Synthesis of Vixotrigine, a Use-Dependent Sodium Channel Blocker. Part 1: Development of Bulk Supply Routes to Enable Proof of Concept

Article abstract of DOI:10.1021/acs.oprd.0c00382

Two syntheses of vixotrigine are reported. Route 1, adapted from the medicinal chemistry route, enabled rapid delivery of drug substance for clinical development. Route 2, which was developed to address many of the limitations of Route 1, was used to manu

Full text of DOI:10.1021/acs.oprd.0c00382

pubs.acs.org/OPRD
Article
Synthesis of Vixotrigine, a Use-Dependent Sodium Channel Blocker.
Part 1: Development of Bulk Supply Routes to Enable Proof of
Concept
Gerard Giblin,* Adrian Heseltine, William Kiesman, David MacPherson, James Ramsden, Ravi Vadali,
Michael Williams, and David Witty
Cite This: https://dx.doi.org/10.1021/acs.oprd.0c00382
Read Online
Metrics & More
Article Recommendations
ACCESS
ABSTRACT: Two syntheses of vixotrigine are reported. Route 1, adapted from the medicinal chemistry route, enabled rapid
delivery of drug substance for clinical development. Route 2, which was developed to address many of the limitations of Route 1, was
used to manufacture pilot quantities of API. Key features of Routes 1 and 2 are the generation of a chiral ketone intermediate from
an (S)-pyroglutamic acid derivative and catalytic reduction to introduce the second stereogenic center into the API with high
stereoselectivity. Route 2 was developed to address the purification burden attributable to “benzyne-derived” impurities generated in
Route 1. The improved process eliminated the possibility of the formation of these impurities, substantially improved the
stereoselectivity in the reduction of the alternate cyclic imine intermediate 22, and gave a pilot plant process with significantly
enhanced yield and throughput.
KEYWORDS: stereoselective, hydrogenation, benzyne, Grignard
(Figure 1), a use-dependent sodium channel blocker, which
has shown efficacy in a Phase II study for pain associated with
INTRODUCTION
■
Several drugs currently used in treating conditions such as pain
and epilepsy, including lamotrigine and carbamazepine, owe
their mechanism of action, at least in part, to nonselective
blockade of voltage-gated sodium channels.1 However, their
utility is limited by side effects and multiple drug−drug
interactions that require careful and prolonged titration to the
safe, effective clinical dose. Hence, there is a need for sodium
channel blockers with improved safety profiles. More recently,
there has been a resurgence of effort aimed at discovering
novel, sodium channel blockers for the treatment of pain,
driven by discoveries in human sodium channel genetics.2
A research program was initiated to identify compounds
with improved efficacy and safety based on a state-dependent
profile and ion channel selectivity. A high-throughput screen-
ing program resulted in the identification of a state-dependent
sodium channel pharmacophore from which a class of benzyl
glycinamides was elucidated. Certain analogues showed
preferential blockade of the inactivated state of the sodium
channel and therefore potential utility in treating pain, a
condition associated with high-frequency neuronal firing.
These analogues also spared the inhibition of low-frequency
nerve impulses associated with normal physiology, indicating
the prognosis for the development of safer efficacious
compounds for this target.
Figure 1. Vixotrigine.
trigeminal neuralgia and an excellent safety profile.4 In this
paper, we describe the successful efforts to define a synthetic
process for large-scale manufacturing.
RESULTS AND DISCUSSION
■
Adaptation of the Medicinal Chemistry Route for
Initial Bulk Supply. The medicinal chemistry synthesis
(Route 1) shown in Scheme 1 was adapted and implemented
to furnish drug substances for preclinical studies and early
Phase I clinical studies.5 Several changes to the process were
necessary as the process was scaled up.
Received: August 21, 2020
A lead optimization program focused on conformationally
constrained benzyl glycinamides to enhance the selectivity and
potency at sodium channels, to minimize off-target interactions
including hERG, and to provide a suitable developability
profile.3 This work led to the identification of vixotrigine 1
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
© XXXX American Chemical Society
A

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 1. Adapted Medicinal Chemistry Route (Route 1)
The key issue for Step 1, O-alkylation of 4-bromophenol 3
using commercially available 2-fluorobenzyl bromide 2, was the
purity of 2. Commercial supplies of 2 contained no more than
0.1% of the 4-fluoro isomer, but up to 4% of the 3-fluoro
isomer, and fate and purge studies demonstrated poor purge of
the isomeric 3-fluoro-impurities in the downstream process.
High-performance liquid chromatography (HPLC) methods
were developed for the separation and quantification of both
the 3-fluoro isomers and 4-fluoro isomers in the downstream
process. A stringent specification for the 3-fluoro isomer level
in 2 was not introduced at this stage of development, but use
tests demonstrated that a 3-fluoro isomer level of ≤0.3% gave
acceptable downstream impurity levels. Based on the use test
data, batches of 2 containing levels of this impurity >0.3% were
rejected. A simple precipitation method was developed for the
isolation of ether 4, which reduced the Step 1 cycle time, and
avoided the use of hexane, which was used in the medicinal
chemistry process. The Step 1 reaction mixture filtrate was
concentrated and the product precipitated by addition of 2%
aqueous NaOH. Filtration followed by washing of the filter
cake with water afforded 4 in high yield and purity. With these
improvements, pilot-plant-scale batches were run, furnishing
up to 58 kg of 4 in 91% yield as a white solid of >99.4% purity,
with the level of the 3-fluoro isomer below 0.2% and other
impurities at levels of no more than 0.3%.
was demonstrated at 30−75 kg scales, but several drawbacks
were identified, which resulted in significant cost or quality
issues.
The major issues associated with the Grignard reagent
generated from 4 relate to the formation of significant levels of
impurities. Among the impurities are 10 and 11, which appear
to be formed via Grignard addition to a benzyne impurity
intermediate,8 as shown in Scheme 2. The major isomer is 10,
and it is presumed that 11 is formed at far lower levels due to
steric hindrance at the 2-position of the benzyne intermediate.
The biaryl impurity 12, as shown in Figure 2, is formed by the
Wurtz homocoupling of the Grignard reagent.9
The benzyne-derived impurities 10 and 11 are difficult to
purge at Step 2 and are carried through the route to related
impurities in the final API step. Due to the persistence,
elaboration, and carry-through of 10 and 11 in Route 1,
substantial effort was invested to improve the profile and
selectivity of the Grignard reaction. A range of alternative
ethereal solvents was evaluated (2-methyl THF, cyclopentyl
methyl ether, and a toluene−THF mixture), but THF
remained the solvent of choice. Addition of the Grignard
reagent derived from 4 to a −55 °C solution of 5 was preferred
to the reverse addition of 5 to the similarly cooled Grignard
reagent. Reverse addition of 5 to the Grignard reagent led to a
considerable increase in the level of the homocoupled impurity
12 (9.3%). Alternative methods to generate a Grignard or
Grignard-related reactant were also attempted. Among these
were Knochel’s method of catalysis with lithium salts via
iPrMgCl/LiCl exchange,10 formation of the arylzinc reagent,11
use of the iodoaryl substrate, formation of the organo-
magnesium “ate complex”,12−14 and a range of magnesium-
In Step 2, a chemoselective addition of the Grignard reagent
derived from 4 to the commercially available chiral
pyrrolidinone ester 56,7 provided the ketone 6, incorporating
the (2S) chiral center of vixotrigine. Control of the cryogenic
reaction (−50 to −60 °C) and formation of process-related
impurities became major challenges on scale-up. The process
B
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 2. Putative Mechanism of Formation of Benzyne Impurities
cyclization to the pyrrolidine imine 7. After completion of the
reaction and neutralization, the DCM layer was concentrated.
In the first bulk campaign, hexane was added as the antisolvent,
and the precipitated solid was isolated in 82−85% yield.
