamate bond formation catalyzed by lipases. Bioorg. Med. Chem.
(
Harmonay, L.; Al-Assaad, A.-S.; Ayers, M.; Nebozhyn, M.; Loboda, A.;
Hirsch, H. A.; Zhang, T.; Shi, B.; Merkel, C. E.; Angagaw, M. H.; Wang,
Y.; Long, B. J.; Lennon, X. Q.; Miselis, N.; Pucci, V.; Monahan, J. W.;
Lee, J.; Kondic, A. G.; Im, E. K.; Mauro, D.; Blanchard, R.; Gilliland, G.;
Fawell, S. E.; Zawel, L.; Schuller, A. G.; Strack, P. MCL1 and BCL-xL
Levels in Solid Tumors Are Predictive of Dinaciclib-Induced Apoptosis.
PLoS One 2014, 9, e108371. (b) Cidado, J.; Boiko, S.; Proia, T.;
Ferguson, D.; Criscione, S. W.; San, M. M.; Pop-Damkov, P.; Su, N.;
Roamio, F. V. N.; Sekhar, R. C. C.; D’Santos, C. S.; Shao, W.; Saeh, J. C.;
Koch, R.; Weinstock, D. M.; Zinda, M.; Fawell, S. E.; Drew, L.
AZD4573 Is a Highly Selective CDK9 Inhibitor That Suppresses MCL-
4
1
999, 7, 2189−2197. (d) Skupinska, K. A.; McEachern, E. J.; Baird, I.
R.; Skerlj, R. T.; Bridger, G. J. Enzymatic Resolution of Bicyclic 1-
Heteroarylamines Using Candida antarctica Lipase B. J. Org. Chem.
2
003, 68, 3546−3551. (e) González-Sabín, J.; Gotor, V.; Rebolledo, F.
CAL-B-catalyzed resolution of some pharmacologically interesting β-
substituted isopropylamines. Tetrahedron: Asymmetry 2002, 13, 1315−
1320. (f) Choi, Y. K.; Kim, M. J.; Ahn, Y.; Kim, M.-J. Lipase/Palladium-
Catalyzed Asymmetric Transformations of Ketoximes to Optically
Active Amines. Org. Lett. 2001, 3, 4099−4101. (g) López-Serrano, P.;
Jongejan, J. A.; van Rantwijk, F.; Sheldon, R. A. Enantioselective
acylation of α-aminonitriles catalysed by Candida antarctica lipase.
Tetrahedron: Asymmetry 2001, 12, 219−228. (h) Sánchez, V. M.;
Rebolledo, F.; Gotor, V. Candida antarctica Lipase-Catalyzed Doubly
Enantioselective Aminolysis Reactions. Chemoenzymatic Synthesis of
1
2
(
and Induces Apoptosis in Hematologic Cancer Cells. Clin. Cancer Res.
020, 26, 922−934.
3) Barlaam, B.; Casella, R.; Cidado, J.; Cook, C.; De Savi, C.;
Dishington, A.; Donald, C. S.; Drew, L.; Ferguson, A. D.; Ferguson, D.;
Glossop, S.; Grebe, T.; Gu, C.; Hande, S.; Hawkins, J.; Hird, A. W.;
Holmes, J.; Horstick, J.; Jiang, Y.; Lamb, M. L.; McGuire, T.; Moore, J.
E.; O’Connell, N.; Pike, A.; Pike, K. G.; Proia, T.; Roberts, B.; Martin,
M. S.; Sarkar, U.; Shao, W.; Stead, D.; Sumner, N.; Thakur, K.;
Vasbinder, M. M.; Varnes, J. G.; Wang, J.; Wang, L.; Wu, D.; Wu, L.;
Yang, B.; Yao, T. Discovery of AZD4573, a potent and selective
inhibitor of CDK9 that enables short duration of target engagement for
the treatment of hematological malignancies. J. Med. Chem. 2020, 63,
3
-Hydroxypyrrolidines and 4-(Silyloxy)-2-oxopyrrolidines with Two
Stereogenic Centers. J. Org. Chem. 1999, 64, 1464−1470.
