Synthesis of Vixotrigine, a Voltage- And Use-Dependent Sodium Channel Blocker. Part 2: Development of a Late-Stage Process

Article abstract of DOI:10.1021/acs.oprd.0c00427

As vixotrigine (1) entered a later clinical phase for trigeminal neuralgia (Zakrzewska, J. M.; et al. Lancet Neurol. 2017, 16, 291-300), the development of a sustainable late-stage process was required to meet the supply needs for formulation optimization

Full text of DOI:10.1021/acs.oprd.0c00427

pubs.acs.org/OPRD
Article
Synthesis of Vixotrigine, a Voltage- and Use-Dependent Sodium
Channel Blocker. Part 2: Development of a Late-Stage Process
Robbie Chen, Vincent Couming, John Guzowski, Jr., Erwin Irdam, William F. Kiesman,
Daw-Iong Albert Kwok, Wenli Liang, Tamera Mack, Erin M. O’Brien, Suzanne M. Opalka,
Daniel Patience, Stefan Sahli, Donald G. Walker,* Frederick Osei-Yeboah, Chaozhan Gu, Xin Zhang,
̈
Markus Stockli, Thiemo Stucki, Hanspeter Matzinger, Roman Kuhn, Michael Thut, Markus Grohmann,
Benjamin Haefner, Joerg Lotz, Michael Nonnenmacher, and Paolangelo Cerea
Cite This: https://dx.doi.org/10.1021/acs.oprd.0c00427
Read Online
Metrics & More
Article Recommendations
ACCESS
ABSTRACT: As vixotrigine (1) entered a later clinical phase for trigeminal neuralgia (Zakrzewska, J. M.; et al. Lancet Neurol. 2017,
16, 291−300), the development of a sustainable late-stage process was required to meet the supply needs for formulation
optimization, phase 3 clinical trials, and registration stability batches (this is the expected commercial formulation). In this article, we
describe how the process was streamlined from the early supply route (Giblin, G.; et al. Org. Process Res. Dev. 2020, DOI: 10.1021/
acs.oprd.0c00382) and a comprehensive control strategy was established. Process improvements included improving safety and
scalability for a temperature-sensitive Grignard reaction, simplifying unit operations, removal of heterogenous conditions, and route
redesign to afford a high yielding, one-pot sequential alkylation and amidation. Improvement in the salt formation step, combined
with wet milling, resulted in improved particle properties with enhanced flow properties of the final active pharmaceutical ingredient.
The process mass intensity was improved 65% while maintaining drug substance purity at more than 99.8%. This new process has
been scaled up to generate metric ton quantities of drug substance.
KEYWORDS: flow chemistry, telescoped reactions, wet milling, manufacturing process
INTRODUCTION
PROCESS DEVELOPMENT STRATEGY
■
■
Chronic pain can make life challenging for sufferers and leave
patients with significantly disrupted lives.1 Vixotrigine (1)
(Figure 1) is a nonopioid, voltage- and use-dependent sodium
Of the two routes previously developed and implemented for the
manufacture of vixotrigine, the pilot scale process (Scheme 1)
was developed for the early phase delivery of clinical trial
materials,3 however, further improvements were necessary to
address larger-scale process considerations. These included a
reduction in cost of goods, a decrease in process mass intensity
(PMI),4 shorter cycle times, and scale increases because of a
projected high-volume demand for vixotrigine. The first step of
the synthesis was considered challenging for the high volume
needs of late-stage manufacturing given the capacity constraints
for large-scale cryogenic (−70 °C) reactions. Engineering
solutions were examined to address these challenges in batch
reactors and a flow process was also developed (step 1). An
additional critical optimization included increasing the
selectivity and throughput of a selective one-pot catalytic
imine reduction and benzyl group hydrogenolysis, with a
concomitant Boc protection of the pyrrolidine formed, to afford
intermediate 6. Optimization of step 6 was deemed necessary
Figure 1. Structure of vixotrigine.
channel blocker currently being studied for the management of
neuropathic pain conditions and has shown efficacy in a phase 2
clinical study for pain associated with trigeminal neuralgia.2 Two
manufacturing processes for vixotrigine were described in the
preceding article, an enabling kilo lab route adapted from the
medicinal chemistry synthesis and an improved pilot scale
process for early and midphase clinical supplies. To support
phase 3 studies and late-stage manufacturing, a more robust
process was required that would enable delivery of metric ton
quantities of 1. Here, we describe our studies leading to the
current manufacturing process for the synthesis of vixotrigine.
Received: September 25, 2020
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
© XXXX American Chemical Society
A

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 1. Pilot Scale Process for Vixotrigine (1)
Scheme 2. Observed Process Impurities in the Grignard Formation
because of the difficulty of removing p-TsOH and the time-
RESULTS AND DISCUSSION
■
consuming amidation reaction using methanolic ammonia. An
opportunity was seen to reorder the alkylation−deprotection−
amidation sequence, bringing two basic steps in tandem to allow
for a one-pot process saving solvent and processing time. The
HCl salt formation from the early supply route resulted in a drug
substance with a small particle size distribution (PSD)
unsuitable for the formulation process. Alternative salt-forming
conditions and wet-milling technology were chosen to engineer
the particles to meet the desired flow properties.
The Grignard reagent formation and reaction with pyrogluta-
mate 3 to generate 4 had several scale-dependent challenges,
including volume inefficiency (PMI = 32), moderate yields, and
requirement of cryogenic reaction temperatures. The clinical
timeline did not allow enough time to develop, demonstrate, and
scale a novel flow process to address these challenges, so the
batch process was optimized to supply materials quickly, and, in
parallel the development of a flow process to completely remove
the cryogenic temperature requirements was undertaken.
Grignard Reaction: Batch Processing. The pilot scale
process needed improvement in a few key areas for successful
B
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
scale up. First, the process utilized a high-temperature dry
activation protocol involving heating iodine and magnesium
turnings at 105 °C for 3 h to prepare Grignard reagent 2b at a 17
kg scale. This procedure could lead to inhomogeneous
activation and was not preferred for scale up. In addition, the
reaction quench was complex, and the workup required a
lengthy solvent concentration of the methyl tert-butyl ether
(MTBE) mixture after extraction prior to precipitation with
hexane. The Grignard addition reaction between 2b and 3 was
performed in tetrahydrofuran (THF) at −65 °C using 1.75
equiv of 1-(benzyloxy)-4-bromobenzene (2a) and after
quenching and workup, the product was isolated from
MTBE/heptane to provide 4 in 78% yield, with 98.7% w/w
purity. The major process impurity in this initial run was the
Wurtz5 homo-coupled impurity 14 (Scheme 2), observed at
0.68% w/w.
The focus of process optimization was to ensure safety,
streamline reagent choice with attention to stoichiometry, and
reduce reaction and workup volumes to maximize the reaction
throughput. The dry magnesium metal and iodine initiation
protocol was replaced with one using 1 M DIBAL-H in toluene.6
The safety of the initiation was improved, and the use of 1,2-
dibromoethane was eliminated. Furthermore, the improved
process reduced the amount of magnesium to 1.05 equiv relative
to 2b thereby reducing the quenching solvent volumes during
cleaning. Summarized below in Table 1 is a comparison of the
pilot scale and improved magnesium activation protocols.
During process exploration, Grignard-related impurities from
reactions with oxygen [10, 11, and 12 (produced from 11 in step
2)] or moisture (13) and homo-coupling (14) were observed
(Scheme 2).7 The assay value for the Grignard reagent 2b was
stable when under a nitrogen atmosphere. However, upon
ceasing the nitrogen feed and exposing the solution to air, an
immediate and sustained increase in phenol impurity 10, THF
adduct impurity 11, and des-bromo impurity 13 was observed
(Scheme 2). Therefore, stringent atmospheric controls were
developed to avoid the introduction of oxygen in both the
Grignard formation phase as well as the cooling phase prior to
the addition of 3. To ensure air-free conditions, THF was heated
under reflux in a nitrogen atmosphere for degassing prior to
charging to the reactor. Check valves were also installed on vent
lines to further minimize air exposure. Reaction of 3 with itself to
form 15, and multiple addition products as shown in Scheme 3
were another source of yield loss. To address this issue, low-
temperature conditions were used and reduced equivalents of
the Grignard reagent 2b were implemented.
The earlier direct quench of the Grignard addition reaction
with 2-PrOH at −60 °C was replaced by a simplified reverse
quench whereby THF/2-PrOH was prepared in a separate
reactor and the −60 °C reaction mixture was added keeping the
quench solution above −20 °C. This simple change used the
lower temperature of the Grignard reaction solution to control
temperature during the quench. Addition of water followed by
adjustment to pH 6−7 with 50% aqueous acetic acid followed by
a final wash of the organic phase with brine completed the
workup. After a filtration to remove the magnesium particles, the
solution was taken directly for crystallization from THF/
heptane. The elimination of the lengthy MTBE solvent swap in
favor of a simple crystallization from the partially concentrated
organic reaction solution and the direct processing of wet 4
forward into the next step streamlined the process further.
