Developing a multistep continuous manufacturing process for (1R,2R)-2-amino-1-methylcyclopentan-1-ol

Article abstract of DOI:10.1021/acs.oprd.0c00405

A multistep continuous manufacturing process to synthesize (1R,2R)-2-amino-1-methylcyclopentan-1-ol (2) was developed. The step 1/2 flow process mitigated the safety hazard associated with high exothermicity during epoxidation, the epoxide-opening reaction, and the low onset of the epoxide intermediate. Process improvements included a hybrid plug flow reactor (PFR)/continuous stirred tank reactor (CSTR) reaction and a continuous quench/work-up process in step 1; a continuous reaction, extraction, washing, and thin-film distillation work-up in step 2; and a continuous trickle bed hydrogenation in step 3, which provided the desired product with high purity and high ee. The end-to-end continuous process provided significant advantages of cost-saving, minimization of manufacturing space and utilities, and reduction of the cycle time.

Full text of DOI:10.1021/acs.oprd.0c00405

pubs.acs.org/OPRD
Article
Developing a Multistep Continuous Manufacturing Process for
(1R,2R)‑2-Amino-1-methylcyclopentan-1-ol
Shengquan Duan,* Xichun Feng, Miguel Gonzalez, Scott Bader, Cheryl Hayward, Tomislav Ljubicic,
Jiangping Lu, Jason Mustakis, Mark Maloney, Joseph Rainville, and Xin Zhang
Cite This: https://dx.doi.org/10.1021/acs.oprd.0c00405
Read Online
Metrics & More
Article Recommendations
ACCESS
ABSTRACT: A multistep continuous manufacturing process to synthesize (1R,2R)-2-amino-1-methylcyclopentan-1-ol (2) was
developed. The step 1/2 flow process mitigated the safety hazard associated with high exothermicity during epoxidation, the
epoxide-opening reaction, and the low onset of the epoxide intermediate. Process improvements included a hybrid plug flow reactor
(PFR)/continuous stirred tank reactor (CSTR) reaction and a continuous quench/work-up process in step 1; a continuous reaction,
extraction, washing, and thin-film distillation work-up in step 2; and a continuous trickle bed hydrogenation in step 3, which
provided the desired product with high purity and high ee. The end-to-end continuous process provided significant advantages of
cost-saving, minimization of manufacturing space and utilities, and reduction of the cycle time.
KEYWORDS: continuous manufacturing, epoxidation, epoxide opening, chiral resolution, trickle bed hydrogenation
resolution by (S)-2-(3,5-dinitrobenzamido)-2-phenylacetic
acid (DNBPG) gives 7 at 95−97% ee and deprotection by
hydrogenation, followed by exposure to p-toluenesulfonic acid
provides compound 2 as the tosylate salt.
1. INTRODUCTION
Cyclin-dependent kinases (CDKs) are a class of cellular
enzymes that play an important role in cell cycle progression.1
Many CDK inhibitors have been evaluated as cancer targets in
the preclinical and clinical studies since the FDA’s approval of
palbociclib in 2015.2 Compound 1 (Figure 1) is a CDK 2/4/6
To support a rapid acceleration for the project, it was
determined that the route would be insufficient to provide the
required quantity of compound 2 for large-scale API
production. While the lack of a commercial source of the
resolution reagent is problematic, the biggest limitations of the
approach on a larger scale are step 2 epoxide opening and step
1 epoxidation. Although the regioselectivity of the step 2
reaction is consistent (at about 5:1 favoring 5), the reaction
requires using water as the solvent,5 and a competitive
hydrolysis reaction was observed at lower temperatures.
Conducting the reaction at elevated temperatures reduces
diol (8) formation but presents a safety hazard since
compound 4 has a very low onset (49 °C even when a base
is present) and high thermal event (506 kJ/kg). Furthermore,
when the reaction was carried out on a larger scale in an
autoclave at 100 °C, a complicated reaction mixture was
obtained, which impacted the subsequent chiral resolution. In
addition, step 1 epoxidation went through a bromohydrin
intermediate, which always gave a mixture with multiple
impurities, and the purification of the epoxide intermediate was
problematic due to the aforementioned safety hazard.
Figure 1. CDK 2/4/6 inhibitor (1) and amino alcohol (2).
inhibitor currently being developed by Pfizer as an oncology
candidate.3 One of the key synthetic challenges for this
compound is the synthesis of a chiral amino alcohol building
block (2).
There is limited literature precedence on the synthesis of
compound 2,4 and several chiral synthetic routes were explored
during development with limited success. The initial route1 for
early clinical supply is illustrated in Scheme 1. The synthesis
starts from the bromination of methylcyclopentene by 1,3-
dibromo-5,5-dimethylhydantoin (DBDMH), followed by base-
mediated cyclization of bromohydrin to provide racemic
epoxide (4). Opening of the epoxide using benzylamine
gives the racemic benzylamino alcohol (5), with ∼5:1
regioselectivity, favoring the desired regioisomer. Chiral
Received: September 8, 2020
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
© XXXX American Chemical Society
A

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 1. Initial Route to Tosylate of Compound 2
Scheme 2. New Route to Compound 2
In recent years, continuous manufacturing processes have
gained momentum in the pharmaceutical industry due to their
many advantages compared to the traditional batch process.6
One important aspect of a continuous process is the increased
safety for thermal hazard reactions by minimizing reaction
volumes and maximizing heat-transfer rates. It also allows for
wider-temperature operation windows and higher-pressure
capabilities. For the problems we encountered in the above
synthesis, a continuous manufacturing process could provide
the best solution. Herein, we describe a new multistep
continuous manufacturing process, which was developed for
the preparation of compound 2 on a large scale.
experiments, the reactivity was too low and the approach
was abandoned. Enzymatic epoxidation was attempted with
chloroperoxidase8 and lipase (Novozyme 435),9 but lower
purity and yield were often obtained, and the same purification
challenge as in the halohydrin route was expected.
The most cost-efficient method identified for the epox-
idation is to use catalytic methyltrioxorhenium (MTO) and
aqueous H2O2 (Scheme 3).10 MTO is stable, easily accessible,
Scheme 3. Epoxidation of Methylcyclopentene and the Diol
Impurity
2. RESULTS AND DISCUSSION
To circumvent the problems identified in the discovery
synthesis, a new route was quickly developed and demon-
strated at the lab scale (Scheme 2). The synthesis starts from
epoxidation of methylcycolpentene using methyltrioxorhenium
(MTO) and H2O2 as the oxidant, followed by epoxide opening
with benzylamine under basic conditions. The chiral resolution
is achieved using the more readily available (S)-mandelic acid,
and the final debenzylation affords compound 2 as a mandelic
acid salt. To respond to the urgent clinical need, we quickly
developed this route into a process that was capable of
producing hundreds of kilograms of amino alcohol 2 using a
continuous process. The details of the process development of
each step are discussed below.