Process development of this step focused on evaluating the
optimal control of pH during workup to avoid emulsion
formation and replacement of n-hexane as an antisolvent with a
safer alternative. Neutralization with aqueous KHCO3 to pH
6.0−7.0 eliminated the emulsion and maintained the chemical
yield above 80%, while heptane performed well as an
antisolvent. These changes were implemented on a pilot
plant scale for two batches with 25−30 kg inputs of ketone 6
and furnished pyrrolidine imine 7 in 78 and 88% yields as off-
white solids. The two main impurities were the cyclized
derivatives of the benzyne-derived impurities 10 and 11
generated in Step 2, the levels of which were only modestly
reduced by the Step 3 crystallization process. For example,
impurity 10 in the stage 2 product was present at 2.7% and its
cyclized derivative was present at levels of 2.5% in the stage 3
product. The remaining issues for pilot plant operation of
Route 1 were the prolonged neutralization time (up to 19 h to
maintain the vessel temperature below 10 °C and control
effervescence) and the requirement of large volumes of TFA
and DCM, which were undesirable from a safety and
environmental perspective.
Figure 2. Step 2 homocoupled impurity.
activating methods such as DIBAL15 and TMSCl.16 All of
these methods either failed to promote the formation of the
Grignard or resulted in low step yields and/or inferior impurity
profiles.
The isolation procedure for ketone 6 was also evaluated. An
extensive screening of solvents for precipitation and isolation
resulted in identification of a ternary mixture of ethyl acetate/
IPA/hexane 2:2:6. This solvent system provided the most
effective purge of the major benzyne impurity 10 to below 4%.
The pilot-plant-scale batches were run implementing these
improvements, providing up to 28 kg of ketone 6 in 54% yield
as a pale-yellow solid with 10 at levels <3%. Despite these
improvements, the yield of the Grignard reaction was modest
and the reaction was still generating levels of benzyne-derived
reaction impurities that were difficult to purge and control in
downstream processing through to the final API.
The Step 3 acid-catalyzed removal of the Boc protecting
group with TFA in dichloromethane (DCM) resulted in
C
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 3. Impurities in the route 1 Step 4 process.
a
Table 1. Route 1 Kilo-Lab (Batch 1) and Pilot Plant (Batches 2−5) of Vixotrigine
specification
batch 1
batch 2
batch 3
batch 4
batch 5
description
batch size (kg)
slightly colored
1.5
slightly colored
4.2
pale yellow
8.9
slightly brown
9.1
off-white
15.2
vixotrigine assay %w/w
organic impurities (% area)
individual specified:
benzyne-derived impurity 18
benzyne-derived impurity 19
acid 16
Trans-isomer 17
unspecified impurities
total impurities
97.0−102.0
98.9
98.5
99.8
100
100
NGT 0.15
NGT 0.5
NGT 0.5
NGT 1.0
NGT 0.15
NGT 3.0
0.05
0.14
0.15
0.69
<0.05
1.1
<0.05
0.39
<0.05
0.42
<0.05
0.9
<0.05
<0.05
0.41
<0.05
0.16
<0.05
0.82
<0.05
<0.05
0.41
<0.05
0.17
0.05
0.85
<0.05
0.34
<0.05
0.25
0.05
0.85
ND
enantiomer 20 (% area)
<0.05
<0.05
a
ND, not detected; NGT, not greater than.
Detailed evaluation of the Step 4 hydrogenation was
On the pilot plant scale, the reaction and workup conditions
developed gave 8 in 97−100% chemical yield for inputs of 7 of
8.3 kg. Levels of trans-isomer 13 were between 7 and 10%, and
benzyne-derived impurities arising from 10 and 11 persisted at
levels of 1.8−2.6%.
conducted to increase its chemo- and stereoselectivity. The
reduction of the imine was prone to high and variable levels
(5−10%) of the (5S) trans-diastereomer 13. The suboptimal
chemoselectivity in this reaction resulted in formation of two
impurities: 14, in which the o-fluorobenzyl group was cleaved;
and 15, the ring-opened over-reduction impurity (Figure 3). A
range of different catalysts, catalyst loadings, solvents, and
reaction conditions was evaluated. Replacing Pt/C with Pd/C
resulted in much higher levels of the hydrogenolysis products
14 and 15, while Raney nickel led to an unacceptably slow
reaction, with 72% unreacted pyrrolidine imine 7 substrate
remaining after 18 h. A selection of different grades of 5% Pt/C
was screened under standard conditions with only minor
effects on the yield and level of trans-isomer 13. The effect of
hydrogen pressure was evaluated, with no significant difference
in the impurity profile between 15 and 60 psi, although higher
pressures reduced the time required for a complete reaction.
There was a tendency for levels of hydrogenolysis products to
increase at 80 psi, and an increased Pt/C catalyst loading of
10% increased the amount of the ring-opened hydrogenolysis
product 15 formed. Asymmetric hydrogenation was not
examined at this stage as a route to minimize the trans
impurity; however, a solvent screen was undertaken though no
solvent was found that improved this transformation.
Step 4 was thus partially optimized. Hydrogenation of
pyrrolidine imine 7 to install the (5R) stereogenic center was
carried out using 5% Pt/C (5% w/w) under 50−60 psi
hydrogen pressure in ethyl acetate and trifluoroacetic acid (1.0
equiv). After workup, the pyrrolidine 8 was isolated in
consistently high weight yields (95−97%) as an oil, but the
stereoselectivity of the reaction remained difficult to control.
The content of the unwanted (5S) trans-diastereoimer 13
varied from ∼6 to 10% with some tendency for further erosion
of stereoselectivity as the scale increased. In addition, varying
levels of products arising from hydrogenolysis of the 2-
fluorobenzyl ether (phenol 14) and/or the benzylic amine
bond (primary amine 15) were produced (see Figure 3).
In Step 5, the ester 8 in methanol was treated with 7N
methanolic ammonia solution at an ammonia pressure of 3−4
kg/cm2 for a period of 10−12 h. The original workup
comprised multiple washes and distillations, required 16 h to
complete the crystallization, and prolonged drying to remove
toluene. Altogether, these operations required a cycle time of
∼50 h, and yields of crystallized 9 were variable, in the range of
50−65%. Laboratory optimization of this step focused on
simplification of the postreaction processing and development
of an improved crystallization procedure. The changes
resulting from this optimization included use of toluene rather
than ethyl acetate to remove methanol via codistillation,
elimination of brine washes, and replacement of toluene with
n-heptane for washes of the filter cake, to improve drying and
reduce material loss. As a consequence of the above
optimization, the overall process cycle time was reduced to
32 h, and chemical yields were consistent in the range of 60−
63% in the laboratory. Levels of the unwanted trans-isomer 17
in 9 were reduced from 9 to 2% by the crystallization. On a
pilot plant scale, 9 was isolated in more variable yields (54−
63%) as a light-brown solid having trans-isomer 17 contents of
between 0.8 and 3.8% and benzyne-derived impurity levels
between 0.2 and 0.8%.
For the Step 6 salt formation, a filtered ethyl acetate solution
of free base 9 was treated with ethyl acetate/HCl and slurried
at 0−5 °C to furnish vixotrigine 1. Kilo-lab batches produced
vixotrigine in 94−96% yield, with ∼97% purity. The
recrystallization of 1 in Step 7 was performed by suspending
the crude solid in 5% aqueous ethanol, heating to 65−72 °C to
effect solution, seeding at 50−52 °C, and then aging at 5−10
°C before isolation by filtration. The resulting final vixotrigine
API 1 was typically isolated in 80−85% yield, with 99% purity.
The final product was colored (yellow/brown), and attempts
D
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 4. Key impurities in Route 1.
Scheme 3. Route 2 Manufacturing Process for Vixotrigine
to reduce color with charcoal treatment (various grades) or
with alternative solvents were all unsuccessful. The inability to
control the color of API manufactured using Route 1
necessitated the use of color-masking in the drug product
(DP) manufacture, with yellow dye coatings applied to both
active and placebo tablets.
Pilot plant-scale batches run under the above conditions are
summarized in Table 1. Key impurities were the benzyne-
derived impurities, arising from the Step 2 Grignard reaction,
and the trans-isomer 17, arising from suboptimal control of the
Step 4 hydrogenation (see Figure 4).
resulted in an overall improvement in the impurity profile of
the final API.