Although the average yield of the optimized Grignard reaction
sequence remained similar (74%), Wurtz impurity 14 was not
detected in 4 and the volume efficiency almost doubled (PMI
from 32 to 17).
Table 1. Comparison of Magnesium Activation Protocols
Scheme 3. Proposed Impurities in Step 1 Impacting the Mass Balance
C
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 2. Overview of the flow process flow design for Grignard addition and workup.11
Grignard Reaction: Flow Processing. Utilization of
suitable flow process equipment enables precise temperature
control by fast mixing and effective heat exchange.8 These
benefits often result in improved quality of the isolated product
and more robust manufacturing processes.9,10 Given the
projected high volume demand for vixotrigine, it was important
to avoid the large-scale Grignard initiations and limit the need
for cryogenic reactor capacity required for batch operation.
Various equipment configurations were considered for all unit
operations in the preparation of 4 (Grignard reagent formation,
Grignard addition, quench, and phase separation) except the
final crystallization, which adapted the THF/heptane protocol
from the batch process to isolate the product. A flow setup was
designed (Figure 2) utilizing:
concentration of 0.83 M and the setup was run for 18 h to
provide 10 kg of the Grignard solution to process forward in the
addition step.
For reaction of 2b with pyroglutamate 3, the flow setup
consisted of a plug flow reactor to allow efficient mixing of two
feed streams and a mixer/heat exchanger. Three reagent streams
(2b, 3, and quench solution) were charged from stainless steel
vessels and regulated by mass flow metering valves to achieve the
desired stoichiometry (1.25 equiv 2b/1.0 equiv 3 for the
reaction and 1.0 equiv 1 M sulfuric acid for the quench). The
pyroglutamate 3 solution was precooled by passing through a
heat exchanger to reach a temperature of −8 °C. The
pyroglutamate 3 and 2b were mixed and passed directly into a
mixer/heat exchanger that efficiently removed heat generated
during the reaction. The entire residence time was under 8 s
before quenching with 1 M sulfuric acid.
The concentration and temperature of the quench were
chosen to completely dissolve the magnesium sulfate instanta-
neously upon mixing. After the sulfuric acid quench, phase
separation was achieved with centrifugal extractors followed by a
brine wash. Direct collection of the THF product stream into a
reactor and crystallization of 4 after the addition of heptane
antisolvent gave a wet cake of intermediate 4 that after drying
produced a total of 1.77 kg (84%) based on 3. Summarized
below in Table 2 is a comparison of the improved batch and flow
process results for producing ketone 4.
• a continuous stirred tank reactor (CSTR) for Grignard
formation
• a plug flow reactor for Grignard addition and quench
• centrifugal extractors for phase separations and washes
CSTR strategies12 for the formation of Grignard reagents
have been published previously. For this project, a setup was
established to specifically suit the conversion of 2a to 2b. The
main challenges are related to charging magnesium and reaction
monitoring. The key challenge for the addition of Grignard
reagent 2b to 3 in flow was implementing a reactor design that
avoided clogging and incorporated the quench. The final
equipment train featured a two tank CSTR setup for Grignard
formation. In the first tank, the Grignard formation occurred.
For this, a 500 mL vessel was fitted with:
The PMI for the flow process was comparable to the batch
process; however, flow processing avoids cryogenic conditions
and affords flexibility in scaling. Work is currently ongoing to
scale up the flow process using manufacturing equipment.
• a solid addition device for magnesium addition
• an inlet for delivery of 2a
• a sampling port
• a product removal tube for transfer of the contents to a
second tank to settle solids before further transfer
Table 2. Comparison of the Improved Batch and Flow
Process Results for Producing 4
Capitalizing on learnings from the batch process, the entire
reaction was kept under an inert atmosphere (Scheme 2). The
advantage of a CSTR setup was that fresh 2a and magnesium
could be charged continuously to an already activated tank of the
Grignard reagent, obviating the need for initiation with each
charge of fresh reactants. Initiation was confirmed by measure-
ment of an exotherm and observation of a change in color. Once
observed, the starting feeds of bromide 2a in THF and
magnesium were added at such a rate that allowed for a 60
min residence time for Grignard formation. The setup was
designed in a way to allow transfer of the product solution via a
settling tank to storage without displacement of magnesium
particles. This feed solution of 2b in THF was prepared at a
parameter
scale (kg 3)
reaction temperature
isolated 4 (kg)
% yield
PMI
purity
residual 14 (Wurtz)
improved batch
284.9
flow
1.20
∼10 °C
1.77
84
a
b
<−50 °C
c
375
74
c
17
97.0
ND
15
98.6
ND
d
e
f
f
a
Internal reaction temperature maintained during the addition of 2b
b
c
to 3. Reaction temperature measured before the quench. Based on
d
wet cake loss on drying (LOD). Wet cake assay (% w/w vs a
characterized reference standard). Assay (% w/w). Not detected.
e
f
D
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Imine Formation. Acid-promoted deprotection of 4 and
subsequent cyclization gave imine 5 in a single step and although
high yielding for pilot scale, areas for further development
included enhanced impurity control, use of low cost alternative
acids, and optimization of the workup volumes. Significant
excesses of TFA (8.9 equiv) and the tedious neutralizing quench
workup contributed to lengthy unit operations and generation of
unnecessary waste. The major impurities observed in isolated
imine 5 were the hydrolysis product 16 (≤0.50%) and
enantiomer 17 (≤0.16%) (Figure 3).
was least pronounced at pH 7.5 as both lower and higher pH
values (6.5 and 8.5) resulted in more rapid rates of
epimerization. Higher hold temperatures enhanced epimeriza-
tion rates at all pH ranges, particularly at temperatures above 25
°C. It is likely that protonation of the imine nitrogen in 5
enhances the acidity of the ester α-proton on the stereogenic
center, promoting increased epimerization to 17. With the
results of these experiments in hand, the wet cake of 4 was first
washed with ACN, and the reaction temperature was kept at <26
°C to suppress the formation of acid 16. A two-stage addition of
quench solution was used, first with a rapid addition to achieve
pH 3 (to ensure control of acid 16) followed by a slower charge
to the target pH of 7.5 (to avoid epimerization to enantiomer
17). Finally, a water wash of the wet cake was added to ensure
purging of acid 16 to the mother liquor.
Replacement of aqueous potassium bicarbonate with
ammonium hydroxide cuts workup volumes by 85% and
eliminated off-gassing. A cooling crystallization of 5 was
performed in ACN/2-PrOH (5:1 v/v) using water as the
antisolvent (2 V), with seeding. The implemented process
changes afforded 475 kg batches of crystalline 5 in >85% yield,
>99.5% purity, and >99.5% stereochemical purity [high-
performance liquid chromatography (HPLC) area %]. The
step 2 PMI was lowered from 35 to 20.
Synchronized Imine Reduction, Amine Protection,
and Debenzylation. The conversion of imine 5 to phenol 6 in
the early pilot route involved three sequential chemical
reactions, as shown in Scheme 4: reduction of imine 5 to give
pyrrolidine 18, followed by in situ Boc protection of the
pyrrolidine nitrogen to produce 19 and then a reductive cleavage
of the benzyl ether to afford phenol 6.
Impurities observed during process optimization and scale up,
listed in their order of decreasing importance, are shown in
Figure 5 and include the pyrrolidine ring-opened impurity 20,
the impurity 21 arising from the reaction of the phenolic OH
group in 6 with Boc-anhydride and the trans-epimer 22. Other
impurities of concern included the phenol 23, arising from
reductive cleavage of the benzyl group in 18 prior to the
formation of the N-Boc intermediate 19, the amino ester 24,
from ring-opening reduction of the C−N benzylic bond in 23,
and the phenol 25, from reductive cleavage of the benzyl group
in the imine 5 prior to the reduction of the imine function itself.
When the hydrogenation was performed in the absence of
Figure 3. Major process impurities for the synthesis of 5.
Solvent screening identified acetonitrile (ACN) as an optimal
toluene replacement for imine formation, allowing for increased
volume efficiency [3 V (L/kg starting material) vs 6.5 V
respectively]. Additionally, any low level Wurtz impurity 14
carried over from the previous step was insoluble in ACN and
could be removed easily by filtration. Methanesulfonic acid
readily replaced TFA and the conversion to imine 5 was
accomplished with only 2.9 equiv, which reduced the quench
volume and cycle time.
Attention was then turned to focus on the parameters
controlling levels of impurities 16 and 17. Formation of the acid
impurity 16 increased if the reaction temperature was >30 °C, if
the water content within the reaction was high, or if the apparent
pH was held at <1 during the quench procedure. Specifically,
when the reaction was run at 35 °C, an increase of 16 to 3.4 area
% was observed. The addition of adventitious water (1.0 equiv)
to the reaction produced 2.1 area % of 16. Optimization
experiments on the workup found that when 25% of the quench
solution (4.6 N NH4OH) was added (apparent pH < 1) and the
reaction was held at 15 °C for 12 h, acid 16 was observed at 13.3
area %. However, when 40% of the quench solution was added
(apparent pH 3) and the reaction was held at 15 °C, the level of
16 observed was only 0.26 area %.
A hold time study was conducted to determine the impact of
pH and temperature on epimerization (Figure 4). Racemization
Figure 4. Growth of epimerization product 17 as a function of pH and temperature.