2.1. Step 1Epoxidation. Epoxidation of methylcyclo-
pentene through a halohydrin intermediate was effective to a
certain extent, but the purity of the product was not
satisfactory for direct telescoping to the next step. Isolation
and purification of the crude product were proven necessary to
ensure an acceptable purity profile for the successful execution
of the resolution step. Purification by vacuum distillation was
also feasible but presented safety risks on a larger scale.
In an effort to improve the process and avoid the classical
resolution, (R,R)-Jacobsen’s catalyst [(R,R)-(−)-N,N’-bis(3,5-
di-tert-butylsalicylidene)-1,2-cyclohexanediaminomanganese-
(III)chloride] was explored briefly, but it failed to provide
satisfactory stereoselectivity.7 This was moderately improved
using a modified chiral Mn-Salen catalyst, but in these
and the oxidation of the olefin proved to be both highly active
and highly selective. The addition of catalytic amounts of
pyridine suppresses the ring opening and diol formation,
resulting in reaction mixtures of high purity that can be used
directly in the subsequent step without additional purification.
In the initial laboratory-scale experiments, substoichiometric
H2O2 (0.95 equiv) was used as neat reactions, resulting in
crude reaction mixtures being carried over to the next step
without the quench and work-up process.
However, the process safety testing of this reaction indicated
that the heat of the reaction was 686 kJ/mol and an adiabatic
temperature of 1075 °C was observed. Elongated reaction
times and slow addition of H2O2 assisted in heat control, but
both changes resulted in higher levels of diol and other
impurities. Through additional experiments, it was found that
the addition of tetrahydrofuran (THF) to the reaction mixture
alleviated the heat output, with no detrimental effects on the
reaction impurity profile. It was found that the addition of 2
vol of THF to the reaction greatly reduced the exotherm to
218 kJ/mol and the adiabatic temperature increased to 350 °C.
While this increased the safety of performing the reaction on a
larger scale, the reaction was still limited to volumes of ≤80 L
to meet the safety threshold, resulting in a total of 15 smaller
batches for a 50 kg delivery.
B
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 4. Epoxidation Mechanism
2.1.1. Flow Epoxidation. To address these limitations, a
flow process was studied for possible application and
increasing scale-up possibilities for the reaction. An initial
assessment of a plug flow reactor (PFR) gave low conversion
due to the formation of a biphasic layer and insufficient mixing.
While the continuous stirred tank reactor (CSTR) gave similar
results to those in the batch reaction, the heat dissipation
capacity was not ideal.
To understand the scalability of this reaction, we tried to
establish a kinetic model based on the literature mechanism.11
However, this task was not trivial, as quantification of the
reaction product and intermediates proved challenging.
Neither NMR nor infrared (IR) was suitable due to the high
noise/signal ratio, while GC was abandoned due to epoxide
product decomposition during analysis. Finally, we constructed
a model (Scheme 4) generated from a multiportion feed
experiment in an RC1. The confidence intervals in this model
were not in the expected range, but the model was consistent
with the empirical data generated in the lab. The empirical
model was sufficient for us to model the mixing in scale-up
equipment.
The preliminary assessment suggested that better mixing
and a longer residence time were beneficial for the reaction
conversion and the higher reaction temperature was
detrimental. Low conversions were observed with PFR due
to poor mixing, but the use of a static mixer (helix-type) helped
the conversion only to an extent. Increasing the static mixer
length from 10 to 170 cm while keeping the residence time at 9
min failed to improve the conversion further, and the residence
time became the critical factor for the further conversion of the
reaction.
For the PFR + static mixer system, a high flow rate (∼0.106
m/s) was required to achieve the desired mixing. When the
system was scaled down (using 1.7 mm OD helical type in a
DN3 (0.5 mm wall thickness) tubing), 60 min residence time
was needed for full conversion, which translated into an
extremely long PFR for scale-up purposes.
since it provided better mixing and a longer residence time.
The combination of PFR + CSTR was tested, providing
comparable results to the laboratory batch process. In total,
80−85% conversion of the methylcyclopentene and a low assay
yield of 80% were consistently obtained with the use of 0.95
equiv of H2O2.
2.1.2. Step 1Quench and Work-up. In our initial
hypothesis, the diol impurity (8) was formed due to excess
H2O2 reacting with MTO, which then reacts further with the
epoxide. Additional studies suggested that the diol was prone
to form under acidic conditions, and in the reaction, the pH
became acidic following quenching with sodium bisulfite.
Although NaHSO3 is slightly basic, the reaction between
NaHSO3 and H2O2 resulted in an overall reduction in pH. A
typical pH after quenching was measured at 4−5 if 1.2 equiv of
H2O2 was used. This resulted in up to 50% hydrolysis of the
desired product at this pH. When 2.0 equiv of H2O2 was used,
the pH was further lowered to 2−3 and 100% product was
completely hydrolyzed. Switching the quench from NaHSO3 to
Na2SO3 solved this problem nicely, as we observed a neutral
pH (6−7), which greatly minimized the hydrolysis, and less
than 8% diol was formed when 1.5 equiv of H2O2 was used
under these conditions.
The impact of this change is significant. First, excess H2O2
can be used instead of substoichiometric amounts, which
maximizes the utilization of methylcyclopentene and improves
the in situ yield to more than 90%. Second, it also allowed us
to reduce the quantity of the MTO catalyst, one of the major
cost contributors for the new route. It was found that the
catalyst could be reduced to as low as 0.0016 equiv if 1.5 equiv
of H2O2 was used and the residence time was increased to 80
min (Table 1).
For step 1 quenching, a continuous process using a PFR was
selected for efficient heat removal. While perfluoro alkoxy
resins like Teflon have a reduced heat-transfer capacity
compared to metals, they can allow efficient heat release at
localized points while minimizing cost. Therefore, a Teflon coil
(4 mm ID) was used and packed with ∼1.5 mm glass beads as
a static mixer. The residence time was set at 3−4 min, and no
temperature increase was observed in the receiving tank during
quenching, suggesting adequate heat removal. An assay yield of
88.2% with 4.6% diol impurity was observed post work-up.
While neither PFR nor CSTR gave satisfactory results, a
combination of the PFR + CSTR system seemed to provide a
better solution. It was envisioned that the PFR coil could be
used for managing most of the reactive heat at the initial stage
of the reaction as it was more efficient for heat exchange, and
the CSTR could be used to drive the reaction to completion
C
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
0.0032 equiv of MTO and 1.2 equiv of H2O2 for the scale-up,
which allowed us to manufacture 82 kg per day.
Table 1. Step 1 Catalyst Loading Study
cat.
loading
no. (equiv)
equiv of
H2O2
(equiv)
residence
time
(min)
The flow mode configuration is shown in Figure 2. Reaction
PFR and a cascading three-stage CSTR were used for step 1,
while a combination of a PFR followed by a CSTR was used
for quenching. The quenched product was continuously
transferred to a series of rotating disk columns (RDC),
where the product was phase-separated in the first column
followed by a continuous washing of the organic phase with a
14% Na2SO3 aqueous solution in the second column. The
organic phase after column 2 was collected and directly used in
step 2. To our satisfaction, during the scale-up run, there was
no starting material (SM) residue and the assay yield was
91.8%, while the diol impurity was controlled to 4.7%.
