Development and manufacture of the inosine monophosphate dehydrogenase inhibitor merimepodib, VX-497

Article abstract of DOI:10.1021/op800060h

A process for the manufacture of merimepodib (VX-497), an inosine monophosphate dehydrogenase (IMPDH) inhibitor, has been developed and efficiently scaled to produce clinical supply. The process comprises five steps, incorporating simple and robust chemis

Full text of DOI:10.1021/op800060h

Organic Process Research & Development 2008, 12, 666–673
Development and Manufacture of the Inosine Monophosphate Dehydrogenase
Inhibitor Merimepodib, VX-497
Adam R. Looker,†,* Benjamin J. Littler,‡ Todd A. Blythe,† John R. Snoonian, Graham K. Ansell,§ Andrew D. Jones,¶
Phil Nyce,† Minzhang Chen,† and Bobbianna J. Neubert†
Chemical DeVelopment, Vertex Pharmaceuticals Incorporated, 130 WaVerly Street, Cambridge, Massachusetts 02139, U.S.A.
and 11010 Torreyana Road, San Diego, California 92121, U.S.A.
Abstract:
by way of the acyl imidazole 2. This intermediate was not
isolated; instead it carried forward into the coupling with phenyl
chloroformate to provide carbamate 5. Nitroaromatic 6 was
reduced by catalytic hydrogenation to yield the left-hand portion
of VX-497, aniline 7. The final stage involved the coupling of
5 and 7 under base-mediated conditions to form the urea
functionality, as well as purification by slurry via the trisolvent
mixture EtOH/acetone/H2O.
However, the analysis of the route revealed several undesir-
able features in preparation for commercial route selection.
These included: (a) capricious yield and purity obtained during
previous production of carbamate 5, (b) excess CDI which led
to largely insoluble and difficult to remove ureas, most notably
9 (Scheme 2), (c) compound 3 having physical properties
uncharacteristic of a robust raw material (semisolid at ambient
temperature and prone to decomposition), (d) need for high
catalyst loading in the reduction of 6, and (e) purification of
the API by slurry instead of a controlled crystallization.
This report shows the development of VX-497 chemistry
into a commercial process. The first section of this paper will
detail work that identified a strong raw material supply chain,
along with more robust chemistry. Next, the reaction parameter
screening process for the nitro reduction of 6 will be discussed.
To conclude, the work that went into the identification of a
reliable coupling and purification procedure, which proved a
highly robust process amenable to commercial-scale production,
will be presented.
A process for the manufacture of merimepodib (VX-497), an
inosine monophosphate dehydrogenase (IMPDH) inhibitor, has
been developed and efficiently scaled to produce clinical supply.
The process comprises five steps, incorporating simple and robust
chemistry that ultimately yielded 96.5 kg with a purity of 100%
(by HPLC analysis) and 99.7% w/w assay. Highlights of the
process are the effective use of production-scale phosgene, ma-
nipulation of Schotten-Baumann reaction conditions to give a low
pH procedure that avoids a critical impurity, and the use of online
tools to better identify parameters of the API purification.
1. Introduction
Merimepodib (VX-497)1 8 is an inhibitor of the inosine
monophosphate dehydrogenase (IMPDH) enzyme, providing
for the basis of its immunosuppressive activity.2
At Vertex Pharmaceutical’s Chemical Development Depart-
ment, the strategy for approaching incoming development
projects is two-fold: (a) advancing the medicinal chemistry route
towards a streamlined, scalable process enabling early-phase
programs, (b) preparing for later-phase work, including registra-
tion, validation, and commercialization. The initial route (often
referred to as the supply route) is characterized by quick delivery
and scalability, while later-phase work is focused on making
the process as robust and efficient as possible.
After early VX-497 development work, our supply route of
synthesis (Scheme 1)1,3 involved the coupling of (S)-3-hydroxy-
tetrahydrofuran (1) with 3-aminobenzylamine (3), using car-
bonyl diimidazole (CDI) to furnish the intermediate aniline 4
2. Results and Discussion
2.1. Making Use of Key Advantages of Phosgene. The
basis of a new development strategy was finding a more robust
starting material to replace the diamine 3. Several 3-nitrobenzyl
adducts were most appealing because they were commercially
available, offered stability, and were crystalline at ambient
temperature. Although using such a starting material would
require adding a reduction step to our synthetic sequence for
the nitro group, it afforded greater control over chemoselectivity
and introduced a stronger supply chain.
Several possible routes using 3-nitrobenzyl adducts were
explored, including two utilizing isocyanate 11 (Scheme 3).