E
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 4. Transformations in the Conversion of 5 to 6 in Step 3
Figure 5. Impurities formed in step 3.
a
Table 3. Selected Catalyst Screening Results (2 h)
entry
catalyst
JM catalog no.
5 (starting material)
22 (trans)
20 (ring-opened)
19 (OBn ether)
6 (product)
b
1
2
3
4
5% Pd/C type 487 (7.5 wt %)
20% Pd(OH)2/C (1 wt %)
5% Pt/C (1 wt %)
113045
ND
ND
6.3
0.17
0.19
ND
ND
0.56
0.31
ND
ND
68.9
78.0
93.6
14.5
30.3
21.7
0.09
0.18
A401002-20
B103032-5
5R58
5% Pd/C (1 wt %)
85.3
a
b
Area % purity by HPLC. Not detected.
Boc2O, impurities 23 (65 area %) and 24 (34 area %) are the
major observed species; when the anhydride was not present in
excess, these impurities were observed at higher levels.
After initial evaluation of the pilot scale conditions, the
following parameters were targeted for additional studies:
reducing the Boc2O charge, lowering the catalyst charge below
7.5 wt % or, changing the catalyst, decreasing the reaction time
and eliminating multiple distillations in the solvent swap from
MeOH to EtOAc/heptane for product isolation.
Small-scale catalyst-screening experiments were performed
with a variety of commercially available Pd/C and Pt/C catalysts
from Johnson Matthey13 to find a more active catalyst than 5%
Pd/C type 487 paste and to address both the quantity of the
catalyst charge and the long reaction time required for step 3
(Table 3).
Pt-based catalysts did not debenzylate well and produced 19
as the major reaction product and, therefore, these catalysts were
not pursued further. Of the Pd/C catalysts screened, the use of
Pearlman’s catalyst [20 wt % Pd(OH)2/C] at a loading of 1 wt %
(entry 2) gave encouraging results as compared with the use of
5% Pd/C type 487 at a loading of 7.5 wt % (entry 1). Further
optimization studies indicated that the use of 3 wt % Pearlman’s
catalyst resulted in consistent completion of the reaction within
2−5 h. Pearlman’s catalyst appeared to produce a more active
Pd(0) catalytic species in situ in comparison to the use of the 5%
Pd/C type 487 catalyst.14 In addition, Pearlman’s catalyst
increased the rate of the reduction reaction in step 3 and was not
susceptible to catalyst deactivation in alcoholic solutions.15 After
that, catalyst selection attention was turned to optimizing the
charge of Boc-anhydride. Ultimately, 1.01 equiv of Boc-
anhydride afforded the best impurity profile for reactions that
went to completion (≤0.15 area % 19 with respect to 6). The
effect of added acetic acid on the impurity profile was briefly
studied with the tepid results. For example, in one reaction
employing 0.6 equiv, a reduced amount of 20 was observed (2.6
vs 3.6 area %); however, at the expense of more 23 (2.0 vs 0.68
area %); levels of 22 in each reaction were similar (0.31 vs 0.39
area %). In general, temperature and the agitation rate had little
impact on the conversion rate to product 6 and product impurity
profiles at reaction completion (reactions on 50−75 g scale, at
25 °C, 750 rpm; 35 °C, 500 rpm; and 35 °C, 750 rpm all afforded
6 in >93 area %).16
The pilot procedure for isolation of 6 required three
distillations to solvent swap from MeOH to EtOAc, after
which a charge of n-heptane induced crystallization. Crystal-
lization of phenol 6 directly from MeOH was investigated and
gave purified 6 in 90−92% yield from 1:1 (v/v) MeOH/water (4
V). In all experiments, phenol impurities 20 and 22 and Boc-
phenol impurity 21 were completely purged to the mother
liquor.
The optimization of step 3 resulted in the replacement of the
hydrogenation catalyst 5% Pd/C type 487 with Pearlman’s
catalyst at a lower loading (3 vs 7.5 wt %) and an improved direct
final product crystallization from MeOH/water. In the initial 30
kg scale-up run of optimized step 3 conditions, a strong
exotherm from 25 to 35 °C was observed within the first 10 min.
F
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
An unusually high level of C−N hydrogenolysis impurity 24 was
observed (10.4 area %) at 1.5 h and this contributed to a low
yield (78%). In a second 30 kg run, the reaction temperature was
more carefully controlled (22−27 °C) and the charge of Boc-
anhydride was slightly increased to 1.03 equiv. In this run, the
expected yield of phenol 6 was obtained. The results for these
initial 30 kg scale-up runs are summarized below in Table 4.
manufacturing operations could be greatly improved using a
telescoping approach (Scheme 5).19
In the pilot scale process, K2CO3 was used as the base for the
alkylation. Therefore, the possibility of using K2CO3 as the base
for the step 4 alkylation and a subsequent amidation was
explored. This idea was tested and proven to be feasible. To
compound 6 and K2CO3 in ACN (5 V) was added 1.05 equiv of
bromide 7. The heterogeneous mixture was heated at 80 °C until
compound 6 was ≤2 area % (typically 16 h to 2 d). The batch
was cooled to 45 °C and filtered, washed with ACN (2 V), and
the filtrate was carried to the amidation step. To the filtered
batch, 2.0 equiv of K2CO3 and formamide (2 V; 20 equiv) were
added and the mixture was heated to 80 °C until compound 8
was ≤2 area % (typically 16−30 h). To the batch was added
water (11 V) at 75−80 °C, followed by cooling to room
temperature over 8 h, filtration, washing with water, and drying
at 55 °C. To shorten the reaction time, the reaction
temperatures for both the alkylation and amidation were raised
to 90 °C. This process was scaled to 110 kg (6 batches). The
average impurity levels observed (Figure 6) for 8 (intermedi-
ate), 27, and 28 were 1.6, 2.1, and 2.2 area %, respectively, before
isolation. After isolation, 26 was obtained in ∼90% molar yield
and ∼96 area % purity; observed levels for 8, 27, and 28 were
0.27, 1.2, and <0.05 area %, respectively.
Table 4. Summary of the Results for 30 kg Scale Up Runs
Producing Phenol 6
Boc2O
(equiv)
residual Pd
(ppm)
a
a
entry
% yield assay purity residual 21
1
2
1.01
1.03
78
89
99.5
100.1
99.7
99.8
0.20
0.16
<10
10
a
Area % by HPLC.
While previous scale ups were performed in headspace
reactors, the commercial manufacturer chosen for further scale
up uses BCT Loop reactors for improved heat removal and mass
transfer efficiencies.17,18 Because previous studies were
performed at pressures of ≤4 bar, additional experiments were
evaluated at pressures of 10−40 bar.
At an increased hydrogen pressure of 10−20 bar, a lower level
of impurity 20 was observed, while at ≥25 bar, significant
amounts of another impurity, 19 (3−5 area %), were observed,
indicating a change in the relative rates of reaction of the desired
benzyl group cleavage and the C−N ring opening hydro-
genolysis. Based on these results, an operating pressure range of
14−16 bar was selected for use.
To eliminate melting of Boc-anhydride prior to its charging,
and stability concerns associated with the melted reagent, the
use of a stable, 90 wt % solution in THF was adopted.
After reaction completion, the reactor was vented and its
contents were heated to 40−50 °C and filtered. A hold time of
the reaction slurry at 3−7 °C and a reduction of the water
content improved the impurity profile of the crystallized product
by keeping impurity 20 in the solution, promoting efficient
purging. In the plant, step 3 was run at a 320 kg scale twice daily
and reaction solutions were combined for isolation of phenol 6
as a single batch. Table 5 below summarizes the results for a
completed registration stability campaign. The step 3 PMI was
also reduced from 20 to 9.
Alkylation and Amidation. Step 4 in the pilot scale process
converted phenol 6 to benzyl ether 8 using 2-fluorobenzyl
bromide (7) in the presence of K2CO3 in refluxing acetone.3 The
reaction time for this step typically exceeded 50 h. The desired
product was crystallized after workup and repetitive codistilla-
tions from heptane, which resulted in high solvent usage. In the
three chemical transformations for steps 4, 5, and 6, the
alkylation, Boc-deprotection, and amidation sequence of
reactions start under basic alkylation conditions followed by
acidic Boc deprotection and then back to basic amidation
conditions. If the two basic reactions were sequential using the
same solvent and base, the PMI and efficiency of the
The drawbacks of this alternative process are: (1) two
separate charges of potassium carbonate (1.5 equiv for the
alkylation and 2.0 equiv for the amidation) and a filtration
between reactions were needed; (2) because of the hetero-
geneity of the reaction mixture, this one-pot process did not
scale up consistently in terms of the reaction time for both the
alkylation (6−17 h) and the amidation (12−20 h).