2.2. Step 2Epoxide Opening and Chiral Resolution.
Direct aminolysis of the epoxide 4 with ammonia (7 N in
methanol) was attempted, as it gave racemic 2 directly without
debenzylation. The reaction worked under elevated pressures
and temperatures using purified epoxide. However, the
intrinsic safety hazard rendered by this approach was deemed
unsuitable for scale-up under batch conditions. Flow
conditions were considered, but very slow kinetics for the
transformation was observed. Thus we focused our effort on
epoxide opening using benzylamine (Scheme 5).
conversion
(%)
results post
quenching
1
2
3
0.0064
0.0032
0.0016
0.95
5
30
80
100
product, 83%; SM,
0; diol, 4.98%
product, 92%; SM,
0.6%; diol, 3.3%
product, 82.02%;
SM, 7.39%; diol,
4.84%
1.2
88
1.5
86.7
With the newly developed continuous process, only an 8 L
flow reactor (2 L of PFR + 6 L of CSTR) and 1 day were
needed for a 100 kg scale production. In comparison, the batch
process on the same scale required 10 batches and at least 5
days to deliver. In addition, the continuous work-up process
(e.g., quenching, extraction) helped to improve productivity
effectively by further reducing the cycle time by about 2−3
days.
For scale-up, we needed to balance the cost-saving from
catalyst usage along with the throughput of the process since
the minimal catalyst loading (0.0016 equiv) required a longer
residence time. In addition, considerable amounts of heat
needed to be removed at the lowest catalyst loading using 1.5
equiv of H2O2. The decision was ultimately made to choose
2.2.1. Epoxide Opening Reaction Design. A sealed
autoclave was initially used to carry out the epoxide opening
Figure 2. Step 1 scale-up flow process configuration.
D
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 5. Epoxide Opening with Benzylamine and Chiral Resolution
Figure 3. Step 2 work-up unit of operations in the batch process.
reaction and resulted in complete conversion after 18 h at 100
°C. Excess benzylamine was needed to reach reaction
completion; however, a competition involving epoxide
hydrolysis and polymerization side reactions was significant.
In addition, the unreacted benzylamine and formed impurities
proved to be a challenge for removal, requiring chromato-
graphic purification of the intermediate prior to the chiral
resolution.
A continuous flow process was evaluated and showed
promising results.12 It was found that a solvent of THF/water
(4:2 vol) could provide a homogeneous mixture that was
suitable for PFR. The reaction mixture was stable up to 24 h in
the presence of dilute NaOH; therefore, a single feed stream
was suitable. Without the need to consider mixing, the impact
of the reaction temperature, pressure, and residence time on
conversion/regioselectivity was evaluated. Using a Propel
segmented reactor, it was found that the best conversion and
selectivity can be achieved at 210−220 °C under 3.8−4.5 MPa
with 1 h residence time.
For scale-up, the conditions generated in the lab allowed an
easy adaptation on the scale. The reaction temperature range
was investigated from 180 to 220 °C, and the best conditions
identified were about 200 °C. At lower temperatures (180 °C),
higher levels of unreacted start material (more than 5%) were
observed, while at higher temperatures (220 °C), the impurity
level increased and led to lower in situ yields. The optimum
residence time was 60−90 min since lower conversion was
obtained at 30 min residence time. With the easy scale-up of
the PFR conditions for the step 2 reaction, we turned our focus
to the continuous work-up design.
were excluded for further investigation based on the solubility
of the product in these two solvents, and DCM was not
considered due to the insolubility of L-mandelic acid in DCM.
Among the other three solvents, when the extract after brine
wash was used directly for chiral resolution, no solids were
observed in 2-MeTHF and only benzylamine salt was isolated
from MTBE and IPAc. Suspecting that residual water in the
crude solutions may have been the cause, Karl−Fischer (KF)
analysis was performed, which indicated 5.2−5.8% water in
these solutions. By distillation of these solutions to NMT 1%
H2O, the desired product was obtained, although benzylamine
salt was present in the isolated material.
These experiments indicate that azeotropic distillation is
required to control KF before the chiral resolution. Using a
conventional batch process, this would require five iterations of
distillation to achieve the desired KF and posed a time and
throughput challenge to our manufacturing operations on a
larger scale. To this end, continuous extraction, continuous
washing, followed by a continuous concentration process was
developed. RDCs like the ones used in the continuous work-up
of step 1 were used to develop the continuous extraction and
washing process.
THF and MTBE (3 vol) were evaluated as the solvents for
continuous extraction. Both showed <0.1% product loss in the
aqueous phase and good recovery of the product in the organic
phase. However, there was no differentiation between the
product and the residual BnNH2, which was also completely
extracted to the organic phase. The KF of the organic phase
with MTBE (5.6%) as the solvent was slightly lower than that
with THF (6.5%) as the solvent.
2.2.2. Step 2Continuous Work-up Design. In the batch
process, multiple units of operations were required before
chiral resolution (Figure 3).
Since the KF of the organic phase was to be controlled to
NMT 1%, continuous concentration using thin-film evapo-
ration (TFE) was evaluated. In our initial conditions using a
flow rate of 33 g/min and a distillation temperature of 40−41
°C, it was found that 5 iterations of distillation were also
needed to control the KF to NMT 1% for both THF and
MTBE. Further studies showed that when the flow rate was
reduced to 27.8 g/min and the temperature increased to 46−
47 °C, the dehydration efficiency was greatly improved and
required only two iterations of distillation. MTBE was
ultimately selected as the solvent for continuous extraction
The extraction/wash/distillation operations relied on a long
residence time and high energy consumption, so our initial
effort was to simplify the work-up process. A short sequence of
extraction, washing, and chiral resolution was ideal. Six
different solvents (tetrahydrofuran (THF), acetone, 2-methyl-
tetrahydrofuran (2-MeTHF), isopropylacetate (IPAc), tert-
butylmethylether (MTBE), and dichloromethane (DCM))
were first screened for the extraction. The THF and acetone
E
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 4. Step 2 reaction and work-up flow process configuration.
and distillation as it contained lower water in the organic
phase.
our attention to the amount of the resolving agent employed. It
was found that a slight increase of mandelic acid to 0.6 equiv
helped slightly in the recovery of our product and that
increasing the equivalents further (0.7) was not beneficial and
could be due to an intrinsic equilibrium among the many
components in the mixture. Although the acetone conditions
were selected for scale-up, the material loss was deemed
unacceptable and the resolution needs to be redesigned to
improve the yield of this step.
On a larger scale, we envisioned running the chiral
resolution in a CSTR, instead of developing a more
technologically complicated continuous crystallization process.
The product/acetone solution and mandelic acid/acetone
solution were mixed in a CSTR, and the resulting slurry
overflowed into an agitated filter dryer (AFD) to complete the
filtration and drying. Two parallel AFDs were used in
production to maintain continuity of the process by alternating
between the vessels.