Starting with 3-nitrobenzylchloride 10, treatment with KOCN
at elevated temperatures in polar aprotic solvents such as DMF
could generate the intermediate isocyanate. Alcohol 1 was added
to the reaction mixture, trapping the intermediate to form
nitrocarbamate 12. The carbamate formed in relatively good
conversion, but there was significant difficulty in isolating clean
product due to the typically poor purity profile of these reactions.
* Author for correspondence. E-mail: adam_looker@vrtx.com.
† Vertex Pharmaceuticals, Inc. Cambridge, MA.
‡ Vertex Pharmaceuticals, Inc. San Diego, CA.
Sepracor, Inc. Marlborough, MA.
§ Abbott Laboratories, Worcester, MA.
¶ Concert Pharmaceuticals, Inc. Lexington, MA.
(1) (a) Armistead, D. M.; Badia, M. C.; Bemis, G. W.; Bethiel, R. S.;
Frank, C. A.; Novak, P. M.; Ronkin, S. M.; Saunders, J. O. (Vertex
Pharmaceuticals, Inc.). U.S. Patent 6,344,465, 2002; Chem. Abstr.
2002, 136, 151157. (b) Armistead, D. M.; Badia, M. C.; Bemis, G. W.;
Bethiel, R. S.; Frank, C. A.; Novak, P. M.; Ronkin, S. M.; Saunders,
J. O. (Vertex Pharmaceuticals, Inc.). U.S. Patent 6,054,472, 2000;
Chem. Abstr. 2000, 132, 28897. (c) Armistead, D. M.; Badia, M. C.;
Bemis, G. W.; Bethiel, R. S.; Frank, C. A.; Novak, P. M.; Ronkin,
S. M.; Saunders, J. O. (Vertex Pharmaceuticals, Inc.). U.S. Patent
5,807,876; Chem. Abstr. 2002, 136, 151157.
(2) (a) Sorbera, L. A.; Silvestre, J. S.; Castan˜er, L. M. Drugs Future 2000,
25, 809. (b) Jain, J.; Almquist, S. J.; Shlyakhter, D.; Harding, M. W.
J. Pharm. Sci. 2001, 90, 625.
(3) Herr, R. J.; Fairfax, D. J.; Meckler, H.; Wilson, J. D. Org. Process
Res. DeV. 2002, 6, 677.
666
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development
10.1021/op800060h CCC: $40.75
2008 American Chemical Society
Published on Web 07/18/2008

Scheme 1. Original supply route
Scheme 2. Dimer from excess CDI
Scheme 3. Use of isocyanate intermediate
In addition, the thermal stability of 10 at elevated temperature
was of concern.4 Alternatively, attempts to prepare 11 through
a phosgenation reaction with commercially available 3-ni-
trobenzylamine HCl (13) using diphosgene at reflux in toluene
led to decomposition.
The use of a strategy similar to that of the supply route, but
using phosgene in the place of CDI to produce chloroformate
14, ultimately led to a plant-scale synthesis (Scheme 4). That
phosgene could be detected down to low ppm levels by GC
analysis using a diethylamine derivitization procedure was a
definite advantage to the synthesis.5 Additionally, excess phos-
gene could be removed by vacuum distillation if needed, an
excellent way to control the byproduct urea dimer 15. The best
conditions for the phosgenation of 1 required the use of 10 mol
% pyridine. Adding a stochiometric amount of base to the
reaction mixture allowed for the conversion requirement to be
met, but it decreased purity. Analytical results indicated that
the solution yield of 14 was >98%. The yield dropped to ∼95%
with a distillation cycle to remove unreacted phosgene; when
repeated to meet the phosgene specification of e20 ppm, the
yield further dropped to ∼92%. The major impurity was the
easily rejected carbonate 16 (Figure 1). Chloroformate 14 could
also be synthesized using the convenient phosgene substitute
triphosgene, as this had also been scaled up effectively.6
For the coupling of 14 with 3-nitrobenzylamine HCl (13),
initial investigations made use of nonaqueous conditions: Et3N
in EtOAc, using distilled chloroformate. These conditions
afforded a good yield of the product, but required initial free-
(4) Pitt, M. J.; Battle, L. A. Bretherick’s Handbook of ReactiVe Chemical
Hazards; Urben, P. G., Ed.; Butterworth-Heinemann: Woburn, MA,
1995; Vol. 1, p 878.
(5) Steger, J. L.; Coppedge, E. A.; Johnson, L. D. Proceedings of the
EPA/A&WMA International Symposium: Measurement of Toxic and
Related Air Pollutants; Research Triangle Park, NC, May, 1996.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
667

Scheme 4. Phosgene as CDI replacement
Scheme 5. Final conditions for 12
12 by the telescoped process in 87% yield (96% solution yield)
with a purity of 99.80%. The major impurity, as before, was
identified as 15. With these positive results in hand, we focused
efforts on the crystallization process. To this end, the parameters
of temperature, concentration, and hold times yielded the
following results: longer hold times and varied temperatures
showed no major effect; more dilute concentrations led to higher
rejection levels of 15. Encouraged by the success of the
recrystallization procedure mentioned earlier using isopropanol
to reduce 15 to very low levels, it was tried as a cosolvent.