In the interest of using a soluble common base for sequential
alkylation and amidation steps, it was found that commercially
available 25% NaOMe in MeOH was a convenient and
inexpensive reagent. An alkylation condition was developed
using 1.06 equiv NaOMe/1.10 equiv 7 in dimethylformamide
(DMF, 3 V) at 25 °C that reached completion in 1−2 h (vs >16 h
at 80 °C for the K2CO3 process). ACN was replaced with DMF
as this enhanced both reaction kinetics and volumetric
efficiency. Following the alkylation step, the telescoped
amidation reaction was initiated by adding to the alkylation
mixture, 24 equiv of formamide and 1.8 equiv of NaOMe/
MeOH at 25 °C.20 Formamide was added prior to the addition
of NaOMe/MeOH to avoid epimerization. The reaction went to
completion (<1.0 area % of 8) in 4 h at 25 °C and similar to the
pilot scale process, excess bromide 7 from the alkylation was
consumed by N-alkylation of formamide. It was found that there
was a clear influence of the equivalents of formamide on the level
of all three major impurities, that is, 8, 27, and 28 (Figure 7).
The level of acid impurity 28 was 2.3 area % with 16 equiv of
formamide versus 1.4 area % using 24 equiv. There was a more
pronounced effect of the equivalents of formamide on the
amount of epimerization impurity 27 (Figure 7). As shown,
when 24 equiv of formamide was used, there was significantly
less epimerization, likely attributed to a lower basicity.
Table 5. Summary of the Scale-Up Results for Step 3
a
a
phenol 6/batch
575−606 kg
% yield
purity
assay
impurities
residual Pd
<5 ppm
86−92%
≥99.8
≥99.6 wt %
≤0.19% 20
≤0.08% 21
a
Area % by HPLC.
G
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 5. Alternative Reaction Sequence to 1 Free Base via Boc-Protected Amide 26
Figure 6. Major impurities observed in telescoped alkylation and ester amidation.
Figure 7. Dependence of the formation of epimerization impurity 27 on reaction time with varying amounts of formamide.
Additionally, when 8 equiv of formamide was used, unreacted 8
did not decrease below 2.7 area % after 17 h. However, when 24
equiv of formamide was used, 8 dropped to <0.7 area % in 3 h.
In summary, compared to the pilot process, the NaOMe one-
pot process (Scheme 6) greatly improved the yield of 26 from 90
to 96% and the assay from 96.4 to >99.8%. In terms of
production efficiency, the optimized process resulted in a >50%
reduction in the production time and a 2-fold decrease in the
volume.
Deprotection. The pilot process used p-TsOH/MeOH to
cleave the Boc group of 8 to provide 9 as its p-TsOH salt
(Scheme 1). Although the yield and purity of salt 9 was high, the
H
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 6. Telescoped Alkylation−Amidation One-Pot Process
Scheme 7. Deprotection of 26 Using MsOH
process required six extractions (using NH4OH, Na2CO3, and
brine) to remove p-TsOH from the product stream, followed by
two distillations to replace MTBE with MeOH. In addition,
these conditions required development of an analytical method
for quantification of residual methyl p-toluenesulfonate, a
potential mutagenic impurity, in 9. To replace p-TsOH, HCl
was considered a favorable choice for the potential benefits of
easy removal, low cost, and HCl salt formation at the same time.
In one scaled up batch (160 kg 26), HCl/2-PrOH was used to
cleave the Boc group and form HCl salt 1 in 98% yield. In terms
of quality, this process did not purge process impurities and gave
1 with nonoptimal physical properties. A recrystallization to
purge impurities and control physical properties was performed,
but the preference was to control impurities during free base
formation and physical properties during the final salt formation.
Therefore, this one-pot process condition was not further
pursued.
properties, preblending and roller compaction were required to
achieve a blend uniformity suitable for the tableting process.
Improvement of the drug substance bulk solid properties would
allow for a simpler direct compression (DC) process that would
eliminate the need for roller compaction.
It was evident that a desolvation process needed to be avoided
to achieve a larger particle size, so 2-PrOH was examined as a
replacement for ethanol. The salt formation carried out in 2-
PrOH produced the anhydrous HCl salt 1 that did not proceed
through a solvate, however, the PSD was still too small to meet
formulation needs. Interestingly, it was discovered that replacing
gaseous HCl with 6 N aqueous HCl in 2-PrOH produced
substantially larger particles as shown in the micrographs in
Figure 8.
After optimization for both yield and purity, 1.3 equiv of
MsOH in ACN (3 V) at 35−45 °C was selected for Boc cleavage
to 1 (Scheme 7). Precipitation of 9 as its MsOH salt was
observed during the reaction; however, to streamline the
process, the MsOH salt was neutralized, in situ, at 50 °C by
charging 1.0 N NH4OH solution (5 V). As the reaction
generated 1 equiv each of CO2 and isobutylene, an off-gassing
safety study was conducted prior to scale up. To control the off-
gassing for large-scale production, MsOH was charged slowly
over 30 min at 25 °C and the reaction was heated up stagewise to
45 °C. Nucleation occurred toward the end of the NH4OH
addition, after which the mixture was then cooled to 0 °C,
filtered, washed with water, and dried to give vixotrigine-free
base in 96% molar yield with >99.8 area % purity. These
conditions were applied to 700 kg scale batches.
Final Salt Formation and Particle Engineering. The
pilot process salt formation step (Scheme 1) produced a
metastable ethanol solvate of the HCl salt 1. Although
transformation into the desired anhydrous form was observed
after drying of the solid, the temperature-driven desolvation
process was time consuming and led to shattering of the crystals
producing small particles and a wide PSD. These fines did not
flow well during the tableting process. To address the poor flow
Figure 8. PLM images showing differences of 1 produced by HCl gas in
2-PrOH and by aqueous HCl in 2-PrOH (approximately 12% water).
Drug substance particles within a target d50 range of 80−110
μm flowed better and could eliminate the need for roller
compaction during drug product manufacturing. Because of low
drug substance solubility in 2-PrOH at ambient temperature, the
salt formation was performed by direct precipitation of the HCl
salt at a higher temperature after dissolving the free base of 1.
The desired particle size in the precipitation process was
achieved by controlling the agitation and aqueous acid (1.0
I
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
equiv) addition rates. The results from plant campaigns and a
comparison with development study lots are summarized in
Table 6.
a
Table 6. Precipitation Process Scale Up Comparison
500 L
reactor
1560 L
reactor
1 L reactor
batch size (kg)
0.1
45
60
ratio of impeller to tank diameter
aq HCl addition time (min)
max agitation speed (rpm)
0.60
2−60
1000
0.56 0.64
b
13.5
100
30
90
max power supplied by the agitation per
unit mass slurry (W/kg)
8
0.51 1.47
b
agitation (rpm)
350−800
100
90
c
product PSD (μm)
d
d
d10
d50
d90
13−19
60−92
165−215
22
108
252
11, 13
86, 96
239,
d
257
a
HCl salt formation was performed by adding 1 equiv of 6 N aq HCl
to a free base solution in 6 volumes of 2-PrOH at 65 °C, cooling to
0−5 °C, and filtering and washing the filter cake with 1.5 volumes of
b
c
2-PrOH. Range used in development studies. Listed values are
d
volume-based percentiles. Batch 1 and batch 2.
The aq HCl/2-PrOH process did indeed give larger particles;
however, the powder blend resulting from the new drug
substance was prone to powder consolidation and poor flow
when held for ordinary staging periods prior to compression
and, therefore, required immediate compression after blending.
Further examination of the powder suggested that the increased
cohesiveness of the active pharmaceutical ingredient might be
because of the presence of large amounts of fines (Figure 10B).
Despite the ability of the alternate precipitation method to
produce larger particles, fines were still present, the proportion
of which varied from batch to batch. Consistency in PSD and the
proportion of fines within the bulk solid were challenging to
reproduce in larger vessels with varying impeller and reactor
geometries. In order to address the lack of consistency in the
particle size, a wet-milling process with temperature cycling was
developed in which the product slurry was recirculated through
a high-shear mixer, consisting of rotor and stator screens with
interchangeable configurations, as depicted in Figure 9. The
completion of milling was controlled by in-line particle
monitoring (focused beam reflectance measurement), the
number of volume turnovers, or a combination of both.
In practice, the salt formation would produce large crystals,
wet-milling would follow by circulation of the slurry through a
high shear mixer, and then thermal cycling of the wet milled
slurry to remove fines prior to the final product isolation.
Conditions for each of these stages were selected for
development studies, which evaluated the isolated material’s
bulk and flow properties and tableting feasibility (Figure 10).
The selected final drug substance conditions were demon-
strated successfully in large-scale campaigns and produced
batches ranging from 120 to 584 kg that met target PSD values
(Table 7) for drug product formulation and other key bulk solid-
state attributes. This allowed implementation of a DC tableting
process for drug product formulation.
Figure 9. Wet mill configuration.
The work described above resulted in a significant improvement
in the scalability and robustness of the process and provided a
drug substance with preferable solid-state attributes necessary
for a DC tableting process. The improvements made for each
step of the process are as follows:
• the Grignard reaction stoichiometry was optimized,
reducing the charge of aryl bromide 2a by 17%. Enhanced
air-free conditions ensured impurity control while
switching from a direct quench to a reverse quench of
the reaction mixture, using THF/2-PrOH, allowed for the
most efficient utilization of reactors. The flow process
described was developed to both enhance process safety
for the Grignard formation and remove the reliance on
cryogenic reactors during the Grignard addition step.