In summary, a more efficient continuous work-up process
was developed for step 2 that included a continuous extraction
column for extraction and washing, and thin-film evaporation
(TFE) was used in the continuous concentration to reach the
desired water content. The continuous work-up process
provided a product with a similar yield (15−22%) and ee
(99−100%), as well as levels of the BnNH2 residue and
product loss compared to the batch process.
The flow mode configuration for step 2 reaction and work-
up is shown in Figure 4. A PFR (coil 1) was used for the
reaction, while extraction column 1 was used for continuous
extraction with MTBE and washing with a NaCl aqueous
solution. Extraction column 2 was used for continuous washing
with the NaCl aqueous solution and phase separation. TFE
was used for continuous distillation, and the solution after TFE
was directly used for chiral resolution.
2.2.3. Step 2Chiral Resolution. In the early phase of the
project, (S)-2-(3,5-dinitrobenzamido)-2-phenylacetic acid
(DNBPG) was used for the chiral resolution. However, this
resolving agent was difficult to obtain on the scale, and a
reslurry was required to upgrade the chiral purity to acceptable
specifications. A screening study was conducted to find
additional resolution reagents, and four chiral acids were
identified that gave acceptable er: (1S,3R)-camphoric acid, S-
mandelic acid, R-Boc-TIC-OH, and Z-L-phenylglycine. Based
on the commercial availability and cost, S-mandelic acid was
selected for further optimization. On the laboratory scale, the
best results were obtained with 20 vol of acetone and 0.52
equiv of mandelic acid.
Although the mandelic acid resolution process gave the
product with high chiral and chemical purity, it was found that
the loss of the product in the mother liquor was significant
(amount to 18−22%). The typical composition of the mixture
after work-up was complicated with ∼74% of racemic 5, 11%
of regioisomer 6, 6% of benzylamine, and 8% of other
impurities. All of these components can form salts with the
resolution reagent, which significantly complicated our under-
standing of the crystallization process. Initial attempts to
isolate racemic 5, either as a free base or a salt, did not provide
any benefit, as it added an additional unit of operation and the
recovery was poor.
Additional optimization on the chiral resolution conditions
was attempted, starting with the use of R-mandelic acid to
remove the undesired enantiomer before adding S-mandelic
acid for resolution. While more than 50% of the undesired
isomer was successfully removed using R-mandelic acid, the
resolution of the enriched material did not improve and a
substantial loss in the mother liquor was still observed. We
next focused on screening the resolution solvent, and it was
observed that no solvents were effective in minimizing the
mother liquor loss. In many cases, our attempts resulted in the
coprecipitation of BnNH2 mandelic acid salt. We then turned
Comparing the processes for step 2 (Table 2), it should be
noted that approximately 3−4 fewer unit operations are used
Table 2. Comparison of Equipment Use and Cycle Time for
Step 2
cycle
time
(days)
process
no. of unit operations
equipment
batch
process
8−9 concentrations; 5−6 3000 L glass-lined reactor ×
∼8
extractions; 1 washing;
1 chiral resolution
2; 1000 L glass-lined
reactor × 1
flow
process
5−6 concentrations; 2−3 3000 L glass-lined reactor ×
∼6
extractions; 1 washing;
1 chiral resolution
1; continuous extraction
column × 2; TFE × 1
in the flow process compared to the batch mode, which has a
direct impact on the cost of operation. For an 80 kg scale
campaign, the cycle time was reduced from 8 to 6 days and
benefitted from a significantly reduced manufacturing footprint
as well.
F
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
2.3. Step 3Trickle Bed Hydrogenation. To prepare
the desired amino alcohol 2, 7·mandelic acid salt post
resolution is hydrogenated using a heterogeneous palladium
catalyst (Scheme 6). Under batch conditions, the debenzyla-
Pd/C catalyst with 625−675 μm support size was selected
because of lower pressure drops across the reactor. The larger
particle catalyst was tested with batch processing conditions to
understand the reaction rate. >99% conversion was achieved
after 16 h at 40−50 °C. When the reaction temperature was
increased to 58−62 °C, the reaction went to completion in 10
h, which suggested faster kinetics at higher reaction temper-
atures. The trickle bed hydrogenation conditions were
evaluated with the equipment setup shown in Figure 5. The
starting material solution was prepared by dissolving 7·
mandelic salt in 10 vol of MeOH.
Scheme 6. Hydrogenolysis of 7·Mandelic Acid Salt
The flow process was tested initially at 85 °C (entry 1, Table
3). The pressure was set at 0.5−0.6 MPa (5−6 bar). The flow
rate of the starting material solution was set at 2.5 g/min. The
H2 flow rate was set at 0.1−0.2 L/min, and the estimated
residence time was ∼10 min. At full conversion, a crude purity
of 98.67% and ee of 99.96% were observed. Under these
conditions, the residence time could be further reduced to 5
min by increasing the flow rate, producing similar conversions
(Table 3, entry 2).
The flow rate of the starting material can be further
increased for throughput improvement (Table 3, entries 3−5).
When the flow rate was increased from 2.5 to 12.5 g/min, the
conversion slightly decreased from >99.9 to 99.4%. When the
flow rate was maintained at 12.5 g/min, and the pressure was
increased from 0.5 MPa (5 bar) to 0.7 MPa (7 bar), the
tion of intermediate 7 reaches complete conversion in 6 h
using 5% Pd/(OH)2/C as the catalyst and a loading of 1.5% at
40−50 °C under 4−6 bar hydrogen pressure.
Trickle bed hydrogenation has been used in other industries
for decades13,14 and in recent years has been gradually adopted
by the pharmaceutical industry.15−18 Considering the intrinsic
process safety hazard associated with the use of H2 on a large
scale and the flammability of Pd(OH)2/C, it is logical to
extend continuous processing to the hydrogenation step.
A few Evonic larger particle size catalysts were evaluated
following an internal evaluation, and a type of impregnated
Figure 5. Diagram for step 3 trickle bed hydrogenation.
G
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Table 3. Trickle Bed Hydrogenation Evaluation
no.
T (°C)
P (MPa)
residence time (min)
SM flow rate (g/min)
H2 flow rate (L/min)
equiv ratio H2/ 7
IPC
1
2
3
4
5
6
7
85−88
85−88
85−88
85−88
85−88
85−88
65−68
0.5
0.5
0.5
0.5
0.5
0.7
0.5
10
5
3.3
2.5
2
2.5
5
7.5
10
12.5
12.5
2.5
0.1−0.2
0.1−0.2
0.1−0.2
0.1−0.2
0.2−0.3
0.2−0.3
0.1−0.2
6.4−12.8
3.2−6.4
2.2−4.4
1.6−3.2
2.6−3.6
3.6−5
conversion, 100%
conversion, >99.9%
conversion, 99.9%
conversion, 99.8%
conversion, 99.4%
conversion, 99.5%
conversion, 99.9%
2
10
6.4−12.8
Scheme 7. Improved Continuous Process to 2·Mandelic Acid Salt
conversion returned back to >99.9%. It was also found that the
reaction temperature could be reduced from 85−88 °C to 65−
68 °C if the flow rate of the starting material was kept at 2.5 g/
min; the reaction still gave 99.9% conversion (Table 3, entry
7).
high exothermicity of the reaction and achieved desired
conversions. Switching the quench from NaHSO3 to Na2SO3
was a key finding that not only minimized the hydrolysis but
also allowed the use of excess H2O2 to drive the reaction to
completion and reduce the catalyst loading. In step 2, a
continuous reaction and work-up were designed and
demonstrated on the scale. In step 3, a continuous trickle
bed hydrogenation was developed for debenzylation, and this
process is expected to accrue the cost-saving benefit in the
future.