Fortuitously, using low concentrations of isopropanol in toluene
gave excellent rejection while not sacrificing a significant
amount of product to the mother liquor. There were now two
ways to control the content of the critical impurity 15 in our
process (a) by removing unreacted phosgene through vacuum
distillation prior to the coupling and (b) by rejection to the
mother liquor in the crystallization process. Process robustness
was ensured by stress testing at every major operation. Tem-
perature of the aqueous washes post coupling was the only
factor found that needed to be controlled to a strict level: due
to the lower solubility of 12 in toluene, product crystallized
during the workup if a reaction temperature of 50 °C ( 10 °C
was not kept. Production efforts9 of this optimized step (Scheme
5) were realized in three batches with a total yield of 120.6 kg
(84%), and an average purity of 99.89% and 100.0% w/w assay
meeting all required specifications. Interestingly, the key
impurity 15 was not detected in any batch.
basing to achieve a clean purity profile. The main impurity
formed was urea dimer 15.7 To avoid the need for distillation
of 14, telescoping was investigated. Overall purity of carbamate
12 when using the crude chloroformate 14 was excellent,
generating only minor amounts of 15. For precautionary reasons,
a rework was explored in case this critical impurity could not
be removed to sufficient levels. A recrystallization from
isopropanol effectively removed the dimer below the limit of
detection of the analytical method and provided for good
recovery.
Figure 1. Major phosgenation impurity.
Hoping to avoid the extra process of desalting 13, the
Schotten-Baumann8 reaction was explored. Initial focus was
placed on acetate solvents, with isopropyl acetate (IPAc) having
emerged as the most promising candidate, providing for 12 in
good yield and purity using aqueous Na2CO3. The next step
was to determine if the conversion of 1 to 14 could be performed
in IPAc, thus allowing for an easy transition in the telescoped
process. Unfortunately, it was found that IPAc was not stable
under the highly acidic conditions: in situ, it formed isopropyl
chloroformate leading to newly observed impurity 17 upon
coupling (Figure 2).
2.2. Preparation of Mixed Bis-carbamate 5. The goal for
the plant production of 5 was to telescope the solution of
intermediate 4 produced after catalytic hydrogenation directly
into the coupling with phenyl chloroformate. A solubility screen
showed that 5 had relatively low solubility in IPAc, so the
coupling in EtOAc was swapped to IPAc in order to isolate 5.
Development of the catalytic hydrogenation of 12 proved
simple, and it was rapidly found that 4 could be reliably formed
under standard Pd/C-catalyzed conditions.
Figure 2. Isopropyl chloroformate adduct.
Both aqueous and nonaqueous reactions were evaluated for
the coupling reaction of 4 with phenyl chloroformate. Com-
pound 5 was obtained in ∼70% yield and >99.0% purity using
Et3N in EtOAc, but these anhydrous conditions were not
preferred because the reaction mixture was a thick slurry due
to the formation of Et3N HCl.
Returning to toluene for both the phosgenation chemistry
and the Schotten-Baumann coupling, initial reactions gauging
feasibility found this solvent to be highly successful in producing
(6) Greater than 150 kg of 14 was outsourced as the raw material for
later campaigns, prepared by the use of triphosgene.
(7) It has been shown in laboratory studies that 14 reacts with HCl to
give 1 and phosgene to a minor extent, which goes on to react in situ
to form 15. Thus, free-basing conditions were required to remove the
HCl.
(9) A phosgene generator utilizing carbon monoxide and chlorine was
used to supply scale-up requirements. The phosgene stream was
pumped directly into a 630 L glass-lined reactor used for the
phosgenation chemistry.
(8) (a) Schotten, C. Ber. 1884, 17, 2544. (b) Baumann, E. Ber. 1886, 19,
3218.
668
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Scheme 6. Optimized Procedure to 5
slowly, was of particular concern because slow addition would
be needed in the plant to control the reaction exotherm and the
rate of carbon dioxide off-gassing. The data indicated that 18
forms only under basic conditions when excess bicarbonate is
present (entries 2 and 5), but not under acidic conditions when
excess chloroformate is used (entries 4 and 7).