Implementation of Route 1 in production provided batches
of API of between 1.5 and 15.2 kg with typical purity levels of
98.5−99.0%. API was a solid ranging in color from off-white to
light brown. Assays were developed for 20, the enantiomer of
vixotrigine (typical level in API < 0.05%), 16, the acid
degradant of vixotrigine (typical level <0.05−0.15%), 17, the
trans-isomer (typical level 0.17−0.69%), and 19 (typical levels
0.14−0.41%) and 18 (typical levels <0.05%), from carry-
through of the benzyne-derived impurities.
The process modifications described above, including the
Route 1 performed adequately to provide API for preclinical
improved control of the formation of trans-isomer 17 in Step 5,
toxicology studies and early clinical studies. However, this
E
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 5. Impurities arising from Route 2 Step 1 and in subsequent processing.
route had several drawbacks that precluded its use for later-
stage development and commercialization
Since high-quality benzyloxy-4-bromobenzene with less than
0.15% of either the 2-Br or 3-Br isomer could be purchased,
Step 1 in the Route 2 process became the Grignard addition to
N-Boc-methyl-(S)-pyroglutamate 5. For the first pilot plant
campaign, the conditions for the Grignard reagent formation
and reaction used for Route 1 were retained. The product
crystallization solvent was changed to MTBE/heptane. Two
batches provided 24.6 kg of the pyroglutamate adduct 21, with
both an improved yield (78 vs 54%) and HPLC purity (98.7 vs
95.5%) relative to the analogous step in the Route 1 process. In
the absence of benzyne-derived side products, the major
impurity in 21 was the homocoupled product 269 at about
0.7%, with no other impurities above 0.2%. The level of dimer
26 is reduced in the isolation of 22 in Step 2, and the
downstream impurity 27 is readily purged in the Step 3
isolation, justifying a specification of not more than 1.0% 26
within isolated 21.
For the second campaign, the input of 5 was scaled up over
fourfold to 37 kg per batch, with between five and seven
reactions run for each of Route 2 campaigns two, three, and
four before pooling for product isolation. The further benzyl
ether side products 28 (from Grignard oxidation)18 and 29
(from Grignard quenching) were identified as low-level step
impurities that were purged in subsequent steps (see Figure 5).
The product quality was maintained across these campaigns,
with 900 kg of 21 produced in ∼73% yield for campaigns two
to four. For the fifth and sixth Route 2 campaigns, the supply
of Grignard product 21 of the required quality was outsourced.
The first new route Step 2 campaign retained the route 1
process to cleave the Boc group and effect cyclization to the
pyrrolidine imine 22, but with reduction in the TFA excess
(from 11 to 8.9 equiv) and volume of DCM (from 10 to 6.5
mL/g). Thus, a 21.1 kg input of 21 produced the crystalline
imine 22 in 90% yield. The more environmentally acceptable
toluene, also at 6.5 mL/g, was used rather than DCM for the
remaining five campaigns. This revised process scaled
smoothly, with inputs of 21 from 117 up to 562 kg producing
imine 22 of comparable quality in yields of between 87 and
91%. Apart from the homocoupled impurity 26 (levels of 0.1−
0.5%), which is purged in Step 3, the major impurity was the
acid 30 (present at levels that were in the range of 0.30−
0.76%), formed by hydrolysis during the workup. The fate of
acid 30 in the process was studied, and the acid analogue of the
API is a specified impurity. The imine 22 is prone to
epimerization, so a chiral assay was developed, and the
enantiomer 31 (see Figure 6) was consistently controlled to
levels below 0.1%.
• The yield of the Grignard addition reaction (Step 2) was
poor.
• The stereoselectivity of the catalytic hydrogenation to
reduce the imine 7 to install the 5R stereocenter was
inadequate (80−88% de) and low-yielding due to the
extensive downstream processing to achieve acceptable
purity. Moreover, levels of impurities arising from
hydrogenolysis side reactions (forming 14 and 15)
were difficult to control. The need for crystallization in
Step 5 to improve the diastereomeric purity of 8 by
reducing levels of the trans-isomer had a profound
negative impact on the overall yield of the final drug
substance.
• Benzyne-derived impurities 10 and 11, formed in Step 2,
were transformed and carried through to the final API.
Control of the level of 18 from the minor Step 2 isomer
11 was achieved. However, purging of 19 from the major
Step 2 impurity 10 was more challenging, leading to a
high burden for purification in the later steps of the
process, which was operationally cumbersome.
• The appearance of the final, isolated API was variable,
with respect to color. This required that the variability in
color of the API within the drug product be masked with
a special dye coating.
Development of the Improved Second Pilot Plant
Process. Early introduction of the 2-fluorobenzyl group in the
enabling Route 1 process led to two major issues on scale-up.
First, the benzyne-derived side products formed during
preparation of the Grignard reagent were not purged efficiently
in subsequent steps and required time-consuming purifications
that reduced the overall process yield. Second, the 2-
fluorobenzyl bromide starting material 2 is costly. Considering
that the overall yield of Route 1 is only 18%, 2 alone
contributed about 30% to the cost of the Route 1 raw material.
The pursuit of an improved process to supply vixotrigine API
therefore focused on introduction of the 2-fluorobenzyl group
later in the process, necessitating the selection of an
appropriate phenol protecting group for the Grignard step.
Following a laboratory evaluation, O-benzyl protection was
selected, with the route depicted in Scheme 3 chosen for
development and scale-up. A key advantage of the use of the
benzyl protecting group was the ability in the Step 3 process to
reduce the imine, capture the newly formed amine with the N-
Boc protecting group, and then cleave the O-benzyl group.17
F
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Table 2. Route 2 Pilot Plant Results for the Step 3 Process
Step 3
yield
(%)
Route 2 input of
no. of
output of
campaign 22 (kg) reactions 23 (kg)
HPLC purity (%)
1
2
3
4
5
6
13.7
121.24
178.46
178.5
178.6
331.5
2
12.46
137.67
164.3
161.8
161.3
284.3
86.5
90.6
88.6
87.3
86.5
99.78 and 99.78
99.64 and 99.77
99.64 and 99.85
99.82 and 99.91
99.96 and 99.96
99.99
a
Figure 6. Impurities in the Route 2 Step 2 process.
5
7
a
a
a
b
7
7
The key third step of the process was developed to
sequentially accomplish three transformations in the same
reaction vessel (Scheme 4).
c
13
82.3
a
For each of these campaigns, worked-up sub-batches were pooled to
provide two portions for isolation. For this campaign, individual 25−
26 kg-scale reactions were pooled for isolation as a single batch. The
product was initially isolated in 87.3% yield and met all specifications
except for a 1.9% level of 33. The OOS level of 33 resulted from
overenrichment of the crystallization solution in n-heptane due to
excessive removal of EtOAc during the distillation. Reprocessing in
the EtOAc/heptane crystallization solvent system provided 23, which
met the specification for the level of 33 in 94.7% recovery.
b
c
1. Reduction of the imine to generate the cis-pyrrolidine
intermediate.
2. Boc protection to give 32.
3. Hydrogenolysis to cleave the O-benzyl group to give 23.
The aryl C−N bond of the pyrrolidine intermediate is
susceptible to hydrogenolysis, leading to a ring-opened
impurity 33, but Boc protection of the nitrogen atom inhibits
benzylic C−N bond cleavage. Careful choice of the grade of
Pd/C catalyst ensured that ring-opening prior to Boc addition
was minimized while achieving good cis-/trans-selectivity in the
imine reduction. The cis-/trans-selectivity of the Route 1 Pt-/
C-catalyzed imine reduction reaction was only about 7:1
(Scheme 1), but with 5% Pd/C type 487 paste, a cis-/trans-
selectivity of >30:1 was achieved in methanol in the Route 2
process, minimizing the level of trans-isomer 34 formed. A fate
and purge study showed that a 0.4% level of 32 in the Step 3
product 23 was converted to API containing 0.08% des-fluoro
impurity. An in-process control (IPC) level of not more than
(NMT) 0.5% 32 was therefore set for the reaction, to ensure
that the downstream level of the des-fluoro impurity met its
specification level of NMT 0.15% in the final API.