Finally, a mixture with er 80:20 was tested to understand the
process tolerance of the enantiomeric impurity. Again, the
reaction reached 99.8% conversion within 10 min. Post
isolation, UPLC analysis indicated a purity of 96.89% and an
improvement in ee (99.2%). This suggested that the excess
undesired enantiomer can be purged further in the final
isolation due to the presence of mandelic acid, and this
provides an opportunity to boost the yield in step 2.
The crystallization of 2 proved straightforward. After the
solvent was switched from MeOH to acetonitrile (7−9 vol),
the product crystallized upon cooling to 20−30 °C. The
MeOH residue was controlled to below 1%, minimizing
product loss in the ML. A demonstration batch at a 4.8 kg scale
was conducted to have a proof of concept for the trickle bed
process. IPC showed the conversion to be >99.9%. After
crystallization and isolation, the material was of excellent ee
(100%) and good purity (98.7%) in 73.1% isolated yield.
While the flow hydrogenation process only provided a slight
benefit in yield compared to the batch process, it delivered the
major advantage of a safer process. For the flow process, H2 is
released in smaller amounts continuously and the diminished
size of the reactor (1−30 L) greatly minimized the risk brought
by handling H2. The cost was another important factor. As the
reactor and conditions remain the same, the cost is solely
dependent on the catalyst cost. For one-time use, the catalyst
cost for the flow process was 3× higher than the batch process
at a 50−100 kg scale. However, our development work has
demonstrated that no catalyst deactivation was observed, even
after multiple uses. In the longer term, the cost of the catalyst is
expected to be significantly lower with an increase in volume
and reuse of the catalyst for multiple campaigns.
Table 4 summarizes the cycle time for each step and the
number of batches needed to deliver 100 kg of the product.
Table 4. Cycle Time Comparison between the Batch
Process and the Flow Process
cycle time
process
process
step 1
step 2
step 3
total
batch
30 batches,
16 days
4 batches,
23 days
1 batch,
8 days
47 days
flow
process
1 batch, 4 days 1 batch,
16 days
1 batch,
7 days
27 days
Compared to the batch process, the flow process is capable of
delivering the same amount of material in nearly half the time.
The average cost-saving per kg of the final product for the
continuous process is about 30%. The new process also
demonstrates a significant advantage due to the reduced
utilization of manufacturing space and equipment while
simultaneously reducing the utilities needed to deliver the
same amount of material with comparable quality.
4. EXPERIMENTAL SECTION
All reactions were performed under a nitrogen atmosphere
unless otherwise stated. All reagents purchased from vendors
were used as received.
1
In step 1, H NMR was used to track the ratio of SM/
3. CONCLUSIONS
product during the reaction and determine the in situ reaction
yield. NMR data was collected using a Bruker AV NEO
400MHz spectrometer with probe Z163739_0076 (PI HR-
400-S1-BBF/H/D-5.0-Z SP).
In summary, an end-to-end continuous process was developed
for the manufacturing of compound 2 (Scheme 7). In step 1, a
combined PFR + CSTR system was designed to handle the
H
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
In step 2, the reaction and starting material consumption
was monitored by GC. GC conditions: Restek RTX-5 Amine;
30 m × 0.32 mm × 1.0 μm; P/N:12354; 250 °C; flow 2.0 mL/
min; 1 μL injection volume; carrier gas, helium; detector gas,
H2/Air/N2 = 40:400:30 (mL/min); gradient 40 °C for 2 min;
ramp at 15 °C/min to 260 °C, hold for 5 min; diluent,
methanol. The received mixture and final product were
analyzed by UPLC. UPLC conditions: Waters HSS T3, 2.1
× 100 mm; 1.8 μm; 25 °C; flow 0.4 mL/min; λ = 210 nm; 1
μL injection volume; A, 0.1% MSA in water; B, acetonitrile;
gradient 5% B to 95% B in 7.5 min, re-equilibrate to 5% B in
0.1 min, run time 10 min; diluent, 10:90 acetonitrile/water.
The enantiomeric purity of step 2 product was monitored by
chiral SFC. Chiral SFC conditions: Chiralpak IC; 4.6 × 250
mm2; 5.0 μm; 40 °C; flow 2.0 mL/min; λ = 210 nm; 5 μL
injection volume; A, CO2; B, 0.1% MSA and 0.1% IPAm in 50
ACN/50 MeOH (v/v); gradient 20% B held in 0.5 min, ramp
to 40% B in 8.5 min, re-equilibrate to 20% B in 0.5 min, run
time 15 min; diluent, methanol.
In step 3, the reaction was monitored by reverse-phase
UPLC. UPLC conditions: Waters HSS T3; 2.1 × 100 mm2; 1.8
μm; 25 °C; flow 0.4 mL/min; λ = 210 nm; 1 μL injection
volume; A, 0.1% MSA in water; B, acetonitrile; gradient 5% B
to 95% B in 7.5 min, re-equilibrate to 5% B in 0.1 min, run
time 10 min; diluent, 10:90 acetonitrile/water. The chiral
purity of step 3 product was monitored by reverse-phase
UPLC after Marfey’s derivatization. Derivatization method:
dissolve 4.40 mg of Marfey’s reagent to 2 mL of ACN; 100 μL
of sample solution was placed in a total recovery HPLC vial
along with 50 μL of 1:1 water/ACN, 150 μL of Marfey’s
reagent solution, and 40 μL of 1 M NaHCO3. The solutions
were heated at 40 °C for 1 h and then quenched with 40 μL of
1 M HCl and 120 μL of more ACN to help solubilize any
solids that precipitated. UPLC conditions: BEH C18; 2.1 × 50
mm; 1.7 μm; 40 °C; flow 0.6 mL/min; λ = 210 nm; 1 μL
injection volume; A, 0.1% trifluoroacetic acid in water; B,
acetonitrile; gradient 15% B to 45% B in 7.5 min, re-equilibrate
to 15% B in 0.5 min, run time 10 min; diluent, 1:1 acetonitrile/
water.
continuous extraction column 1 was cooled to 15−25
°C. The continuous extraction column 2 was cooled to
5−15 °C.