Entry 5 demonstrates the importance of performing stress
testing prior to going to the plant and fully understanding the
reaction mechanism of byproduct formation. Urea 18 did not
form for the first 4 h after the addition of phenyl chloroformate
but then formed at a steady rate. An investigation showed that
this window of stability correlated to excess unreacted phenyl
chloroformate initially preventing the formation of 18 by
reacting rapidly with any 4 liberated by basic hydrolysis
(Scheme 7).
Table 1. Effect of NaHCO3, Chloroformate and Dosing
Time on the Formation of 18
NaHCO3 PhOCOCl dosing time after end
entry (equiv)
(equiv)
time of dosing (h)
18 (%)
The production of relatively high levels of 18 under the 6 h
dosing of Table 1, entry 3, was due to the fact that the phenyl
chloroformate was added dropwise with a significant time
interval between individual drops. During the time between the
drops of reagent there was no phenyl chloroformate present to
convert 4 back to 5, and so 18 was formed. Sodium bicarbonate
could therefore not be used because any interruption in dosing
would lead to unacceptable levels of urea 18 and loss of the
batch.
The only viable solution was to perform the reaction under
acidic conditions at all times but to avoid the strongly acidic
conditions at the end of the reaction that hydrolyzed the ethyl
acetate and formed a single phase that could not be worked
up. After many buffer systems were considered, it was realized
that aqueous sodium sulfate (pH 6.5) should be ideal since the
reaction would be acidic throughout and the formation of
NaHSO4 would buffer the acidity at the end of the reaction.
An additional advantage was that sodium sulfate would not
produce any off-gassing, and thus pressure build-up in the
reactor would be avoided. In fact, impurity 18 was not produced
with an addition time up to 2 h. At the end of each reaction,
the pH of the aqueous phases was 0-1, but the phases separated
readily even after stirring for 18 h.
Confident the process was ready for the pilot plant, produc-
tion of 5 led to a total of 111.4 kg in three batches, correspond-
ing to a yield of 89.8% with an average purity of 100.0% and
99.8% w/w assay meeting all required specifications.
2.3. Synthesis of Penultimate Aniline 7. Development
work on the hydrogenation of nitrooxazole 6 focused on a
general parameter screening and investigation of the process.
Attention was first directed at exploring the solvent and Pd/C
catalyst loading, while keeping other variables constant (3 bar
H2, 40 °C).
1
2
3
4
5
0.95
1.10
1.05
0
1.10
10 min
10 min
6 h
1
24
1
24
0
not detected
not detected
0.11
0.95
1.05
1.05
1.10
9.08
0.14
0.15
14
1
10 min
10 min
not detected
not detected
not detected
not detected
0.04
16a
1
1.20
4
6
20
2
18
1.5
18a
0.23
6
7
1.05
1.05
1.10
1.15
10 min
10 min
not detected
not detected
not detected
not detected
a Homogeneous reaction mixture.
Schotten-Baumann conditions using a biphasic mixture of
EtOAc and aqueous Na2CO3 were superior to the anhydrous
conditions due to complete dissolution of the solids throughout
the process and simpler workup procedures. Scheme 6 shows
the initial set of reaction conditions from 12 to 5 that produced
a >90% yield on a 70 g scale.
Stress testing of the reaction mixture, performed as a prelude
to running the reaction in the plant, showed that prolonged
exposure of the product mixture to these basic conditions formed
urea 18: a byproduct known from the initial supply route to be
difficult to remove (Figure 3).
Figure 3. Dimeric by-product.
EtOAc and IPAc proved to be excellent solvents for
conversion, but neither was acceptable for isolation due to
relatively high solubility. Even with a low temperature isolation
(-10 °C), the yield of 7 from IPAc was still only ∼60%. This
could be increased, though, to >90% upon solvent exchange
to heptane, making sure that the final solvent ratio did not
exceed 10% IPAc.
Of significant interest was the need for a suspiciously high
catalyst loading to achieve good purity. At very high loading
(20% w/w Pd/C, dry basis), the reaction performs admirably.
A comparison of aqueous sodium carbonate and sodium
bicarbonate in the Schotten-Baumann reaction clearly showed
the sodium carbonate system did not allow for enough of a
time window following complete conversion with respect to
the formation of impurity 18. This result prompted replacement
of sodium carbonate with sodium bicarbonate.