A range of crystallization solvents was examined to optimize
the purge of impurities after Step 3. An ethyl acetate/heptane
solvent system achieved the best results and enabled efficient
purging of impurities due to the high crystallinity of 23. All of
the pilot plant batches, summarized in Table 2, contained
<0.05% each of 32, 33, and 34 and no detectable 27 arising
from the Step 1 process impurity 26. The high crystallinity and
particle characteristics of 23 isolated from this crystallization
resulted in quick filtration and drying of the solid, establishing
it as a key intermediate and point of purity control for this and
future processes.
Alkylation of the phenol 23 with 2-fluorobenzyl bromide 2
in Step 4 was performed in acetone with potassium carbonate
as the base. The levels of regioisomers of 2-fluorobenzyl
bromide were closely monitored, due to the placement of this
reaction later in the process for Route 2, as compared with
Route 1. Trials found that using 2 spiked with 0.4% each of 3-
fluorobenzyl bromide and 4-fluorobenzyl bromide led to Step
4 product 24 in which the levels of 3- and 4-fluorobenzyl
regioisomers had been reduced two- to fourfold. The 3- and 4-
fluorobenzyl regioisomeric impurities of the final API and its
des-fluoro impurity all had specification levels of NMT 0.15%.
Accordingly, the specifications for the regioisomeric fluori-
nated impurities in 2 were set to NMT 0.2% each, and the
specification for the des-fluoro impurity was set at NMT 0.15%
in 2.
The six Route 2 manufacturing campaigns produced over
1000 kg of 24, with yields ranging from 85 to 89%. The only
impurities identified in 24 at levels of >0.1% in these
campaigns were unalkylated phenol 23 (up to 0.6%) and the
3-fluorobenzyl analogue (<0.2%). In the first campaign, it was
found that phenol 23 did not purge well in the crystallization
of 24, but the 0.6% level seen in one batch was efficiently
purged in downstream processing. For future campaigns, the
Scheme 4. Conversion of 22 to 23 in Step 3
G
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Table 3. Route 2 Pilot Plant Results for Step 5
Route 2 campaign
input of 24 (kg)
number of reactions
output of 25 (kg)
Step 5 yield (%)
HPLC purity of 25 (%)
level of acid 16 (%)
1
2
3
4
5
6
14.0
161.46
192.5
182.9
189.2
327.9
1
2
1
1
1
2
10.38
173.93
204.58
194.74
195.8
96.1
92.0
91.0
91.2
88.6
94.6
99.15
99.82 and 99.71
99.81
99.70
99.68
0.04
0.18 and 0.20
0.17
0.28
0.19
362.6
99.74 and 99.78
0.18 and 0.17
Table 4. Route 2 Pilot Plant Results for Step 6
Route 2 campaign
input of 25 (kg)
number of reactions
output of 9 (kg)
Step 6 yield (%)
HPLC purity (%)
level of acid 16 (%)
1
2
3
4
5
6
10.3
173.8
204.27
197.4
195.8
362.4
1
2
1
1
1
2
9.16
97.97
117.38
113.05
114.1
93.0
89.9
91.7
92.6
93.0
88.0
99.87
99.67, 99.97
99.86
99.87
99.74
0.04
0.17 and 0.02
0.08
0.06
0.11
199.8
99.85, 99.73
0.05 and 0.12
IPC for unreacted 23 was tightened from NMT 1.0% to NMT
0.5%.
Isolation of the Step 5 product as its p-TsOH salt 25 led to
advantages in handling and impurity control. The extra
operations in Step 6, including the salt break, required
additional studies to demonstrate the extent of formation of
p-TsOH esterspotential genotoxic impuritiesand their
purge in subsequent process steps. The Step 6 amidation
developed employed a large excess of ammonia and was one of
the steps that required further attention to improve the volume
productivity and green metrics of the vixotrigine manufacturing
route.
For the first three campaigns for Step 7 using Route 2,
conversion of 9 to the HCl salt 1 was carried out by addition of
1.25 N HCl in ethanol to an ethanol solution of the free base at
25−35 °C. This process provided excellent yield and quality,
but the initially formed ethanol solvate of the HCl salt required
an extended drying time to liberate EtOH and meet the
required residual solvent content acceptance criteria. In
addition, the particle size distribution (PSD) was lower than
preferred at that time for the formulation process, mainly due
to the presence of fines. An alternative process was therefore
developed for campaigns 4 and 5 in which the free base 9 in
IPA solution at 60−65 °C was treated with aqueous HCl and
then cooled to 0−5 °C for collection of the salt 1. This second
process gave a product that dried more rapidly, with a higher
bulk density and a d50 measurement in the required range of
50−100 μm. Following a reassessment of the PSD require-
ment, campaign 6 reverted to the process run in ethanol
solvent, giving four batches with d50 measurements in the 15−
20 μm range. The drying times for these batches were
significantly reduced by a steeper temperature ramp and more
intensive agitation in the Nutsche filter dryer.
In Step 5 of the first campaign run using the Route 2
process, deprotection of the Boc group of 24 used aqueous
phosphoric acid (H3PO4, 85% aq solution) in MTBE at 15−20
°C.19 However, the processing involved relatively large workup
volumes, generated significant amounts of inorganic salts from
the neutralization of the large excess of phosphoric acid, and
gave the free base as an oil after stripping to dryness, with poor
plant volume productivity. A new process was therefore sought
that would enable the direct formation of the Boc-free
intermediate as a suitable salt in a “direct drop process”.
Studies of the best acid and solvent combination resulted in
the selection of p-TsOH in methanol as the procedure used for
future campaigns (see Table 3 summary).
The procedure used for the first Route 2 campaign provided
the free base of 25, which was isolated as an oil with 99.15%
purity, with unreacted 24 as the main impurity at 0.58%. For
subsequent campaigns 2−6, 25 (p-TsOH salt) was isolated in
∼99.7% purity, with acid 16 present at about 0.2% and the 3-
fluoro regioisomer present at a reduced level of <0.05%.
Methyl ester 25 was converted to the amide 9 in Step 6
using a 7N solution of ammonia in methanol. For the first
campaign using the oily free base of 25 as input, the product
was isolated by a solvent swap with toluene, followed by
addition of heptane cosolvent to crystallize 9. For the
subsequent campaigns, the p-TsOH salt of 25 was converted
to its free base in MTBE. A solvent swap replaced MTBE with
methanol for the amidation reaction, and addition of water was
used to crystallize the amide 9. Results for the six campaigns
are summarized in Table 4. All batches of 9 were >99.7% pure
by HPLC. Acid 16 and the trans-amide 17 (Figure 7) were the
only impurities that exceeded the 0.10% level in any of these
batches.
The six Route 2 campaigns produced over 660 kg of 1, with
a Step 7 yield range of 91−95%. The purities of all batches of 1
were >99.8% by HPLC with the acid impurity 16 reduced in
the salt formation to ≤0.07%, all unknown impurities at
<0.10%, and no detectable levels of the trans-amide 17 or of
the enantiomer 15. In addition, the material prepared by Route
2 was white, obviating the need for the color-masking required
for API produced using Route 1 in the manufacture of drug
products.
Figure 7. Impurities in Route 2 Step 5 and Step 6.
H
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
aqueous ammonium chloride (15 kg in 55 L of water) and
aqueous sodium chloride (9.9 kg in 28 L of water) was added
over 2 h maintaining the temperature below −50 °C, and
following completion of the addition, the mixture was
maintained at −50 °C for 20 min before raising the
temperature to 30 °C over 2 h. Water (110 L) was added
followed by aqueous acetic acid to adjust the pH to 7. Ethyl
acetate washes (2 × 138 L) were combined and washed with
aqueous sodium chloride, and the organics reduced under a
vacuum. A mixture of IPA (55 L), ethyl acetate (55 L), and
hexane (165 L) was added, and the crude mass was stirred for
3 h at 30 °C and for 3 h at 5 °C. The resultant solid was
filtered, and the IPA/EtOAc/hexane wash was repeated 2
further times; the resultant solid was dried at 50 °C under a
vacuum to furnish the product (26.8 kg, 53%) as a pale-yellow
solid. 1H NMR (600 MHz, DMSO-d6) δ (ppm): 7.93 (d, 2H),
7.57 (td, 1H), 7.44 (m, 1H), 7.27 (m, 3H), 7.14 (d, 2H), 5.24
(s, 2H), 4.04 (m, 1H), 3.61 (s, 3H), 3.03 (m, 2H), 1.94 (m,
2H), 1.38 (s, 9H).