5. Run the flow process: Pump 1 was connected with
hydrogen peroxide (284.1 kg in total, 35 w/w%, 2924.6
mol, 1.5 equiv) tank, pump 2 was connected with S.M.
solution tank of Ins.1, pump 3 was connected with the
catalyst solution of Ins. 2, and pumps 4 and 5 were
connected with sodium sulfite solution tanks of Ins. 3.
The flow rates were set and tuned automated by an
automated system. The flow rates and retention time of
every stage can be found in Figure 2. Run pumps 1, 2,
and 3 simultaneously, first. IPC samples were taken
periodically at the outlet of reaction coil and reaction
CSTR to track the ratio of SM/product by NMR. When
the mixture in reaction CSTR (stage 3 CSTR) reached
the overflow volume, the reaction mixture would be
transferred into a quenching coil and then into a
quenching CSTR. We started pump 4 when the
transferring started. The quenched mixture from the
quenching CSTR was transferred into the extraction
column 1 by a second transfer pump. The organic phase
in extraction column 1 overflowed into a transfer tank
and then was transferred into the continuous extraction
column 2 by a third transfer pump. We started pump 5
at this time. The organic phase in the extraction column
2 overflowed into the product tank. When all of the
materials in the tanks were pumped, we connected the
pumps into the THF tank to push out and exchange the
reaction mixture in the coils and CSTRs. Also, when all
of the organic phase was collected, we sampled from the
product tank to determine the product assay and the
product solution will be used in the next step without
further purification. In total, 405 kg of colorless solution
was obtained. The assay was 39% analyzed by 1H NMR.
4.2. Preparation of 7·Mandelic Acid Salt by the
Continuous Process.
1. Prepare the benzylamine/catalyst solution: Purified
water (316 kg, 2 V), sodium hydroxide (0.7 kg, 17.5
mol, 0.01 equiv), benzylamine (173.1 kg, 1615.5 mol,
1.0 equiv), and THF (281 kg, 2 V) were charged into a
glass-lined reactor at 20−30 °C and stirred until all
materials dissolved. Then, the mixture was discharged
into the BnNH2 solution tank.
2. Prepare the 20% sodium chloride solution: Purified
water (621.2 kg) and sodium chloride (161.2 kg) were
charged into a glass-lined reactor and stirred at 20−30
°C until all materials dissolved. Then, the mixture was
discharged into two NaCl aq. tanks.
3. Heating: We pumped THF into the coil reactor and
pressurize the coil to 3.5−5.0 MPa by a backpressure
valve. We heated the coil reactor to 195−205 °C.
Meanwhile, we maintained the pressure at 3.5−5.0 MPa.
4. Run the flow process: Pump 1 was connected with the
compound 4 solution (405 kg in total, 39 w/w%, 1609.9
mol, 1.0 equiv) tank (step 1 product solution from
Section 4.1), pump 2 was connected with the BnNH2
solution tank of Ins.1, pump 3 was connected with the
MTBE tank, and pumps 4 and 5 were connected with
sodium chloride solution tanks of Ins. 2. The flow rates
were set and tuned by an automated system. The flow
rates and retention time of every stage can be found in
4.1. Preparation of Epoxide (4) by the Continuous
Process.
1. Prepare the S.M. solution: THF (142.6 kg, 1 V),
pyridine (4.7 kg, 59.4 mol, 0.03 equiv), and 1-
methylcyclopent-1-ene (166.7 kg, 96 w/w%, 1947.9
mol, 1.0 equiv) were charged into a glass-lined reactor at
20−30 °C and stirred until all materials dissolved. Then,
the mixture was discharged into the S.M. solution tank.
2. Prepare the catalyst (MeReO3) solution: THF (142.6
kg, 1 V) was charged into a glass-lined reactor at 20−30
°C and stirred. Methyltrioxorhenium (1.6 kg, 6.4 mol,
0.00328 equiv) was charged into the reactor at 20−30
°C and stirred until all materials dissolved. Then, the
mixture was discharged into the catalyst tank.
3. Prepare the 14% sodium sulfite solution: Purified water
(480.7 kg) and sodium sulfite (80.1 kg, 635.7 mol, 0.33
equiv) were charged into a glass-lined reactor and stirred
at 20−30 °C until all materials dissolved. Then, the
mixture was discharged into two Na2SO3 aq. tanks.
4. Cooling: The reaction coil jacket temperature was
cooled to 5 2 °C and the reaction CSTRs to 5−15
°C. The quench coil jacket temperature was cooled to
5−15 °C and the quench CSTR to 15−25 °C. The
I
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
333 mmol); catalyst, charged palladium; 5% Pd/C
Evonic, 600 μm (120 g, 112.8 mmol). The rough
measurement is ∼1 g of catalyst/0.25″ column height.
Top pack, charged silica gel (top pack) (69.5 g, 30.0 mL,
1160 mmol). The rough measurement is ∼5″ of column
height.
Figure 4. First, pumps 1 and 2 were run simultaneously.
Sampled in a certain interval at the outlet of reaction coil
to track the residue of SM and product purity by GC
and UPLC. The reaction mixture flowed into the first
storage tank then was transferred into the extraction
column 1 by a transfer pump. We started pumps 3 and 4
when the transfer pump was started to wash and extract
the crude product. The organic phase in the extraction
column 1 overflowed into a second storage tank and
then was transferred into the continuous extraction
column 2 by a second transfer pump. We started pump 5
when the second transfer pump was started. The organic
phase in the extraction column 2 overflowed into the
third storage tank. The organic phase in the third storage
tank would be transferred continuously into the thin-film
evaporator (TFE) to remove water azeotropically. When
all of the materials in the tanks were pumped, we
connected the pumps into the THF tank to push out the
reaction mixture in the coil.
3. Pressure test: Start N2 flow at 1000 cc/min and allow to
ramp to the pressure limit; confirm that N2 flow stops
and the N2 valve closes after reaching the pressure limit;
confirm that no leaks are present at any adjusted fitting
with soap solution; monitor the system pressure for
pressure drop (target <1 psi/min loss).
4. Column inertion: Confirm the backpressure valve is in
manual mode and set to −5%; switch on the N2 flow at
400 cc/min and flow gas for 1 min; confirm proper feed
and column output valve positions; switch on the
methanol feed at 50 cc/min; confirm that the pump is
primed; confirm flow to waste jug and check for leaks;
wait for 10 min to the N2 inert system.
5. Hydrogenation: Bring the system to operating con-
ditions under nitrogen: 725 psi (50 bar), 96 cc/min
solvent flow, circulator temperature of 65 °C. Switch the
gas feed to H2; set the flow valve to the auto mode;
equilibrate the system under H2 gas and process solvent
conditions for 20 min. Switch the feed to process stream
flowing under operating conditions for the duration of
the experiment: liquid feed, 90 mL/min; hydrogen feed,
43.65 mL/min; jacket temperature, 65 °C; system
pressure, 50 bar. Run the reaction until the feed solution
is consumed. Once reaction feed is complete, switch to
methanol to flush the column. The product solution is
combined for crystallization.