An investigation with the purpose of ensuring a robust
process by examining the effect of adding different amounts
of base and chloroformate is summarized in Table 1. Of these
data, entry 3, where the phenyl chloroformate was added very
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
669

Scheme 7. Proposed pathway to the formation of 18
Scheme 8. Proposed pathway of 6 to 20
The manufacture of 7 was performed over two stages. The
first stage utilized nitrooxazole 6 that was already available and
required the need for pretreatment with activated charcoal. The
first batch of this stage ran into immediate problems due to the
in-line filtration setup for the charcoal. The filtration ran from
the bottom of the vessel, through a filter bag, and back into the
top of the vessel. Because of the inefficiency of this filtration,
not all the charcoal was removed from the batch. When the
hydrogenation was performed, the reaction time exceeded the
laboratory protocol considerably (the in-process control (IPC)
specification was met, but was out of trend with respect to
purity). After a laboratory investigation, isolation with one less
solvent exchange was undertaken to increase the amount of
IPAc during crystallization and therefore give the best chance
for impurity removal. In fact, adequate quality material was
obtained in 97.9% and 98.3% w/w assay. The yield though was
lower due to the higher IPAc content in the liquor (81.5% yield).
For the next three batches, the in-line filtration was performed
into a receiving tank, thus ensuring that the filtrate was free of
charcoal. These batches ran smoother while providing quality
material in >99.5% purity and 89% yield. Upon the receipt of
newer lots of 6 that had been pretreated by our suppliers, this
material processed exceptionally in four further batches to give
high quality product in excellent yield (100% purity and 87.6%
yield). All future campaigns will use material pretreated with
charcoal at the vendor, allowing for smoother cGMP processing.
A further understanding of this reduction using React-IR and
other online techniques has led to a more robust and cost-
effective process. Such work will be reported in due course.
2.4. Coupling and Recrystallization to Provide Merime-
podib. The final bond-forming step in the merimepodib (VX-
497) process did not require significant changes from the supply
route, as a robust procedure with simple heating of 7 and 5 in
the presence of base and EtOAc provided for title compound 8
in >90% yield, albeit with the purity lower than specification
(>98%). The process itself was straightforward. During heating,
all materials went into solution and as the reaction progressed,
product crystallized. Cooling and filtration afforded the product
in excellent yield. Purification, however, proved problematic.
Knowing that API solubility is poor in nearly all common
organic solvents, a slurry method originally utilized three
Yet, at <6% there is significant purity reduction. The impurities
consisted of the hydroxylamine, nitroso, azobenzene dimer, and
to a large extent, the azoxybenzene dimer.10 However, when
the reference standard of 6 was hydrogenated using only 3%
catalyst, the reaction performed to give much higher quality
material. Extensive experimentation did not identify the cause
of this difference between the reference standard and other lots,
but it was found that pretreatment with activated charcoal
allowed for acceptable processing of 6. In addition, it was noted
that 6 decomposed in solution while in the presence of light
(an 80% conversion of 6 to 20 occurred in merely 24 h (Scheme
8)). Literature precedence was found for this oxidative frag-
mentation.11 It was proposed that a net 4 + 2 addition of oxygen,
followed by rearrangement and then hydrolysis was the likely
pathway, similar to what Scarpati and co-workers described.
Also noted was a significant rate difference in this oxidative
fragmentation between the reference standard (significantly
slower) and batches known to reduce poorly under the
hydrogenation conditions. Although no direct evidence was
found, the unknown impurity that likely poisoned the catalyst
could also play a role in promoting such decomposition as well.
This charcoal treatment was instituted into the campaign and
the future raw material suppliers were notified of these results
in an effort to have lots of 6 delivered that met a use test
specification. This allowed for a convenient Pd/C loading of
4.5% w/w (dry basis). Finally, after a screening of pressure and
temperature was performed, suitable conditions were found that
used EtOAc as solvent, 4.5% w/w (dry basis) of 5% Pd/C, 7
bar H2 pressure, at 40 °C.
(10) For excellent in-depth reading into nitroaromatic hydrogenation
chemistry, please see: Augustine, R. L. Heterogeneous Catalysis for
the Synthetic Organic Chemist; Marcel Dekker, Inc.: New York, 1996;
p 473.
(11) Lesce, R. M.; Graziano, L. M.; Cimminiello, G.; Cermola, F.; Parrilli,
M.; Scarpati, R. J. Chem. Soc., Perkin Trans. 2 1991, 7, 1085.
670
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

for NMP is 530 ppm.12,13 In spite of RC1 recrystallization
studies and extensive evaluation of washing conditions and
drying temperature, the NMP level could not be reduced below
700 ppm.
Microscope photos of the solids produced by this procedure
showed areas of crystal agglomeration that were thought to trap
NMP so that it could not be removed by washing or prolonged
drying under vacuum. Agglomerates were thought to form
because the solution was highly supersaturated at the time of
the primary nucleation event. A more controlled and gradual
crystallization was thus developed to favor crystal growth over
nucleation during the primary crystallization and to prevent
entrapment of NMP. An experiment was performed that reduced
the initial methanol charge at 50 °C from 8.4 vol to 4.05 vol in
order to reach the cloud point with the minimum volume of
antisolvent. Seeds of 8 were added, and the mixture was stirred
for 1 h before the remaining 4.35 vol of methanol was added.