• for the imine formation, the acid and solvent were
changed from TFA/toluene to MsOH/ACN, allowing a
purge of any residual impurity 14 via filtration. The acid
stoichiometry was reduced to one-third of the original
charge, simplifying the neutralization once the reaction
endpoint was reached. The reaction quench was
optimized by the determination of an optimal pH of its
endpoint and ability to be held (7.5), conditions that were
most resistant to epimerization of C-2 of the pyrrolidine
moiety. Aqueous potassium carbonate (10 V) was
replaced by ammonium hydroxide (1.5 V) as the base
for the neutralization, effecting a significant improvement
in volume efficiency (2.5-fold). The number of unit
operations was also significantly reduced.
Scheme 8 depicts the late-stage process developed to support
the production of vixotrigine (1) on a manufacturing scale.
Summary of Process Improvements to Enable
Manufacturing of Vixotrigine at a Late-Stage Scale.
• the hydrogenation catalyst previously used, 5% Pd/C type
487 paste, was replaced with Pearlman’s catalyst [20%
Pd(OH)2/C], based on the latter’s higher activity and
selectivity for the target reactionhydrogenolysis of the
J
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 10. Scanning electron microscopy images of 1 produced from three different processing methods. (A) EtOH/HCl gas typical d50: 15−20 μm.
(B) 2-PrOH/aq HCl precipitation typical d50: 70−110 μm; note the presence of fines around the larger particles. (C) 2-PrOH/aq HCl crystallization
with wet milling and temperature cycling. Typical d50: 80−125 μm; note the absence of fines.
Table 7. Average PSD Obtained from Large-Scale
Manufacturing Campaigns
by replacement with MsOH for Boc deprotection, which
had a dramatic impact on the workup efficiency. The new
three reaction sequence (two processing steps) reduced
the overall PMI from 106 to 32. In addition, the
production time for the combined three steps was
reduced 3-fold. Both process steps 4 and 5 gave 96%
molar yield with >99.8 area % purity at a 600 kg scale.
a
a
a
batch size (kg)
d10 (μm)
d50 (μm)
d
90 (μm)
campaign I
campaign II
120−174
473−584
32
30
102
108
217
256
a
Listed value is a volume-based percentile.
benzyl group. Catalyst loading was reduced from 7.5 to 3
wt %. A BCT Loop reactor was used to optimize mixing
and heat exchange efficiencies. An optimal operating H2
pressure of 15 bar was identified, which provided the best
hydrogenolysis selectivity (C−O vs C−N). Intermediate
6 was directly isolated from the MeOH reaction solvent
after 1.5-fold dilution with water.
• a new HCl salt formation condition was discovered for
producing a consistently large PSD material. A wet-
milling-temperature cycling operation was added to the
process to remove fines and achieve a PSD with desirable
flow properties for formulation.
Overall, drug substance batches, on a scale of up to ∼600 kg,
were able to be reproducibly manufactured, in 54% overall yield,
as compared with 43% for pilot scale route 2. Table 8 shows the
PMI and yield comparison of the pilot route versus the late-stage
route.
• for the alkylation and amidation steps, the sequence of the
transformations was changed, so that the two base-
mediated reactions, formation of 8 then 26, were
consecutive and could be telescoped to an efficient one-
pot process. Problematic removal of p-TsOH was solved
Scheme 8. Late-Stage Process for Vixotrigine (1)
K
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Table 8. PMI and Yield Comparison of the Pilot Route vs the Late-Stage Route
pilot route (cf. Scheme 1)
late-stage route (cf. Scheme 8)
step
reaction
% yield
PMI
step
reaction
% yield
PMI
1
2
3
4
5
6
7
2b + 3 → 4
4 → 5
5 → 6
6 + 7 → 8
8 → 9
9 → 1 FB
1 FB → 1 HCl
78
90
87
88
92
94
96
32
35
20
39
16
51
20
1 (batch)
2
3
4
5
6
2b + 3 → 4
4 → 5
5 → 6
6 + 7 → 26
26 → 1 FB
1 FB → 1 HCl
2b + 3 → 4
74
88
92
96
96
92
84
17
20
9
15
17
9
1 (flow)
15
was added the Grignard solution from reactor B while
maintaining a reaction temperature of <−50 °C. Reactor B
was rinsed with THF (22 kg) into reactor C and the reaction was
aged at <−50 °C for about 1 h. The progress of the reaction was
monitored (HPLC). Upon completion, the reaction was
quenched by transfer of the reactor C solution into a cold
(−20 to 0 °C) solution of 2-PrOH (224 kg) and THF (235 kg)
in reactor D while maintaining the temperature at −20 to 0 °C.
Reactor C was rinsed with THF (53 kg) into reactor D and the
mixture was aged at −20 to 0 °C. Water (712 kg) was added
while maintaining an internal temperature of <20 °C. The pH
was adjusted to 6.5 by the addition of the aq AcOH solution
(50%, ∼170 kg) keeping the temperature below 20 °C. The
temperature was adjusted to 20−30 °C, the mixture stirred for
20−30 min, and the phases separated (55−65 min). The water
phase was discarded and the upper organic layer was treated with
NaCl (42 kg) and water (255 kg). The mixture was stirred for
55−65 min and the phases were separated (55−65 min); the
water phase was discarded. THF (125 kg) was added to the
upper organic phase and the temperature was set to 25 °C. The
mixture was then filtered in circulation over a cartridge filter to
remove residual magnesium turnings. The filtrate was
concentrated under reduced pressure to a residual weight of
1510−1710 kg. The temperature was adjusted to 35−45 °C and
n-heptane (994 kg) was added and the mixture was stirred for
1−2 h at 35−45 °C. The mixture was then cooled to 15−25 °C
over at least 2 h and then to 0−5 °C and stirred for 3−5 h. The
solids were isolated by centrifugation and washed with n-
heptane (291 kg) and then ACN (177 kg) to give 389 kg of the
wet product. Based on loss on drying measurements, 375 kg
(74%) of the title compound was obtained.3
CONCLUSIONS
■
The pilot scale process, route 2, was optimized to make it
amenable to late-stage manufacturing of vixotrigine. The work
done simplified operational aspects and improved volume
efficiency, the impurity profile, and the overall yield. Step 1 run
in a continuous mode was also demonstrated at the kg scale,
providing a foundation for further improvement of the process.
The overall PMI was reduced by 65% while maintaining drug
substance purity at >99.8 area %. Flow properties of the drug
substance were improved from the previous process by choosing
optimal salt formation and wet-milling conditions. This new
process was scaled up to generate 2.5 t of drug substance in a
registration campaign with an average yield of 90% for each
reaction step.
EXPERIMENTAL SECTION
■
General. 1H NMR data were recorded at either 300 or 400
MHz on a Bruker spectrometer. Chemical shifts are expressed as
δ (ppm) values using the residual signals of the solvents as a
reference. LC−mass spectrometry data were generated on an
Agilent Q-TOF 6530 spectrometer, operating in ES+ ionization
mode, coupled to an Agilent 1290 Series HPLC system.
Instrument control and data acquisition were mediated through
the MassHunter data acquisition and analysis software.
Methyl (S)-5-(4-(benzyloxy)phenyl)-2-((tert-
butoxycarbonyl)amino)-5-oxopentanoate (4) (Batch Proc-
ess). Reactor A was charged with THF (1199 kg) and 1-
(benzyloxy)-4-bromobenzene (2a; 450 kg). The solids were
dissolved by heating to 66 °C, and the solution was degassed by
holding at reflux for 10−15 min and cooling to 20−30 °C under
an inert atmosphere of nitrogen. Reactor B was charged with Mg
(43.6 kg) and THF (399 kg), and DIBAL-H in toluene (1 M;
6.21 kg) was added followed by a toluene (3.19 kg) line rinse,
and the mixture was heated to 60−70 °C. To this mixture was
added a part of the 1-(benzyloxy)-4-bromobenzene/THF
solution from reactor A (80−90 kg) and the resulting mixture
stirred for 1 h at 66 °C. After charging additional solution from
reactor A (30−40 kg), the initiation of the reaction was checked,
and the rest of the solution from reactor A was added to reactor
B over 3−4 h while maintaining the mixture under reflux
conditions. Reactor A was charged with THF (52 kg), the
contents adjusted to 66 °C, and held for 10−15 min, then cooled
to 20−30 °C and transferred to reactor B. The contents of
reactor B were kept for 50−70 min at 60−70 °C, then cooled to
20−25 °C. Conversion to the Grignard reagent was checked by
IPC (HPLC). Reactor C was charged with 1-(tert-butyl)2-
methyl(S)-5-oxopyrrolidine-1,2-dicarboxylate (3; 284.9 kg) and
THF (760 kg). The solids were dissolved by heating to 66 °C,
and the solution was degassed by holding at reflux for 10−15
min. The solution was cooled −60 to −70 °C. To this solution
1H NMR (400 MHz, CDCl3): δ 1.52 (s, 9H), 2.10 (m, 1H),
2.30 (m, 1H), 2.95−3.15 (m, 2H), 3.76 (s, 3H), 4.40 (m, 1H),
5.15 (s, 2H), 5.18 (br s, 1H), 7.03 (d, J = 9 Hz, 2H), 7.30−7.50
(m, 5H), 7.95 (d, J = 9 Hz, 2H). MS m/z: calcd for C24H30NO6
[M + H]+, 428.2073; found, 428.2064.