Methyl (2S)-5-[4-(2-Fluorobenzyloxy)phenyl]-3,4-dihydro-
2H-pyrrole-2-carboxylate (7). Ketone 6 (26 kg) and dichloro-
methane (260 L) were stirred at 30 °C under nitrogen for 20
min and cooled to 5 °C. Trifluoroacetic acid (47.7 kg) was
added over 1 h at 5 °C before the temperature was raised to 30
°C for 3 h. The reaction mass was cooled to 5 °C, and a
solution of potassium bicarbonate (57.2 kg) in water (221 L)
was added until the aqueous layer reached pH 7. The
temperature was raised to 30 °C, and after separation, the
aqueous layer was washed with DCM (130 L) and the
combined organics were washed with water (130 L) and
reduced under a vacuum to 1.5 volumes before heptane was
added and sequentially distilled (2 × 52 L). The crude mass
was suspended in heptane (260 L) and stirred for 3 h at 30 °C
before filtration, and drying in vacuo at 45 °C for 10 h gave the
product (14.98 kg, 78%) as a light-brown solid. 1H NMR (400
MHz, DMSO-d6) δ (ppm): 8.16 (m, 2H), 7.60 (td, 1H), 7.46
(m, 1H), 7.34 (m, 2H), 7.27 (m, 2H), 5.32 (s, 2H), 5.25 (m,
1H), 3.77 (s, 3H), 3.57 (m, 2H), 2.60 (m, 1H), 2.34 (m, 1H).
(2S,5R)-Methyl 5-[4-(2-Fluorobenzyloxy)phenyl]-
pyrrolidine-2-carboxylate (8). Imine 7 (8.3 kg) was dissolved
in ethyl acetate (41.5 L), and trifluoroacetic acid (1.88 L) was
added at 30 °C under nitrogen and stirred for 30 min. A slurry
of 5% w/w platinum on carbon (0.25 kg) in ethyl acetate (41.5
L) was added, and the reaction mass was stirred for 5 h under
60 psi hydrogen pressure at 30 °C. After filtering through a
Hyflo bed, the filtrate was washed with sodium bicarbonate
solution, and after separation, the aqueous layer was re-
extracted with ethyl acetate (24.9 L). The combined organics
were evaporated to 1 vol, and the mass was codistilled with
methanol (2 × 24.9 L) to give the product (8.15 kg, 97.6%) as
a dark-brown liquid, which was used directly in the next step.
1H NMR (600 MHz, DMSO-d6) δ (ppm): 7.55 (dt, 1H), 7.41
(m, 1H), 7.34 (m, 2H), 7.23 (m, 2H), 6.97 (m, 2H), 5.12 (s,
2H), 4.09 (dd, 1H), 3.83 (dd, 1H), 3.66 (s, 3H), 2.97 (bs,
1H), 2.04 (m, 2H), 1.94 (m, 1H), 1.52 (m, 1H).
CONCLUSIONS
■
Route 1 to vixotrigine, the enabling process adapted from the
medicinal chemistry synthesis, was used to produce about 40
kg of API to start the development program. Subsequent to
this, an improved process, Route 2, was developed. Route 2
enabled significant improvements in both the purity (from
between 98.5 and 99.0% to >99.8%) and overall yield (from
∼18 to ∼40%) of vixotrigine. During multiple pilot plant
campaigns, the fate, purge, and improved control of process
impurities in the process were established, and Route 2 was
demonstrated to be capable of consistent production of API
with a higher quality. Over 660 kg of vixotrigine was
manufactured in six pilot plant campaigns to enable clinical
proof of concept to be established.
EXPERIMENTAL SECTION
■
Proton magnetic resonance (NMR) spectra were recorded
with TMS as the internal standard using a Bruker 400, 500, or
600 MHz spectrometer and were typically run at 25 °C.
Melting points were determined using a Stanford Research
Systems Optimelt instrument at a heating rate of 1 °C/min
and are uncorrected. Liquid chromatography-mass spectrom-
etry (LC-MS) data was generated on a Waters ZQ Mass
Spectrometer, operating in switched ES+ and ES− ionization
modes coupled to an Agilent 1100 Series HPLC system with
in-line Agilent 1100UV-DAD and Sedere SEDEX 75 ELSD
detection. Instrument control and data acquisition were
mediated through the Waters MassLynx-OpenLynx software
suite.
Route 1. 1-[(4-Bromophenoxy)methyl]-2-fluorobenzene
(4). Potassium carbonate powder (47.08 kg) and acetone (645
L) were warmed to 35 °C under a nitrogen atmosphere, and 4-
bromophenol 3 (41.32 kg) was added followed by 2-
fluorobenzyl bromide 2 (43.00 kg) over 10 min. After stirring
at 35 °C for 10 min, the mixture was heated to 60 °C for 8 h.
The reaction was cooled to 35 °C, and the solid was filtered
and washed with acetone (215 L). The filtrate was evaporated
to a syrupy mass, and water (430 L) was added followed by
sodium hydroxide (2.25 kg) over 30 min at 35 °C. After
stirring for 3 h, the resulting solid was filtered and washed with
water (215 L) and dried for 12 h under a vacuum at 50 °C to
give the product (58 kg, 90.7%) as an off-white solid. 1H NMR
(400 MHz, DMSO-d6) δ (ppm): 7.54 (td, 1H), 7.46 (d, 2H),
7.42 (m, 1H), 7.23 (m, 2H), 7.01 (d, 2H), 5.13 (s, 2H).
Methyl (2S) 5-[4-(2-Fluorobenzyloxy)phenyl]-2-[(tert-
butoxycarbonyl)amino]-5-oxopentanoate (6). Magnesium
turnings (8.14 kg) and iodine (0.11 kg) were charged into a
vessel under a nitrogen atmosphere and heated to 120 °C for 3
h. The mixture was cooled to 30 °C, tetrahydrofuran (THF)
(55 L) and further iodine (0.055 kg) were added, and the
mixture was refluxed for 25 min. A separate reactor under a
nitrogen atmosphere was charged with 4 (55.66 kg) in THF,
and 5% of this solution was added to the first reactor at 65 °C
followed by 1,2-dibromoethane (0.11 kg) to initiate the
reaction. The remaining 4 solution was then added at reflux
over 5 h and then cooled to 30 °C. N-Boc L-pyroglutamic acid
methyl ester (5) (27.5 kg) in THF (10 L) was cooled to −60
°C, and the Grignard solution was added over a period of 5 h
maintaining the temperature below −50 °C. On completion of
the addition, the mixture was maintained at −50 °C for 2 h
and isopropyl alcohol (IPA, 28 L) was added over a period of 2
h maintaining the temperature below −50 °C. A mixture of
(2S,5R) 5-[4-(2-Fluorobenzyloxy)phenyl]pyrrolidine-2-car-
boxamide (9). Methanol (130.4 L) was purged with gaseous
ammonia for 7 h at 0 °C, ester 8 (16.30 kg) was added to the
ammonia solution, and ammonia pressure of 2 kg/cm2 was
established at a reaction temperature of 30 °C for 8 h. A
nitrogen purge was applied for 4 h, and the solvent was
reduced under a vacuum to 2 volumes. The reaction mass was
codistilled with toluene (2 × 49 L), and a further batch of
I
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
toluene (49 L) was added; after stirring at 5 °C for 2.5 h, the
solid was filtered and washed with n-heptane (32.6 L). The
mass was stirred with water (81.5 L) for 45 min and filtered,
washed with water (32.6 L), and dried under a vacuum at 55
°C for 14 h. The product (8.35 kg, 53.7%) was isolated as a
Demineralized water (36 L) was added, and the pH was
adjusted to 6−7 by the addition of 1:1 aqueous acetic acid.