The mixture was cooled and filtered to afford API with 380
ppm NMP. Microscope images revealed well-defined needles
with no agglomeration. Washing was again revisited with this
procedure in an attempt to lower the NMP levels even more,
but once again, the washing proved to have little or no effect.
This study confirmed that low levels of NMP would be
produced by a carefully controlled crystallization.
At this stage in the studies, a focused beam reflectance
measurement (FBRM available from Mettler Toledo) particle
size monitor became available, allowing for key variations of
the NMP/MeOH recrystallization to be re-examined. The
FBRM data is presented in Figure 5. The x-axis is shown in
minutes (a scan was recorded every 15 s). The y-axis is the
number of particle counts per second. The graphs show four
groupings of crystal size based on particle chord length: 1-5
µm (red); 10-23 µm (yellow); 29-86 µm (green); 100-251
µm (blue); 292-1000 µm (violet). Since the crystals of 8 are
needles, the measured chord length is highly dependent on the
crystal orientation, so the data are semiquantitative.
The data presented show that the initial procedure where
8.4 vol of methanol is added in a single portion leads to a single
nucleation-controlled crystallization event (Figure 5a). The data
in b and c of Figure 5 show that adding either crude API (Figure
5b) or recrystallized API (Figure 5c) as seeds to a mixture with
4.05 vol MeOH leads to slow crystal growth with only a minor
difference in the final crystal size distribution. Crude seeds were
preferred for the initial plant campaign because recrystallized
seeds would have to be prepared in a separate operation,
whereas crude material will be directly available from the
coupling. The crystal growth phase is particularly well illustrated
in Figure 5b where the number of counts for the fine particles
(red and yellow lines) decreases in the immediate period after
seeding, while the number of counts for the larger particles
increases in this same time frame. The API was obtained with
NMP levels (200-250 ppm) well within the ICH limit for the
two crystallizations performed at low supersaturation.
Figure 4. VX-497 solubility in NMP/MeOH.
solvents (EtOH, acetone, H2O) as a quick way to obtain
acceptable purity. Most of the development work was focused
on the purification (as described in detail below), but some work
did aim at identifying the optimal conditions for the coupling
procedure.
Optimization of the coupling conditions made use of DoE
and screening tools, through which a slight excess of 7 (1.07
equiv with respect to 5) was recognized as most advantageous.
EtOAc was selected as the best solvent for the coupling, but
with a higher dilution of 15 vol (with respect to the original
conditions of 8 vol) that alleviated stirring problems and gave
more consistent reaction times. In a similar manner, diisopro-
pylethylamine was chosen as base.
A true recrystallization procedure was sought to ensure that
the isolated API consistently met specification and allowed the
batch to be polish filtered in the final step of the process,
removing any accumulated inorganic material. An initial solvent
screen of merimepodib 8 showed that only highly polar aprotic
solvents (such as NMP, DMF, and DMSO) would be suitable
for a polish filtration. A screen of solvents and antisolvents
indicated that mixtures of NMP (solvent) and methanol (anti-
solvent) were best able to reject the main impurity, urea 18.
This system also provided for an effective filtration, whereas
othersystemswithNMPshowedverypoorfiltrationcharacteristics.
A solubility curve was determined for the NMP/MeOH
system (Figure 4), and a ratio of ∼70% MeOH was chosen as
the target final concentration for further optimization based on
potential recovery. The initial recrystallization procedure charged
an NMP solution of 8 into a reactor containing MeOH heated
to 65 °C, but this approach led to an uncontrolled crystallization
event, as well as being operationally difficult to safely control.
An improved procedure dissolved 8 in NMP (5.6 vol) and polish
filtered at room temperature, then heated the filtrate to 50 °C.
Methanol (8.4 vol) was added and the mixture seeded with 8.
At this point, the mixture was a thin, turbid slurry, but no defined
solids could be observed. The mixture was cooled to 20 °C
over 6 h, and additional MeOH (2 vol) was added. The mixture
was then cooled to 0 °C over 4 h and filtered. This provided
excellent purity API (typically 99.9%) in >85% yield, but the
level of residual NMP remaining was variable (typically
800-2000 ppm) and higher than ideal: the ICH option 1 limit
(12) ICH Harmonized Tripartite Guideline Q3C (R3), revised November
2005, available at http://www.ich.org.
(13) Option #1 (530 ppm for NMP) of the ICH Q3C guideline on residual
solvents is based on the assumption of a 10 g daily dose. Option #2
(see section 3.3) allows for the calculation based on actual daily dose.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
671

corresponding to a yield of 78.4% over two steps with an
average purity of 100.0% and 99.7% w/w assay meeting all
required specifications. The average NMP level was 633 ppm.