Methyl (S)-5-(4-(benzyloxy)phenyl)-2-((tert-
butoxycarbonyl)amino)-5-oxopentanoate (4) (Flow Proc-
ess). The CSTR flow setup consisted of a 1 L stirred tank for
reaction, a 1 L settling tank, and a 10 L Schlenk type collection
vessel. The stirred tank was equipped with a solid addition
device, a reflux condenser, and transfer lines for product solution
forward processing. The setup was set to a 500 mL working
volume. Step 1: a stirred tank reactor was precharged with THF
(70 mL) and magnesium (50.8 g, 5 equiv) and the mixture was
stirred at room temperature overnight. The solid addition device
was filled with magnesium. The reaction was initiated by adding
(4-(benzyloxy)phenyl)magnesium bromide 0.77 M solution in
THF (2b; 7.7 g, 5.9 mmol). The jacket temperature was
increased to 55 °C. A solution of 1-(benzyloxy)-4-bromoben-
zene (2a; 0.85 M in THF) was added to the stirred reaction
L
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
vessel. After seven min, the addition of magnesium started at an
equimolar rate. The total amount of magnesium for the entire
run was 175 g (7.18 mol) and was calculated to keep 5 equiv of
magnesium in the stirred tank reactor over the course of the run.
When the liquid level in the tank reached operating volume, the
product solution was forwarded to the settling tank while
maintaining the 500 mL filling level in the CSTR. The
approximate residence time of the solution in the jacketed
reactor was 62 min. The product was transferred into the settling
tank (unstirred), held for the residence time of 1 h, and
subsequently transferred to a final collection vessel. The entire
process was run for 18 h. Step 2 Grignard addition: the
equipment consisted of a tubular pipe reactor, heat exchanger,
and a series of centrifugal phase separators. The tubular reactor
accommodated mixing of two reagents for the conversion to
methyl (S)-5-(4-(benzyloxy)phenyl)-2-((tert-butoxycarbonyl)-
amino)-5-oxopentanoate (4) and quenching of the product
solution with an acid solution. The centrifugal phase separators
separated the product containing organic phase from the waste
aqueous phase. The reagent solutions (methyl N-Boc-
pyroglutamate 3, Grignard 2b, and sulfuric acid) were
transferred continuously at controlled flow rates from their
respective storage tank to pass through the tubular pipe reactor,
heat exchanger, and finally to the centrifugal extractors.
Reaction/quench/workup: the 0.82 M Grignard solution was
fed continuously from the storage tank at a molar rate of 1.19
equiv and simultaneously a 0.817 M methyl N-Boc-pyrogluta-
mate 3 solution stream was fed continuously through a heat
exchanger to precool it to −8 °C. The tubular reactor for the
Grignard addition reaction was attached to a heat exchange unit
with chiller fluid set at 10 °C. After passing through the reaction
zone, 1.0 M sulfuric acid was introduced at a stoichiometric ratio
to 3. The residence time of the solution from reagent
introduction to acid quench was 8 s. From sulfuric acid
introduction to phase split, the residence time was approx-
imately 80 s. The quenched mixture passed through another
heat exchanger to increase the temperature to 30 °C for the
phase split. This material was directly fed into a centrifugal
extractor to remove the aqueous component. The obtained
organic layer was subsequently mixed with a solution of brine
and sodium bicarbonate in a second centrifugal extractor. The
final product-containing organic layer was collected into a glass
bottle. The process was run for 3.7 h. Crystallization: the
product-containing organic layer was transferred to a 10 L
reactor for solvent swap to a low water content THF/heptane
solvent system by vacuum distillation. A total of 6867 mL THF
(approximately 9.5% v/v) in heptane was added to the reactor
and subsequently distilled in approximately two equal portions
while maintaining distillation under reduced pressure (approx-
imately 600−700 mbar) at temperatures within 60−65 °C to
replace the original solvent (water containing THF). The final
solution obtained (approximately 11.5 L) was cooled to 0−5 °C
with a cooling rate of 0.5 °C/min and the resulting slurry was
filtered, washed with heptane, and dried under vacuum at 60 °C
to obtain 1.765 kg (83.6%) of the title compound.
conversion was checked by IPC (HPLC). The temperature was
adjusted to 0−5 °C and the pH was adjusted to 2−3 by quick
addition of a first part of a solution prepared from water (796 kg)
and aq NH3 solution (30%, 295.4 kg) keeping the temperature
below 30 °C. Using the rest of the diluted aq NH3 solution, the
pH of the mixture was then adjusted more slowly to 7.5 (7−8)
while maintaining a reaction temperature of <25 °C. The phases
were separated, and the upper organic layer was heated to 25−
30 °C. The mixture was filtered over a cartridge filter and the line
rinsed with ACN (23.3 kg). A solution of water (909 kg) and 2-
PrOH (357 kg) was added at 20−25 °C. The solution was
cooled to 15−20 °C and seeded (3.6 kg). The slurry was cooled
to 0−5 °C in not less than 2 h and maintained at 0−5 °C for at
least 30 min. Water (2364 kg) was charged while letting the
temperature rise to a maximum of 20 °C. The mixture was
cooled to 0−5 °C and held for 30−40 min, then warmed to 15−
20 °C, and held for 30−40 min. The slurry was then cooled to
0−5 °C in not less than 1 h and maintained for at least 2 h, then
filtered. The solids were washed with a mixture of water (1062
kg) and 2-PrOH (279 kg) and dried under reduced pressure at
50−55 °C to constant weight to afford 461 kg (88%) of the title
compound.3
1H NMR (400 MHz, CDCl3): δ 2.2−2.42 (m, 2H), 2.97 (m,
1H), 3.15 (m, 1H), 3.79 (s, 3H), 4.91 (m, 1H), 5.13 (s, 2H),
7.02 (d, J = 9 Hz, 2H), 7.30−7.50 (m, 5H), 7.86 (d, J = 9 Hz,
2H). MS m/z: calcd for C19H20NO3 [M + H]+, 310.1443; found,
310.1438.
1-(tert-Butyl)2-methyl(2S,5R)-5-(4-hydroxyphenyl)-
pyrrolidine-1,2-dicarboxylate (6). A hydrogenation reactor was
charged with 20% Pd(OH)2/C (water wet; 9.6 kg), methyl (S)-
5-(4-(benzyloxy)phenyl)-3,4-dihydro-2H-pyrrole-2-carboxylate
(5; 320 kg activity), MeOH (1142 kg), water (16 kg), and di-
tert-butyldicarbonate (90% in THF, 250 kg). The reactor was
pressurized with nitrogen followed by venting (3 times). The
reactor was pressurized with hydrogen followed by venting (3
times). The reactor was pressurized with hydrogen (15 bar).
After about 1 h at 25 °C, the reactor was vented and
repressurized with hydrogen (15 bar). The progress of the
reaction was monitored for completion (HPLC). After about 2
h, the reactor was vented and its contents were heated to 40−50
°C and filtered, and the filtrate was combined with that from a
second identical run. The stream was concentrated in vacuo to
about 2800 L at about 35−40 °C and at about 240 mbar. The
contents of the reactor were cooled to 20−30 °C and the
formation of a suspension was checked by IPC. The slurry was
aged for about 1 h. Water (960 kg) was added over about 1 h.
The slurry was cooled to 3−7 °C over at least 2 h and aged for
about 3 h. The solids were isolated by filtration, washed with 1:4
(v/v) MeOH/water (1226 kg), and dried in vacuo at 50−55 °C
to a constant weight to afford 605 kg (92%) of the title
compound.3
1H NMR (400 MHz, CDCl3) (mixture of rotamers): δ 1.19,
1.43 (2s, 9H), 1.85−2.40 (series of m, 4H), 3.83 (s, 3H), 4.35−
5.50 (series of m, 3H), 6.77 (d, J = 9 Hz, 2H), 7.42 (m, 2H). MS
m/z: calcd for C17H23NNaO5 [M + Na]+, 344.1474; found,
344.1459.