13) (a) Karlsson, S. Development of an Enantioselective Novozym
35 Mediated Acetylation for the Preparation of (1S,3R)-3-
Acetamidocyclohexane-1-carboxylic Acid. Org. Process Res. Dev. 2016,
(
4
̈
0, 1336−1340. (b) Pithani, S.; Karlsson, S.; Emtenäs, H.; Oberg, C. T.
2
Using Spinchem Rotating Bed Reactor Technology for Immobilized
1
5564−15590.
4) Galeta, J.; Tenora, L.; Man, S.; Potácek, M. Dihydropyrrolo[1,2-
b]pyrazoles: withasomnine and related compounds. Tetrahedron 2013,
9, 7139−7146.
5) Amorín, M.; Castedo, L.; Granja, J. R. Self-Assembled Peptide
Tubelets with 7 Å Pores. Chem. - Eur. J. 2005, 11, 6543−6551.
6) (a) Stewart, G. W.; Brands, K. M. J.; Brewer, S. E.; Cowden, C. J.;
Enzymatic Reactions: A Case Study. Org. Process Res. Dev. 2019, 23,
(
1
(
926−1931.
14) Using the bromo-analogue of 8 instead of iodo was attempted at
6
(
an early stage of the route development, but this gave an unacceptable
low yield of 16.
(15) Building block 4 was obtained on multikilogram scale starting
(
from pyrazole. Only minor modifications of the conditions in the first
generation synthesis (Scheme 1) were used. Process descriptions can be
found in the Supporting Information.
Davies, A. J.; Edwards, J. S.; Gibson, A. W.; Hamilton, S. E.; Katz, J. D.;
Keen, S. P.; Mullens, P. R.; Scott, J. P.; Wallace, D. J.; Wise, C. S. Process
Development and Large-Scale Synthesis of a c-Met Kinase Inhibitor.
Org. Process Res. Dev. 2010, 14, 849−858. (b) Newby, J. A.; Huck, L.;
Blaylock, D. W.; Witt, P. M.; Ley, S. V.; Browne, D. L. Investigation of a
Lithium−Halogen Exchange Flow Process for the Preparation of
Boronates by Using a Cryo-Flow Reactor. Chem. - Eur. J. 2014, 20,
2
(
63−271 and references cited therein.
7) Bottles of n-butyllithium were purchased from Aldrich and used
freshly upon arrival. The molarity was assumed to be as specified on the
label.
(8) Znidar, D.; O’Kearney-McMullan, A.; Munday, R.; Wiles, C.;
Poechlauer, P.; Schmoelzer, C.; Dallinger, D.; Kappe, C. O. Scalable
Wolff-Kishner Reductions in Extreme Process Windows Using a Silicon
Carbide Flow Reactor. Org. Process Res. Dev. 2019, 23, 2445−2455.
(9) For example, see: (a) Sieser, J. E.; Maloney, M. T.; Chisowa, E.;
Brenek, S. J.; Monfette, S.; Salisbury, J. J.; Do, N. M. Ir-Catalyzed
Borylation as an Efficient Route to a Nicotine Hapten. Org. Process Res.
Dev. 2018, 22, 527−534. (b) Campeau, L.-C.; Chen, Q.; Gauvreau, D.;
Girardin, M.; Belyk, K.; Maligres, P.; Zhou, G.; Gu, C.; Zhang, W.; Tan,
L.; O’Shea, P. D. A Robust Kilo-Scale Synthesis of Doravirine. Org.
Process Res. Dev. 2016, 20, 1476−1481. (c) Larsen, M. A.; Hartwig, J. F.
O
https://doi.org/10.1021/acs.oprd.1c00058
Org. Process Res. Dev. XXXX, XXX, XXX−XXX