3. Conclusion
A five-step process was developed for the immunosuppres-
sive drug candidate merimepodib (VX-497, 8). Through the
understanding of critical mechanistic pathways, insight into
impurity formation/rejection, and the use of powerful online
tools to better gauge process variables, a highly robust synthesis
was devised.
4. Experimental Section
4.1. General. HPLC conditions: Analyses were performed
using an Atlantis dC18 100 mm × 4.6 mm, 3 µm column (Part
#186001337). The mobile phase consisted of a gradient method
using A: 0.025% (v/v) phosphoric acid (85%)/water and B:
acetonitrile. Assays were determined using high purity reference
standards. NMR spectra were collected using a Bruker 499.87
MHz spectrometer. Solvents and reagents were obtained from
commercial sources and used as is without further purification,
unless otherwise noted. All reactors are standard multipurpose
equipment, either glass-lined or stainless steel. All reactions were
carried out under an atmosphere of nitrogen.
4.2. Carbamic acid [(3-nitrophenyl)methyl]-(3S)-tetrahy-
dro-3-furanyl ester (12). 1 (20.5 kg, 232.7 mol) was charged
to a glass-lined reactor, followed by toluene (174 kg). Pyridine
(1.82 kg, 23.3 mol) was charged to the system. Phosgene (43.2
kg, 436.8 mol) was charged to the reactor while maintaining a
temperature of 20 °C ( 5.0 °C. The reaction was warmed to
25 °C ( 5.0 °C, and stirred for 6 h. 101 kg of solvent was
removed by distillation (20 mbar), followed by the addition of
toluene (72 kg). Distillation removed 76 kg of solvent (20 mbar).
IPC confirms residual phosgene content in solution to be 4 ppm.
The reaction mixture was filtered, and the reactor and cake were
rinsed with toluene (68 kg). To a separate glass-lined reactor
was charged Na2CO3 (23.4 kg, 220.8 mol), followed by water
(145 kg). This was stirred at 25 °C ( 5.0 °C to attain full
dissolution. Charged to this solution was 13 (34.0 kg, 180.3
mol), followed by toluene (262 L). The reaction was heated to
50 °C ( 5.0 °C. To this solution was charged the toluene
solution of 14 previously produced, while maintaining a
temperature of 50 °C ( 5.0 °C. The reaction was stirred at this
temperature for 1 h and sampled for completion. The phases
were separated at 50 °C ( 5.0 °C, and the organic phase (top
layer) was washed with 1 M HCl (41.4 kg) at this same
temperature. The phases were separated at 50 °C ( 5.0 °C,
and the organic phase (top layer) was polish filtered into a
second glass-lined reactor. The solution was distilled to remove
254 kg of solvent at <100 mbar. The reaction was cooled to
40 °C ( 5.0 °C, and isopropanol (25 L) was charged. The slurry
was cooled to 0 °C ( 5.0 °C and stirred 1 h at this temperature.
The product was filtered, washed with toluene/isopropanol (32
L: 8 L), and dried under reduced pressure (40 mbar) at 50 °C
( 5.0 °C. The title compound (41.39 kg, 86.7%) was isolated
as a white solid (99.78% purity, 99.9% w/w assay): 1H NMR
(d6-DMSO): 8.10 (m, 2H); 7.90 (m, 1H); 7.70 (d, 1H); 7.62 (t,
1H); 5.10 (br, 1H); 4.30 (d, 2H); 3.72 (m, 4H); 2.10 (m, 1H);
1.87 (m, 1H).
Figure 5. FBRM data of the crystallization: (a) 8.4 vol MeOH
added in single portion; (b) 4.05 vol MeOH with crude API
seeds; (c) 4.05 vol MeOH with rxd API seeds.
The research performed up to this point in identifying
optimal recrystallization conditions made use of 8 that was of
very high purity (since representative crude API was unavailable
at the time). When the optimized procedure was then applied
to representative crude API, excellent purity was attained, but
the levels of NMP were higher than anticipated (∼600 ppm).
Presumably the trace impurities that were now present in the
test article led to a less perfect crystal lattice resulting in more
NMP becoming entrapped.14 The decision was made to go
forward into production with this procedure, as an acceptance
limit of 1500 ppm was easily justified by the predicted clinical
dose and using option #2 in the ICH guideline for residual
solvents.13
The production of API performed very smoothly in the plant.