6. Crystallization: Distill the crude product solution under
vacuum at 50 °C until ∼6 L. Add acetonitrile (2 × 60 L)
and continue to distill to ∼45 L and make sure NMT
1.0% methanol is in the mixture. Heat the batch to 45
°C until all solids are dissolved. Cool the mixture to 20
°C over 6 h. Filter the slurry, wash the filter cake with
acetonitrile, and dry the product at 40 °C for 12 h. In
total, 2.624 kg of white solids was isolated. Yield, 73.1%;
purity, 98.7%; chiral purity, 100%.
5. Resolution: When the water content in the organic
phase was qualified, we concentrated and exchanged
THF with acetone. The final acetone solution should be
1400−1900 L (9−12 V). The mixture was heated to
40−50 °C. (S)-(+)-Mandelic acid (94.3 kg, 620 mol, 0.6
equiv of the racemic product) was added portion-wise to
the mixture (20−30 kg in each portion) at the interval of
10−20 min. The mixture was maintained at 40−50 °C
for 2−3 h, then cooled to 10−15 °C, and stirred for 10−
15 h. Then, the mixture was filtered, and the collected
Mandelic acid salt of compound 7 was dried. In total,
87.7 kg of white solid was obtained. The 1H NMR assay
was 99.6%. The UPLC purity was 100%, and the ee was
99.2%. The yield was 15.2%.
4.3. Preparation of 2·Mandelic Acid Salt by the Batch
Process. A 3000 L autoclave was charged MeOH (692.8 kg,
10 V) and 7·mandelic acid salt (87.1 kg, 243.7 mol, 1.0 equiv)
and stirred for 1−2 h. Then, Pd(OH)2/C (5%, 17.4 kg, 0.2 g/
g, wet basis) was added. Oxygen was exchanged with nitrogen
eight times and then with hydrogen eight times. The autoclave
was pressurized to 0.6 MPa, and then, the mixture was heated
to 40−50 °C. The mixture was allowed to react at 40−50 °C, P
= 0.4−0.6 MPa. After 6 h, the mixture was sampled for UPLC
analysis to assure compound 7 transformed completely. The
mixture was cooled to 20−30 °C and then filtered with a 1000
L stainless steel Nutsche filter. Methanol (278 kg, 4 V) was
used to rinse the autoclave and the cake. The filtrate and
rinsing liquor were combined in a 1000 L glass-lined reactor.
The mixture was concentrated under vacuum and exchanged
with acetonitrile to remove MeOH. The final acetonitrile
solution should be 530−700 L (6−8 V). The acetonitrile
solution was cooled to 10−20 °C and stirred for 3−5 h. We
filtered, dried, and then collected the white solid (50.7 kg).
The final yield was 76.5%. The assay by1H NMR was 98.5%,
and the ee was 100%.
AUTHOR INFORMATION
Corresponding Author
■
Shengquan Duan − Chemical Research and Development, Pfizer
Worldwide Research and Development, Groton, Connecticut
06340, United States; orcid.org/0000-0002-8126-2592;
Email: Shengquan.duan@pfizer.com
Authors
Xichun Feng − Asymchem Life Science (Tianjin) Co., Ltd.,
Tianjin 300457, P. R. China
Miguel Gonzalez − Asymchem Inc., Morrisville, North Carolina
27516, United States
Scott Bader − Chemical Research and Development, Pfizer
Worldwide Research and Development, Groton, Connecticut
06340, United States
Cheryl Hayward − Chemical Research and Development, Pfizer
Worldwide Research and Development, Groton, Connecticut
06340, United States
4.4. Preparation of 2·Mandelic Acid Salt by the
Continuous Process.
1. Prepare 7·mandelic acid salt solution: 4.8 kg of 7·
mandelic acid salt (13.428 mol) was dissolved in 48 L of
methanol (10 vol).
2. Equipment: A GLFR 1″ ID column with a Julabo chiller
unit was used. The reactor is 39 in. long. Pack the
column. Bottom pack, charged silica gel (20 g, 8.62 mL,
Tomislav Ljubicic − Chemical Research and Development,
Pfizer Worldwide Research and Development, Groton,
Connecticut 06340, United States
J
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
(3) Behenna, D. C.; Chen, P.; Freeman-Cook, K. D.; Hoffman, R. L.;
Jalaie, M.; Nagata, A.; Nair, S. K.; Ninkovic, S.; Ornelas, M. A.;
Palmer, C. L.; Rui, E. Y. CDK 2/4/6 inhibitors. U.S. Patent
US10,233,1882018.
Jiangping Lu − Asymchem Life Science (Tianjin) Co., Ltd.,
Tianjin 300457, P. R. China
Jason Mustakis − Chemical Research and Development, Pfizer
Worldwide Research and Development, Groton, Connecticut
06340, United States; orcid.org/0000-0002-8683-0691
Mark Maloney − Chemical Research and Development, Pfizer
Worldwide Research and Development, Groton, Connecticut
06340, United States
Joseph Rainville − Chemical Research and Development, Pfizer
Worldwide Research and Development, Groton, Connecticut
06340, United States
(4) (a) Wrobleski, S. T.; Brown, G. D.; Doweyko, L. M.; Duan, J.;
Guo, J.; Hynes, J.; Jiang, B.; Kempson, J.; Lin, S.; Lu, Z.; Spergel, S.
H.; Tokarski, J. S.; Wu, H.; Yang, B. V. Pyrrolopyridazine jak3
inhibitors and their use for the treatment of inflammatory and
autoimmune diseases. WO2012125886A12012. (b) Saeed, A.;
Vaught, G. M.; Gavardinas, K.; Matthews, D.; Green, J. E.; Losada,
P. G.; Bullock, H. A.; Calvert, N. A.; Patel, N. J.; Sweetana, S. A.;
Krishnan, V.; Henck, J. W.; Luz, J. G.; Wang, Y.; Jadhav, P. 2-Chloro-
4-[[(1R,2R)-2-hydroxy-2-methyl-cyclopentyl]amino]-3-methyl-ben-
zonitrile: A Transdermal Selective Androgen Receptor Modulator
(SARM) for Muscle Atrophy. J. Med. Chem. 2016, 59, 750−755.
(5) Azizi, N.; Saidi, M. R. Highly Chemoselective Addition of
Amines to Epoxides in Water. Org. Lett. 2005, 7, 3649−3651.
(6) Burcham, C. L.; Florence, A. J.; Johnson, M. D. Continuous
Manufacturing in Pharmaceutical Process Development and Manu-
facturing. Annu. Rev. Chem. Biomol. Eng. 2018, 9, 253−281.
(7) Brandes, B. D.; Jacobsen, E. N. Highly Enantioselective,
Catalytic Epoxidation of Trisubstituted Olefins. J. Org. Chem. 1994,
59, 4378−4380.
Xin Zhang − Asymchem Life Science (Tianjin) Co., Ltd., Tianjin
300457, P. R. China
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.oprd.0c00405
Author Contributions
The manuscript was written through the contribution of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
(8) Geigert, J.; Lee, T. D.; Dalietos, D. J.; Hirano, D. S.; Neidleman,
S. L. Epoxidation of alkenes by chloroperoxidase catalysis. Biochem.
Biophys. Res. Commun. 1986, 136, 778−782.