tert-Butyl (2S,5R)-2-carbamoyl-5-(4-((2-fluorobenzyl)oxy)-
phenyl)pyrrolidine-1-carboxylate (26). A reactor was charged
with 1-(tert-butyl)2-methyl(2S,5R)-5-(4-hydroxyphenyl)-
pyrrolidine-1,2-dicarboxylate (6; 590 kg), anhydrous DMF
(1671 kg), and 2-fluorobenzyl bromide [7; 382 kg, 1.10 equiv
(caution! lachrymator!)]. With good stirring, approximately
349 kg (1.06 equiv; assay corrected) of 30% NaOMe solution in
Methyl (S)-5-(4-(benzyloxy)phenyl)-3,4-dihydro-2H-pyr-
role-2-carboxylate (5). A reactor was charged with methyl
(S)-5-(4-(benzyloxy)phenyl)-2-((tert-butoxycarbonyl)amino)-
5-oxopentanoate (4; 728 kg) and ACN (1715.7 kg) and the
slurry temperature was adjusted to 10−15 °C. Methanesulfonic
acid (474.4 kg) was added in not less than 3 h while maintaining
a reaction temperature of <26 °C. The reaction was maintained
at 20−25 °C for 1 h, then the mixture was cooled 0−10 °C. The
M
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
MeOH was added over 60 min while maintaining the
temperature between 20 and 30 °C. Following the charge, the
line was rinsed forward with MeOH (20 kg) and the batch was
maintained between 20 and 30 °C for at least 1 h. The progress
of the reaction was monitored for completion (HPLC). Upon
completion, formamide (2007 kg) was charged followed by a
line rinse with MeOH (20 kg). Approximately 562 kg (1.7 equiv;
assay corrected) 30% NaOMe solution in MeOH was added
over at least 45 min while maintaining a temperature of between
20 and 30 °C followed by a line rinse with MeOH (20 kg). The
contents of the reactor were maintained at about 25 °C with
agitation for about 4 h. The progress of the reaction was
monitored for completion (HPLC). Upon completion, the
batch was transferred to a second reactor and the equipment was
rinsed forward with MeOH (172 kg). Glacial acetic acid (217
kg) was added to the batch over at least 10 min while
maintaining a temperature of 20−30 °C followed by the
addition of water (590 kg). The batch was heated to 60 °C and
water (2360 kg) was added over at least 2 h with good agitation.
The batch was maintained at 60 °C with agitation for at least 1 h.
The batch was cooled to 0−3 °C over at least 3 h and aged for at
least 1 h. The solids were isolated by filtration and washed with a
mixture of MeOH (280 kg) and water (826 kg) and then with
water (1180 kg). The wet cake was dried to constant weight in
vacuo at 67 °C to afford 728 kg (95.7%) of the title compound.
1H NMR (300 MHz, CDCl3): δ 1.13−1.28 (br s, 9H), 1.93−
2.09 (m, 2H), 2.25−2.34 (m, 1H), 2.48−2.58 (m, 1H), 4.45−
4.53 (m, 1H), 4.60−4.73 (m, 1H), 5.15 (s, 2H), 5.37−5.45 (br s,
1H), 6.93 (d, J = 8.7 Hz, 2H), 7.06−7.24 (m, 5H), 7.28−7.37
(m, 1H), 7.48−7.56 (m, 1H). MS m/z: calcd for C23H28FN2O4
[M + H]+, 415.2033; found, 415.2028.
(2S,5R)-5-(4-((2-Fluorobenzyl)oxy)phenyl)pyrrolidine-2-
carboxamide (1 Free Base). A reactor was charged with tert-
butyl(2S,5R)-2-carbamoyl-5-(4-((2-fluorobenzyl)oxy)phenyl)-
pyrrolidine-1-carboxylate (26; 700 kg) and ACN (1287 kg).
With good agitation, methanesulfonic acid (208 kg, 1.28 equiv)
was added while maintaining a reaction temperature of 20−30
°C followed by ACN (109 kg). The contents of the reactor were
warmed to 40−50 °C over at least 30 min and aged for 2−3 h.
The progress of the reaction was monitored (HPLC). Upon
completion, aq 1.7% NH3 solution (approximately 693 kg) was
added over at least 15 min while maintaining a reaction
temperature of 40−50 °C. The reaction temperature was raised
to 46−55 °C and aq 1.7% NH3 solution (approximately 2772
kg) was added over at least 2 h with good stirring and the mixture
maintained within this temperature range for 1 h. The slurry was
cooled to −3 to 3 °C over at least 3 h and was aged for at least 1
h. The solids were isolated by filtration and washed with a
mixture of ACN (111 kg) and water (1259 kg) and then with
water (1400 kg). The solids were dried in vacuo at 70 °C for 12 h
to afford 507.6 kg (95.6%) of the title compound.3
the filter line were rinsed with 2-PrOH (74 kg). The contents of
reactor B were set to 65−75 °C and purified water (145 kg) was
added. The internal temperature was increased to 75 °C and
20% aq hydrochloric acid solution (approximately 274 kg, 1.00
equiv based on the titrated HCl assay) was dosed at an internal
temperature of 70−78 °C over 20−40 min. The contents were
cooled to 0−15 °C over at least 3 h. The resulting slurry was
transferred to reactor C with an installed wet-mill loop and then
line rinsed with 2-PrOH (37 kg). The slurry was milled (target
20 passes) while keeping the temperature between 0 and 15 °C.
The resulting PSD was controlled by IPC, and the slurry
transferred to reactor D at 0−15 °C. Reactor C and the milling
equipment were rinsed with 2-PrOH (371 kg) into reactor D.
The contents were heated to 60 °C within 3−4 h, held for 10
min at 60 °C then cooled to −2 to 2 °C over 3−4 h. After aging
at −2 to 2 °C for 2−3 h, the product suspension was filtered at
−2 to 2 °C over a pressure filter. The filter cake was washed
under displacement with 2-PrOH (557 kg) via the empty reactor
D. The solids were dried in vacuo at 43 °C for 3−4 h and then to
constant weight at 68 °C (8 h) to afford 485 kg (91.9%) of the
title compound.3
1H NMR (400 MHz, DMSO-d6): δ 1.95−2.20 (m, 2H), 2.30
(m, 2H), 3.35 (s, 1H), 4.30 (m, 1H), 4.61 (m, 1H), 5.18 (s, 2H),
7.10 (d, J = 9 Hz, 2H), 7.18−7.30 (m, 2H), 7.40 (m, 1H), 7.47
(d, J = 9 Hz, 2H), 7.56 (m, 1H), 7.72 (s, 1H), 8.07 (s, 1H), 10.60
(br s, 1H). MS m/z: calcd for C18H20FN2O2 [M + H]+,
315.1509; found, 315.1505.
AUTHOR INFORMATION
■
Corresponding Author
Donald G. Walker − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United
States; orcid.org/0000-0003-1271-7124;
Email: donald.walker@biogen.com
Authors
Robbie Chen − Biogen, Product and Technology Development,
Cambridge, Massachusetts 02142, United States
Vincent Couming − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
John Guzowski, Jr. − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Erwin Irdam − Biogen, Product and Technology Development,
Cambridge, Massachusetts 02142, United States
William F. Kiesman − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Daw-Iong Albert Kwok − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Wenli Liang − Biogen, Product and Technology Development,
Cambridge, Massachusetts 02142, United States
Tamera Mack − Biogen, Product and Technology Development,
Cambridge, Massachusetts 02142, United States
1H NMR (400 MHz, CDCl3): δ 1.68 (m, 1H), 2.06−2.35
(series of m, 3H), 2.51 (br s, 1H), 3.88 (dd, J = 3, 9 Hz, 1H), 4.31
(dd, J = 6, 9 Hz, 1H), 5.15 (s, 2H), 5.57 (br s, 1H), 6.99 (d, J = 9
Hz, 2H), 7.11 (m, 1H), 7.18 (m, 1H), 7.30−7.40 (m, 3H), 7.53
(m, 1H). MS m/z: calcd for C18H20FN2O2 [M + H]+, 315.1509;
found, 315.1505.
(2S,5R)-5-(4-((2-Fluorobenzyl)oxy)phenyl)pyrrolidine-2-
carboxamide hydrochloride (1). Reactor A was charged with
(2S,5R)-5-(4-((2-fluorobenzyl)oxy)phenyl)pyrrolidine-2-car-
boxamide (1 free base; 473 kg) and 2-PrOH (2339 kg). The
contents of the reactor were heated to 70−75 °C, held for 45
min and transferred to reactor B via a filter bag. Reactor A and
Erin M. O’Brien − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Suzanne M. Opalka − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Daniel Patience − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Stefan Sahli − Biogen International, Baar 6340, Switzerland
Frederick Osei-Yeboah − Biogen, Product and Technology
Development, Cambridge, Massachusetts 02142, United States
Chaozhan Gu − STA Pharmaceutical R&D Company Ltd., A
Wuxi AppTec Company, Shanghai 200131, China
N
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
(2) Zakrzewska, J. M.; Palmer, J.; Ettlin, D. A.; Obermann, M.; Giblin,
G. M. P.; Morisset, V.; Tate, S.; Gunn, K. Novel design for a phase IIa
placebo-controlled, double-blind randomized withdrawal study to
evaluate the safety and efficacy of CNV1014802 in patients with
trigeminal neuralgia. Trials 2013, 14, 402.