The product was extracted into MTBE (2 × 45 L), and the
combined extracts were washed with a solution of NaCl (13.5
kg) in water (54 L). The organic layer was concentrated under
a vacuum to about 14 L, cooled to 30 °C, and treated with a
mixture of MTBE (27 L) and heptane (18 L). The mixture was
stirred for 2 h and filtered, and the product was washed with
heptane (18 L) and dried in a vacuum oven (50 °C) for 8 h.
The product 21 (12.25 kg, 78%) was a white solid, mp 127 °C.
LC-MS MH+ = 428 (C24H29N06 requires 427). 1H NMR (400
MHz, CDCl3) δ (ppm): 1.52 (9H, s), 2.10 (1H, m), 2.30 (1H,
m), 2.95−3.15 (2H, m), 3.76 (3H, s), 4.40 (1H, m), 5.15 (2H,
s), 5.18 (1H, br s), 7.03 (2H, d, J = 9 Hz), 7.30−7.50 (5H, m),
7.95 (2H, d, J = 9 Hz).
(S)-Methyl 5-[4-(Benzyloxy)phenyl]-3,4-dihydro-2H-pyr-
role-2-carboxylate (22). Ketone 21 (141.4 kg) and toluene
(919 L) were stirred at 30 °C under nitrogen for 20 min, and
the solution was cooled to 5 °C. Trifluoroacetic acid (334 kg)
was added over 1 h at 5 °C before the temperature was raised
to 30 °C for 3 h. The reaction mass was cooled to 5 °C, and a
solution of potassium bicarbonate (311 kg) in water (882 L)
was added until the aqueous layer reached pH 7, keeping the
temperature below 10 °C. The temperature was raised to 30
°C, and after separation, the upper organic layer was washed
with water (438 L). The separated toluene solution was
reduced under a vacuum to 1.2 volumes, and the concentrate
was maintained at 35−40 °C as heptane (298 L) was added.
The mixture was cooled to 25 °C and stirred for 1 h to effect
crystallization and then cooled to 5 °C and stirred for a further
1 h. Filtration, washing with a chilled mixture of toluene (36 L)
and heptane (72 L), and drying in vacuo at 45 °C for 10 h gave
the product (91.98 kg, 89.9%) as a light-brown solid. 1H NMR
(400 MHz, DMSO-d6) δ (ppm): 8.16 (m, 2H), 7.60 (td, 1H),
7.46 (m, 1H), 7.34 (m, 2H), 7.27 (m, 2H), 5.32 (s, 2H), 5.25
(m, 1H), 3.77 (s, 3H), 3.57 (m, 2H), 2.60 (m, 1H), 2.34 (m,
1H).
1
light-brown solid. H NMR (500 MHz, DMSO-d6) δ (ppm):
1.39 (m, 1H), 1.84 (m, 1H), 2.04 (m, 2H), 3.54 (m, 1H), 4.09
(m, 1H), 5.12 (s, 2H), 6.96 (d, 2H), 7.15 (m, 1H), 7.25 (m,
2H), 7.34 (d, 2H), 7.41 (m, 2H), 7.55 (t, 1H).
(2S,5R) 5-[4-(2-Fluorobenzyloxy)phenyl]pyrrolidine-2-car-
boxamide Hydrochloride (1). Free base 9 (8.2 kg) was
dissolved in ethyl acetate (205 L) at 25 °C, and 1N HCl
solution in ethyl acetate (34 L) was added over 45 min. The
reaction was stirred for 2 h and filtered, washed with cold (5
°C) ethyl acetate (8.2 L), and dried at 60 °C under a vacuum
for 14 h. The product (8.65 kg, 94.6%) was a slightly colored
1
solid. H NMR (400 MHz, DMSO-d6) δ ppm 10.19 (br. s.,
1H), 8.13 (br. s., 1H), 7.94 (s, 1H), 7.60−7.77 (m, 1H), 7.51
(dt, 1H), 7.43 (d, 2H), 7.34−7.41 (m, 1H), 7.23 (d, 1H), 7.18
(dd, 1H), 7.05 (d, 2H), 5.13 (s, 2H), 4.49−4.60 (m, 1H),
4.19−4.28 (m, 1H), 2.17−2.38 (m, 2H), 2.05−2.16 (m, 1H),
1.92−2.03 (m, 1H).
(2S,5R) 5-[4-(2-Fluorobenzyloxy)phenyl]pyrrolidine-2-car-
boxamide Hydrochloride (1). Crude HCl salt 1 (19 kg) in 15
volumes of 5:95 v/v % of aqueous ethanol was heated to 70 °C
until a solution was obtained. The solution was filtered through
a 1 μm polishing filter, cooled to 50 °C, and held to initiate
crystallization. The slurry was cooled in stages to 5 °C, held for
1−2 h, and then filtered. The filter cake was washed with
ethanol at 5 °C, and the product was dried under a vacuum at
72 °C to give the product (15.2 kg, 82%) as an off-white solid.
[α]D = −30.5°. MS: (ES/+) m/z: 315 [MH+], C18H19FN202
1
requires 314; H NMR (400 MHz, DMSO-d6) δ ppm 10.19
(br. s., 1H), 8.13 (br. s., 1H), 7.94 (s, 1H), 7.60−7.77 (m,
1H), 7.51 (dt, 1H), 7.43 (d, 2H), 7.34−7.41 (m, 1H), 7.23 (d,
1H), 7.18 (dd, 1H), 7.05 (d, 2H), 5.13 (s, 2H), 4.49−4.60 (m,
1H), 4.19−4.28 (m, 1H), 2.17−2.38 (m, 2H), 2.05−2.16 (m,
1H), 1.92−2.03 (m, 1H).
(2S,5R)-1-tert-Butyl 2-Methyl 5-(4-Hydroxyphenyl)-
pyrrolidine-1,2-dicarboxylate (23). To a solution of 22 (25
kg) in methanol (100 L) under nitrogen were added 5% Pd on
C catalyst (2.84 kg, type 487 paste) and a solution of di-tert-
butyl dicarbonate (19.3 kg) in methanol (20 L). The reactor
was purged and integrity checked at 4 bar, and the contents
were warmed to 30 °C. The mixture was hydrogenated at 30
°C under 4 bar of hydrogen for about 20 h, the head space was
purged to remove carbon dioxide generated during the process,
and hydrogenation was continued for another 10 h until
hydrogen uptake ceased. The mixture was filtered to remove
the catalyst, which was washed with methanol (10 L). The
combined filtrate and washings was concentrated under
reduced pressure to half volume, and ethyl acetate (75 L)
was added. The solution was concentrated under reduced
pressure to about 70 L, and further ethyl acetate (75 L) was
added. The solution was concentrated under reduced pressure
to 65 L to chase out any remaining methanol, and heptane
(200 L) was added. The mixture was stirred for 2 h at 20 °C,
cooled to 0−4 °C, and stirred for a further 1 h and filtered.
The solid was washed with 4:1 heptane/ethyl acetate (2 × 25
L cooled to 0−4 °C) and dried in a vacuum oven at 40 °C.
The product 23 (22.59 kg, 87%) was a white crystalline solid,
mp 174−175 °C. LC-MS MH+ = 322 (C17H23N05 requires
Route 2. (S)-Methyl 5-[4-(Benzyloxy)phenyl]-2-[(tert-
butoxycarbonyl)amino]-5-oxopentanoate (21). Mg turnings
(2.66 kg) and iodine (36 g) were heated together under
nitrogen at 105 °C for 2−3 h. The mixture was cooled to 30
°C, THF (18 L) and iodine (36 g) were added, and the
mixture was heated to 65 °C. A solution of benzyloxy-4-
bromobenzene (17 kg) in THF (45 L) was added portionwise
over about 6 h under nitrogen to the above suspension of Mg
maintained at reflux, with an addition of 1,2-dibromoethane
(36 mL) made after 30 min to initiate Grignard formation.