The authors declare no competing financial interest.
(9) (a) Wiles, C.; Hammond, M. J.; Watts, P. The development and
evaluation of a continuous flow process for the lipase-mediated
oxidation of alkenes. Beilstein J. Org. Chem. 2009, 5, No. 27.
(b) Tufvesson, P.; Adlercreutz, D.; Lundmark, S.; Manea, M.; Hatti-
Kaul, R. Production of glycidyl ethers by chemo-enzymatic
epoxidation of allyl ethers. J. Mol. Catal. B: Enzym. 2008, 54, 1−6.
(10) Herrmann, W. A.; Fischer, R. W.; Marz, D. W. Methyltriox-
orhenium as Catalyst for Olefin Oxidation. Angew. Chem., Int. Ed.
1991, 30, 1638−1641.
(11) Herrmann, W. A.; Fischer, R. W.; Scherer, W.; Rauch, M. H.
Methyltrioxorhenium(VII) as Catalyst for Epoxidations: Structure of
the Active Species and Mechanism of Catalysis. Angew. Chem., Int. Ed.
1993, 32, 1157−1160.
ACKNOWLEDGMENTS
■
The authors thank Dr. James Gage for helpful discussion and
advice on the different aspects of improving the chemistry of
the process; Dr. Baolin Wu for his technical and business
development support; and Dr. Liu Kai for the overall logistics,
technical, and excellent interaction between Asymchem and
Pfizer teams. The authors also thank Asaad Namatalla and
Jared Van Haitsma for their help in the product isolation and
analysis in the trickle bed demo run and Dr. Jared Piper, Dr.
́
Stephane Caron, and Dr. Nick Thomson for reviewing the
manuscript.
(12) Li, B.; Bader, S.; Guinness, S. M.; Ruggeri, S. G.; Hayward, C.
M.; Hoagland, S.; Lucas, J.; Li, R.; Limburg, D.; McWilliams, J. C.;
Raggon, J.; Van Alsten, J. Continuous flow aminolysis under high
temperature and pressure. J. Flow Chem. 2020, 10, 145−156.
(13) Herskowitz, M.; Smith, J. M. Continuous flow aminolysis under
high temperature and pressure. AIChE J. 1983, 29, 1−18.
(14) AL-Dahhan, M. H.; Larachi, F.; Dudukovic, M. P.; Laurent, A.
High-Pressure Trickle-Bed Reactors: A Review. Ind. Eng. Chem. Res.
1997, 36, 3292−3314.
(15) Hickman, D. A.; Holbrook, M. T.; Misttetta, S.; Rozeveld, S. J.
Successful Scale-up of an Industrial Trickle Bed Hydrogenation Using
Laboratory Reactor Data. Ind. Eng. Chem. Res. 2013, 52, 15287−
15292.
REFERENCES
■
(1) (a) Bruyere, C.; Meijer, L. Targeting cyclin-dependent kinases in
anti-neoplastic therapy. Curr. Opinion in Cell Biol. 2013, 25, 772−779.
(b) Stone, A.; Sutherland, R. L.; Musgrove, E. A. Inhibitors of Cell
Cycle Kinases: Recent Advances and Future Prospects as Cancer
Therapeutics. Crit. Rev. Oncog. 2012, 17, 175−198. (c) Węsierka-
Gądek, J.; Maurer, M.; Zulehner, N.; Komina, O. Whether to target
single or multiple CDKs for therapy? That is the question. J. Cell
Physiol. 2011, 226, 341−349. (d) Shapiro, G. I. Cyclin-dependent
kinase pathways as targets for cancer treatment. J. Clin. Oncol. 2006,
24, 1770−1783.
(2) (a) Barvian, M.; Boschelli, D. H.; Cossrow, J.; Dobrusin, E.;
FAttaey, A.; Fritsch, A.; Fry, D.; Harvey, P.; Keller, P.; Garrett, M.; La,
F.; Leopold, W.; McNamara, D.; Quin, M.; Trumpp-Kallmeyer, S.;
Toogood, P.; Wu, Z.; Zhang, E. Pyrido[2,3-d]pyrimidin-7-one
Inhibitors of Cyclin-Dependent Kinases. J. Med. Chem. 2000, 43,
4606−4616. (b) Rader, J.; Russell, M. R.; Hart, L. S.; Nakazawa, M.
S.; Belcastro, L. T.; Martinez, D.; Li, Y.; Carpenter, E. L.; Attiyeh, E.
F.; Diskin, S. J.; Kim, S.; Parasuraman, S.; Caponigro, G.; Schnepp, R.
W.; Wood, A. C.; Pawel, B.; Cole, K. A.; Maris, J. M. Dual CDK4/
CDK6 Inhibition Induces Cell-Cycle Arrest and Senescence in
Neuroblastoma. Clin Cancer Res. 2013, 19, 6173−6182. (c) Gelbert,
L. M.; Cai, S.; Lin, X.; Sanchez-Martinez, C.; del Prado, M.; Lallena,
M. J.; Torres, R.; Ajamie, R. T.; Wishart, G. N.; Flack, R. S.;
Neubauer, B. L.; Young, J.; Chan, E. M.; I, P.; Cronier, D.; Kreklau,
E.; de Dios, A. Preclinical characterization of the CDK4/6 inhibitor
LY2835219: in-vivo cell cycle-dependent/independent anti-tumor
activities alone/in combination with gemcitabine. Invest. New Drugs
2014, 32, 825−837.
(16) Cole, K. P.; Reizman, B. J.; Hess, M.; Groh, J. M.; Laurica, M.
E.; Coope, R. F.; Campbell, B. M.; Frost, M. B.; Bart, J. L.; Maloney,
T. D.; Johnson, M. D.; Mitchell, D.; Polster, C. S.; Miha, A. W.;
Boukerche, M.; Condor, E. W.; Braden, T. M.; Mitter, R. D.; Heller,
M. K.; Philips, J. L.; Howell, J. R. Small-Volume Continuous
Manufacturing of Merestinib. Part 1. Process Development and
Demonstration. Org. Process Res. Dev. 2019, 23, 858−869.
(17) Ouchi, T.; Battilocchio, C.; Hawkins, J. M.; Ley, S. V. Process
Intensification for the Continuous Flow Hydrogenation of Ethyl
Nicotinate. Org. Process Res. Dev. 2014, 18, 1560−1566.
(18) Fernadez-Puertas, E.; Robinson, A. J.; Robinson, H.;
Sathiyalingam, S.; Stubbs, H.; Edwards, L. J. Evaluation and Screening
of Spherical Pd/C for Use as a Catalyst in Pharmaceutical-Scale
Continuous Hydrogenations. Org. Process Res. Dev. 2020, 24, 2147.
K
https://dx.doi.org/10.1021/acs.oprd.0c00405
Org. Process Res. Dev. XXXX, XXX, XXX−XXX