Xin Zhang − STA Pharmaceutical R&D Company Ltd., A
Wuxi AppTec Company, Shanghai 200131, China
̈
Markus Stockli − Dottikon Exclusive Synthesis AG, Dottikon
5605, Switzerland
Thiemo Stucki − Dottikon Exclusive Synthesis AG, Dottikon
5605, Switzerland
Hanspeter Matzinger − Dottikon Exclusive Synthesis AG,
Dottikon 5605, Switzerland
Roman Kuhn − Dottikon Exclusive Synthesis AG, Dottikon
5605, Switzerland
Michael Thut − Dottikon Exclusive Synthesis AG, Dottikon
5605, Switzerland
Markus Grohmann − Dottikon Exclusive Synthesis AG,
Dottikon 5605, Switzerland
Benjamin Haefner − Evonik Operations GmbH, Hanau
63457, Germany
Joerg Lotz − Evonik Operations GmbH, Hanau 63457,
Germany
Michael Nonnenmacher − Evonik Operations GmbH, Hanau
63457, Germany
Paolangelo Cerea − Olon S.p.A., Milan 20090, Italy
(3) Giblin, G.; Heseltine, A.; Kiesman, W.; MacPherson, D.;
Ramsden, J.; Vadali, R.; Williams, M.; Witty, D. Synthesis of
Vixotrigine, a Use-Dependent Sodium Channel Blocker. Part 1:
Development of Bulk Supply Routes to Enable Proof of Concept.
Org. Process Res. Dev. 2020, DOI: 10.1021/acs.oprd.0c00382.
(4) (a) Process mass intensity (PMI) includes all process materials
including water and work-up chemicals relative to the product.
(b) Roschangar, F.; Zhou, Y.; Constable, D. J. C.; Colberg, J.;
Dickson, D. P.; Dunn, P. J.; Eastgate, M. D.; Gallou, F.; Hayler, J. D.;
Koenig, S. G.; Kopach, M. E.; Leahy, D. K.; Mergelsberg, I.; Scholz, U.;
Smith, A. G.; Henry, M.; Mulder, J.; Brandenburg, J.; Dehli, J. R.;
Fandrick, D. R.; Fandrick, K. R.; Gnad-Badouin, F.; Zerban, G.; Groll,
K.; Anastas, P. T.; Sheldon, R. A.; Senanayake, C. H. Inspiring Process
Innovation Via an Improved Green Manufacturing Metric: iGAL. Green
Chem. 2018, 20, 2206−2211.
̈
(5) Koch, A.; Dufrois, Q.; Wirgenings, M.; Gorls, H.; Krieck, S.;
Etienne, M.; Pohnert, G.; Westerhausen, M. Direct Synthesis of Heavy
Grignard Reagents: Challenges, Limitations, and Derivatization. Chem.
Eur. J. 2018, 24, 16840−16850.
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.oprd.0c00427
(6) Tilstam, U.; Weinmann, H. Activation of Mg Metal for Safe
Formation of Grignard Reagents on Plant Scale. Org. Process Res. Dev.
2002, 6, 906−910.
Author Contributions
(7) Amaya, T.; Suzuki, R.; Hirao, T. Quinonediimines as Redox-
Active Organocatalysts for Oxidative Coupling of Aryl- and
Alkenylmagnesium Compounds Under Molecular Oxygen. Chem.
Commun. 2016, 52, 7790−7793.
This manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript. R.C., V.C., J.G., and T.M. were employees of
Biogen at the time these studies were conducted. R.C. is
currently employed by STA Pharmaceutical R&D Company
Ltd., A Wuxi AppTec Company. V.C. is currently employed by
Juno Therapeutics, Inc. T.M. is currently employed by Sigilon
Therapeutics, Inc. J.G. is currently employed by Translate Bio.
E.I., W.F.K., D.-I.A.K., W.L., E.M.O., S.M.O., D.P., S.S., D.G.W.,
and F.O.-Y. are employees of Biogen and hold stock in Biogen.
C.G. and X.Z. are employees of STA Pharmaceutical R&D
Company Ltd., A Wuxi AppTec Company that was paid by
Biogen to conduct this research. M.S., T.S., H.M., R.K., M.T.,
and M.G. are employees of Dottikon Exclusive Synthesis that
was paid by Biogen to conduct this research. M.N., J.L., and B.H.
are employees of Evonik Operations GmbH that was paid by
Biogen to conduct this research. P.C. is an employee of Olon
S.p.A. that was paid by Biogen to conduct this research.
(8) Baumann, M.; Moody, T. S.; Smyth, M.; Wharry, S. A Perspective
on Continuous Flow Chemistry in the Pharmaceutical Industry. Org.
Process Res. Dev. 2020, 24, 1802−1813.
(9) Pastre, J. C.; Browne, D. L.; Ley, S. V. Flow Chemistry Synthesis of
Natural Products. Chem. Soc. Rev. 2013, 42, 8849−8869.
(10) McWilliams, J. C.; Allian, A. D.; Opalka, S. M.; May, S. A.;
Journet, M.; Braden, T. M. The Evolving State of Continuous
Processing in Pharmaceutical API Manufacturing: A Survey of
Pharmaceutical Companies and Contract Manufacturing Organiza-
tions. Org. Process Res. Dev. 2018, 22, 1143−1166.
(11) The process is not fully continuous; 2b is kept in a storage vessel
prior to addition.
(12) Selected publications: (a) Wong, S.-W.; Changi, S. M.; Shields,
R.; Bell, W.; McGarvey, B.; Johnson, M. D.; Sun, W.-M.; Braden, T. M.;
Kopach, M. E.; Spencer, R. D.; Flanagan, G.; Murray, M. Operation
Strategy Development for Grignard Reaction in a Continuous Stirred
Tank Reactor. Org. Process Res. Dev. 2016, 20, 540−550. (b) Hosoya,
M.; Nishijima, S.; Kurose, N. Investigation into an Unexpected
Impurity: A Practical Approach to Process Development for the
Addition of Grignard Reagents to Aldehydes Using Continuous Flow
Synthesis. Org. Process Res. Dev. 2020, 24, 405−414. (c) Hughes, D. L.
Applications of Flow Chemistry Drug Development: Highlights of
Recent Patent Literature. Org. Process Res. Dev. 2018, 22, 13−20.
(d) Riva, E.; Gagliardi, S.; Martinelli, M.; Passarella, D.; Vigo, D.;
Rencurosi, A. Reaction of Grignard Reagents with Carbonyl
Compounds under Continuous Flow Conditions. Tetrahedron 2010,
66, 3242−3247.
(13) For a listing of Johnson Matthey catalysts available for the
pharmaceutical market, see: Matthey.com Catalysts For The
Pharmaceutical Market|Johnson Matthey. 2020, available online at
https://matthey.com/en/products-and-services/pharmaceutical-and-
medical/catalysts?q=.
(14) Itaya, T.; Morisue, M.; Takeda, M.; Kumazawa, Y. Synthesis of 7-
Methyl-3-β-D-ribofuranosylwye, the Putative Structure for the Hyper-
modified Nucleoside Isolated from Archaebacterial Transfer Ribonu-
cleic Acids. Chem. Pharm. Bull. 1990, 38, 2656−2661.
(15) Walker, D. G. Private conversation with Johnson Matthey
precious metal catalyst personnel.
Funding
This work was funded by Biogen Inc. and by Convergence
Pharmaceuticals Ltd., which has since been acquired by Biogen
Inc.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors wish to sincerely thank Dr. Stuart G. Levy of SGL
Chemistry Consulting, LLC for technical assistance in editing
this manuscript, a service for which he was financially
compensated by Biogen. The authors also wish to thank Dr.
Lin Lin of Biogen for the acquisition of supporting analytical
data.
REFERENCES
■
(1) Obermann, M. Recent advances in understanding/managing
trigeminal neuralgia. F1000Res. 2019, 8, 505.
O
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
(16) Reactions were run in a 1 L Buchi hydrogenator at 5 V.
Conversion was comparable at 1.5 and 4 h for the 35 °C, 500 rpm run.
(17) For an animated presentation of a BCT Loop reactor’s operation,
see: Pratteln-Basel, K. HydrogenationBuss Chemtech AG. 2020,
available online at https://www.buss-ct.com/buss_loop_reactor.html.
(18) For a white paper describing the operating features of a BCT
Loop reactor, see: buss-ct.com 2020, available online at https://www.
buss-ct.com/up/files/PDFs_RT/Brochure_RT_Advanced_Loop_
Reactor_operating_principle.pdf.
(19) Gupton, B. F.; McQuade, D. T. A Holistic Approach to
Streamlining Pharmaceutical Processes: A Conversation. Org. Process
Res. Dev. 2019, 23, 711−715.
(20) (a) Ramirez, A.; Mudryk, B.; Rossano, L.; Tummala, S. A
Mechanistic Study on the Amidation of Esters Mediated by Sodium
Formamide. J. Org. Chem. 2012, 77, 775−779. (b) Wagner, J.; von
Matt, P.; Sedrani, R.; Albert, R.; Cooke, N.; Ehrhardt, C.; Geiser, M.;
Rummel, G.; Stark, W.; Strauss, A.; Cowan-Jacob, S. W.; Beerli, C.;
Weckbecker, G.; Evenou, J.-P.; Zenke, G.; Cottens, S. Discovery of 3-
(1H-Indol-3-yl)-4-[2-(4-methylpiperazin-1-yl)quinazolin-4-yl]-
pyrrole-2,5-dione (AEB071), a Potent and Selective Inhibitor of
Protein Kinase C Isotypes. J. Med. Chem. 2009, 52, 6193−6196.
P
https://dx.doi.org/10.1021/acs.oprd.0c00427
Org. Process Res. Dev. XXXX, XXX, XXX−XXX