After completion of the addition, the mixture was refluxed for a
further 2 h and then cooled to 30 °C.
A separate vessel was charged with N-Boc L-pyroglutamic
acid methyl ester (5) (9 kg) and THF (45 L) under nitrogen,
and the solution was cooled to −65 °C. The above Grignard
solution was added slowly over 2.5 h maintaining the reaction
temperature between −60 and −70 °C and then stirred for a
further 1.5 h at this temperature after completion of the
addition. Isopropanol (9 L) was added to the reaction mixture
at −60 to −70 °C over 1 h, and the mixture was stirred at this
temperature for a further 20 min. A solution of ammonium
chloride (4.9 kg) and NaCl (3.24 kg) in water (27 L) was
added to the reaction mixture at −60 to −70 °C over 1.5 h,
and the mixture was stirred at this temperature for a further 15
min. The reaction was warmed to 30 °C over 2 h.
1
321). H NMR (400 MHz, CDCl3, a mixture of rotamers) δ
(ppm): 1.19 and 1.43 (9H, 2s), 1.85−2.40 (4H, series of m),
J
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
3.83 (3H, s), 4.35−5.50 (3H, series of m), 6.77 (2H, d, J = 9
Hz), 7.42 (2H, m).
and the concentration was continued to remove a further 80 L
of distillate. The reaction mixture was cooled to below 20 °C,
and 7 N ammonia in methanol (312 kg) was added while
maintaining the temperature below 3 °C. The mixture was
stirred for 24 h at 25−30 °C, and then, purified water (400 L)
was added while maintaining the temperature at 25−30 °C;
the mixture was stirred at this temperature for 2 h. The
suspension was filtered, and the solid was washed with a
mixture of water (80 L) and methanol (80 L) and dried in a
vacuum oven at 50 °C. The product 9 (47 kg, 93.7%) was a
white crystalline solid, mp 119.5 °C. LC-MS MH+ = 315
(2S,5R)-1-tert-Butyl 2-Methyl 5-[4-(2-Fluorobenzyloxy)-
phenyl]pyrrolidine-1,2-dicarboxylate (24). To a solution of
23 (12.0 kg) in acetone (120 L) was added powdered
potassium carbonate (15.5 kg), and the temperature of the
slurry was adjusted to 25−35 °C. A charge of 2-fluorobenzyl
bromide (7.75 kg) was added at a rate that kept the
temperature at 25−35 °C, with a line wash of acetone (6 L).
The mixture was heated to reflux and stirred under nitrogen for
30 h. The mixture was cooled to 30 °C and filtered, and the
solids were washed with acetone (60 L). The filtrate plus wash
was distilled under a vacuum at a maximum temperature of 40
°C to a volume of approximately 20−30 L. A charge of ethyl
acetate (48 L) was added, and the solution was distilled under
a vacuum at a maximum temperature of 50 °C until
approximately 20−30 L of the reaction mixture remained
and then cooled to below 25 °C. Ethyl acetate (60 L) and
water (15 L) were added, and the mixture was stirred at 20−30
°C for 15−20 min. After separation, the aqueous layer was
extracted with further ethyl acetate (60 L), and the combined
organic layers were washed with water (60 L). The organic
phase was concentrated under a vacuum at below 50 °C until
approximately 20−30 L remained. Heptane (48 L) was added,
and the solution was concentrated to approximately 20−30 L.
Further heptane (48 L) was added, and the concentration to
20−30 L was repeated to chase out any remaining ethyl
acetate. A 50:1 v/v mixture of heptane/ethyl acetate was
added, and the mixture was heated to 42 2 °C and stirred at
this temperature for 30 min. The mixture was cooled until
crystallization occurred, further cooled to 0−4 °C, stirred for 2
h, and then filtered. The solid was washed with heptane (2 ×
24 L cooled to 0−4 °C) and dried in a vacuum oven at 40 °C.
The product 24 (14.1 kg, 87.9%) was a white crystalline solid,
mp 61.0−61.3 °C. LC-MS MH+ = 430 (C24H28FN05 requires
1
(C18H19FN202 requires 314). H NMR (400 MHz, CDCl3) δ
(ppm): 1.68 (1H, m), 2.06−2.35 (3H, series of m), 2.51 (1H,
br s), 3.88 (1H, dd, J = 3 Hz, 9 Hz), 4.31 (1H, dd, J = 6 Hz, 9
Hz), 5.15 (2H, s), 5.57 (1H, br s), 6.99 (2H, d, J = 9 Hz), 7.11
(1H, m), 7.18 (1H, m), 7.30−7.40 (3H, m), 7.53 (1H, m).
(2S,5R) 5-[4-(2-Fluorobenzyloxy)phenyl]pyrrolidine-2-car-
boxamide Hydrochloride (1). A slurry of 9 (9.0 kg) in ethanol
(118 L) was heated to 50 °C, the resulting solution was
filtered, and the filter was washed with further ethanol (38 L).
The filtrate plus washings were cooled to 20 °C, and a solution
of 1.25 M HCl in ethanol (31 kg) was added over about 1 h
with cooling to maintain the temperature between 20 and 25
°C. The HCl solution was rinsed in with more ethanol (3.5 L),
and the mixture was stirred at ambient temperature for 2 h and
then chilled to 0−5 °C and stirred for another 2 h. The solid
was collected by filtration, washed with ethanol (2 × 17 L) that
had been chilled to about 5 °C, and dried in a vacuum oven at
70 °C. The product 1 (9.6 kg, 95.9%) was a white crystalline
solid, mp 233−234 °C. LC-MS MH+ = 315 (C18H19FN2O2
1
requires 314). H NMR (400 MHz, d6-DMSO) δ (ppm):
1.95−2.20 (2H, m), 2.30 (2H, m), 3.35 (1H, s), 4.30 (1H, m),
4.61 (1H, m), 5.18 (2H, s), 7.10 (2H, d, J = 9 Hz), 7.18−7.30
(2H, m), 7.40 (1H, m), 7.47 (2H, d, J = 9 Hz), 7.56 (1H, m),
7.72 (1H, s), 8.07 (1H, s), 10.60 (1H, br s).
1
429). H NMR (400 MHz, CDCl3, a mixture of rotamers) δ
(ppm): 1.15 and 1.40 (9H, 2s), 1.93−2.32 (4H, series of m),
3.80 (3H, s), 4.25−4.98 (2H, series of m), 5.14 (2H, s), 6.95
(2H, d, J = 9 Hz), 7.05−7.20 (2H, m), 7.30 (1H, m), 7.47
(2H, d, J = 9 Hz), 7.52 (1H, m).
AUTHOR INFORMATION
■
Corresponding Author
(2S,5R)-Methyl 5-[4-(2-Fluorobenzyloxy)phenyl]-
pyrrolidine-2-carboxylate p-Toluenesulfonate Salt (25). To
a stirred solution of 24 (83.5 kg) in methanol (334 L) was
added p-toluenesulfonic acid (45.8 kg). The reaction mixture
was warmed to 50 °C and stirred at this temperature for 3 h.
The mixture was cooled to 40 °C, and MTBE (1160 L) was
added while maintaining the temperature in the 35−40 °C
range. The mixture was stirred for a further 0.5 h at 35 °C,
cooled to between 0 and 5 °C, and stirred at this temperature
for 1 h. The resulting solid was filtered, washed with MTBE
(249 L) that had been chilled to about 5 °C, and dried in a
vacuum oven at 40 °C. The product 25 (89.8 kg, 92.1%) was a
white crystalline solid, mp 171 °C.
(2S,5R) 5-[4-(2-Fluorobenzyloxy)phenyl]pyrrolidine-2-car-
boxamide (9). To a stirred slurry of 25 (80 kg) in MTBE (400
L) was added water (76.4 L) and then 33% aqueous ammonia
solution (63.4 kg). The reaction was stirred for 15 min at 30
°C and allowed to settle for 15 min, and the lower aqueous
layer was removed. The MTBE layer was successively washed
with a solution of sodium carbonate (40 L) in water (400 L),
water (240 L), and then aqueous brine (260 kg). The MTBE
solution was concentrated under reduced pressure at 40 °C to
remove about 300 L of distillate, methanol (80 L) was added,
Gerard Giblin − Convergence Pharmaceuticals (a Biogen
Company), Cambridge CB22 3AT, U.K.;
Email: ged.m.giblin@gmail.com
Authors
Adrian Heseltine − Dr. Reddy’s Laboratories (EU) Limited,
Mirfield, West Yorkshire WF14 8HZ, U.K.
William Kiesman − Biogen Inc., Cambridge, Massachusetts
02142, United States
David MacPherson − Convergence Pharmaceuticals (a Biogen
Company), Cambridge CB22 3AT, U.K.
James Ramsden − Dr. Reddy’s Laboratories (EU) Limited,
Cambridge CB4 0PE, U.K
Ravi Vadali − Dr. Reddy’s Laboratories Ltd., Hyderabad
500049, Telangana, India
Michael Williams − Convergence Pharmaceuticals (a Biogen
Company), Cambridge CB22 3AT, U.K.
David Witty − Convergence Pharmaceuticals (a Biogen
Company), Cambridge CB22 3AT, U.K.; orcid.org/
0000-0002-5897-7370
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.oprd.0c00382
K
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
for selective monosubstitution of dibromoarenes. Tetrahedron Lett.
2001, 42, 4841−4844.
Notes
The authors declare the following competing financial
interest(s): D.W., D.M., and G.G. hold stock in Biogen Inc.
and were employees of Convergence Pharmaceuticals Ltd. and
Biogen Inc. at the time this work was conducted. Their
involvement ended after they left Biogen in 2017, except for
the creation of this article. W.K. is an employee of Biogen Inc.
(15) 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.
(16) Ace, K. W.; Armitage, M. A.; Bellingham, R. K.; Blackler, P. D.;
Ennis, D. S.; Hussein, N.; Lathbury, D. C.; Morgan, D. O.; O’Connor,
N.; Oakes, G. H.; Passey, S. C.; Powling, S. C. Development of an
Efficient and Stereoselective Manufacturing Route to Idoxifene. Org.
Process Res. Dev. 2001, 5, 479−490.
(17) MacPherson, D.; Witty, D.; Giblin, G.; Williams, M. T.; Walker,
D.; Kiesman, W.; Mack, T. Process for preparing alpha-carboxamide
pyrrolidine derivatives. WO2016102967A12015.
ACKNOWLEDGMENTS
■
We would like to acknowledge Julie Cornish and Joanne
O’Hare of Dr. Reddy’s Laboratories Limited at their Mirfield,
U.K., site for their analytical contributions and the chemistry
team at Dr. Reddy’s Laboratories Ltd., Hyderabad, India, for
laboratory investigations.
(18) He, Z.; Jamison, T. F. Continuous-flow synthesis of
functionalized phenols by aerobic oxidation of Grignard reagents.
Angew. Chem., Int. Ed. 2014, 53, 3353−3357.
(19) Li, B.; Berliner, M.; Buzon, R.; Chiu, C. K-F.; Colgan, S. T.;
Kaneko, T.; Keene, N.; Kissel, W.; Le, T.; Leeman, K. R.; Marquez,
B.; Morris, R.; Newell, R.; Wunderwald, S.; Witt, M.; Weaver, J.;
Zhang, Z.; Zhang, Z. Aqueous Phosphoric Acid as a Mild Reagent for
Deprotection of tert-Butyl Carbamates, Esters, and Ethers. J. Org.
Chem. 2006, 71, 9045−9050.
REFERENCES
■
(1) Curia, G.; Biagini, G.; Perucca, E.; Avoli, M. Lacosamide: a new
approach to target voltage-gated sodium currents in epileptic
disorders. CNS Drugs 2009, 23, 555−568.
(2) Wu, M.-T.; Huang, P.-Y.; Yen, C.-T.; Chen, C.-C.; Lee, M.-J. A
Novel SCN9A Mutation Responsible for Primary Erythromelalgia and
Is Resistant to the Treatment of Sodium Channel Blockers. PLoS One
2013, 8, No. e55212.
(3) Witty, D. R.; Alvaro, G.; Derjean, D.; Giblin, G. M. P.; Gunn, K.;
Large, C.; Macpherson, D. T.; Morisset, V.; Owen, D.; Palmer, J.;
Rugiero, F.; Tate, S.; Hinckley, C. A.; Naik, H. Discovery of
Vixotrigine: A Novel Use-Dependent Sodium Channel Blocker for the
Treatment of Trigeminal Neuralgia. ACS Med. Chem. Lett. 2020, 11,
1678−1687.
(4) Zakrzewska, J. M.; Palmer, J.; Morisset, V.; Giblin, G. M. P.;
Obermann, M.; Ettlin, D. A.; Cruccu, G.; Bendtsen, L.; Estacion, M.;
Derjean, D.; Waxman, S. G.; Layton, G.; Gunn, K.; Tate, S. Safety and
efficacy of a Nav1.7 selective sodium channel blocker in patients with
trigeminal neuralgia: a double-blind, placebo-controlled, randomised
withdrawal phase 2a trial. Lancet Neurol. 2017, 16, 291−300.
(5) Alvaro, G.; Bergauer, M.; Giovannini, R.; Profeta, R. Prolinamide
derivatives as sodium channel modulators. WO2007042239A12007.
́
(6) Najera, C.; Yus, M. Pyroglutamic acid: a versatile building block
in asymmetric synthesis. Tetrahedron: Asymmetry 1999, 10, 2245−
2303.
(7) Ezquerra, J.; Pedregal, C.; Rubio, A.; Valenciano, J.; et al.
General Method for the synthesis of 5-arylpyrrole-2-carboxylic acids.
Tetrahedron Lett. 1993, 34, 6317−6320.
́
(8) García-Lopez, J.-A.; Greaney, M. F. Synthesis of biaryls using
aryne intermediates. Chem. Soc. Rev. 2016, 45, 6766−6798.
(9) Yus, M.; Behloul, C.; Guijarro, D. In Formation of Biaryls by
Homocoupling of Grignard Reagents, 7th Electronic Conference on
Synthetic Organic Chemistry, 2003.
(10) Krasovskiy, A.; Knochel, P. A LiCl-Mediated Br/Mg Exchange
Reaction for the Preparation of Functionalized Aryl- and Hetero-
arylmagnesium Compounds from Organic Bromides. Angew. Chem.,
Int. Ed. 2004, 43, 3333−3336.
(11) Piller, F. M.; Metzger, A.; Schade, M. A.; Haag, B. A.;
Gavryushin, A.; Knochel, P. Preparation of Polyfunctional Arylmag-
nesium, Arylzinc, and Benzylic Zinc Reagents by Using Magnesium in
the Presence of LiCl. Chem. - Eur. J. 2009, 15, 7192−7202.
(12) Hawkins, J. M.; Dube, P.; Maloney, M. T.; Wei, L.; Ewing, M.;
Chesnut, S. M.; Denette, J. R.; Lillie, B. M.; Vaidyanathan, R.
Synthesis of an H3 Antagonist via Sequential One-Pot Additions of a
Magnesium Ate Complex and an Amine to a 1,4-Ketoester followed
by Carbonyl-Directed Fluoride Addition. Org. Process Res. Dev. 2012,
16, 1393−1403.
(13) Inoue, A.; Kitagawa, K.; Shinokubo, H.; Oshima, K. Selective
Halogen−Magnesium Exchange Reaction via Organomagnesium Ate
Complex. J. Org. Chem. 2001, 66, 4333−4339.
(14) Iida, T.; Wada, T.; Tomimoto, K.; Mase, T. Tributylmagnesium
ate complex-mediated novel bromine−magnesium exchange reaction
L
https://dx.doi.org/10.1021/acs.oprd.0c00382
Org. Process Res. Dev. XXXX, XXX, XXX−XXX