A total of 96.5 kg of 8 was produced in four batches,
(14) (a) Mullin, J. W. Crystallization, 3rd ed.; Butterworth-Heinemann Ltd.:
Oxford, 1993; p 172. (b) Okamoto, M.; Hamano, M.; Igarashi, K.;
Ooshima, H. J. Chem. Eng. Jpn. 2004, 37, 1224.
672
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

4.3. Carbamic Acid [[3-[(Phenoxycarbonyl)amino]phen-
yl]methyl]-(3S)-tetrahydro-3-furanyl ester (5). 12 (31.8 kg,
119.4 mol) was charged to a stainless-steel hydrogenator,
followed by 5% Pd/C (50% wet, 0.5 kg, 0.75% dry basis).
Inerting with nitrogen was performed to 4 bar. EtOAc (254 L)
was charged, and inerting with nitrogen performed again to 4
bar. The reaction was stirred to a setpoint of 25 °C ( 5.0 °C.
The reactor was pressurized to 6 bar with hydrogen gas, and
stirred at 25 °C ( 5.0 °C for 7 h. After removal of hydrogen
pressure, and inerting with nitrogen, the reaction slurry was
filtered in-line to a receiver with the aid of an EtOAc (64 L)
rinse. To a separate glass-lined reactor was charged water (223.0
kg), followed by anhydrous Na2SO4 (37.4 kg, 263.3 mol). This
mixture was stirred at 30 °C ( 5.0 °C to attain complete
dissolution. The solution of 4 previously made was charged to
the solution. The temperature was increased to 50 °C ( 5.0
°C, and phenyl chloroformate (20.9 kg, 133.5 mol) was charged
in a continuous fashion over 30 min while maintaining a
temperature of 50 °C ( 5.0 °C. The reaction was stirred at this
temperature for 4 h, and sampled for completion. The phases
were separated at 50 °C ( 5.0 °C, and the organic phase (top
layer) washed with water (127.5 kg) at this same temperature.
The phases were separated at 50 °C ( 5.0 °C, and the organic
phase (top layer) was distilled to remove 240 kg of solvent at
<50 °C. Isopropyl acetate (276.2 kg) was charged to the reactor,
and this mixture was distilled to remove 277 kg of solvent at
<50 °C. Isopropyl acetate (276.3 kg) was charged to the reactor,
and this mixture was distilled to remove 280 kg of solvent at
<50 °C. The resulting suspension was cooled to 15 °C ( 5.0
°C and stirred for 30 min at this temperature. The product was
filtered, washed twice with isopropyl acetate (55.1 kg, 55.3 kg),
and dried under reduced pressure (20 mbar) at 45 °C ( 5.0
°C. The title compound (39.2 kg, 91.8%) was isolated as a white
solid (100.0% purity, 99.8% w/w assay): 1H NMR (d6-DMSO):
10.20 (br, 1H); 7.85 (br, 1H); 7.40 (m, 4H); 7.25 (t, 2H); 7.20
(d, 1H); 6.95 (m, 1H); 6.75 (m, 1H); 5.15 (m, 1H); 4.15 (d,
2H); 3.75 (m, 2H); 3.70 (m, 2H); 2.10 (m, 1H); 1.90 (m, 1H).
4.4. Benzenamine 3-Methoxy-4-(5-oxazolyl) (7). 6 (9.98
kg, 45.3 mol) was charged to a stainless-steel hydrogenator,
followed by 5% Pd/C (50% wet, 0.91 kg, 4.5% dry basis).
Inerting with nitrogen was performed to 3.5 bar. IPAc (87.1
kg) was charged, and inerting with nitrogen performed again
to 3.5 bar. The reaction was stirred to a setpoint of 40 °C (
5.0 °C. The reactor was pressurized to 7 bar with hydrogen
gas, and stirred at 40 °C ( 5.0 °C for 5 h. After removal of
hydrogen pressure, and inerting with nitrogen, the reaction slurry
was cooled to 20 °C ( 5.0 °C. The reaction slurry was filtered
in-line to a receiver with the aid of an IPAc (18.2 kg) rinse.
The solution was distilled to remove 90 L of solvent at <40
°C. Heptane (68.0 kg) was charged to the reactor, and this
mixture was distilled to remove 94 L of solvent at <40 °C.
Heptane (68.0 kg) was charged to the reactor, and this mixture
was distilled to remove 94 L of solvent at <40 °C. The resulting
suspension was cooled to 20 °C ( 5.0 °C and stirred for 1 h at
this temperature. The product was filtered, washed with heptane
(15.1 kg), and dried under reduced pressure (15 mbar) at 45
°C ( 5.0 °C. The title compound (7.27 kg, 84.4%) was isolated
as a yellow solid (100.0% purity, 99.8% w/w assay): 1H NMR
(d6-DMSO): 8.20 (s, 1H); 7.32 (d, 1H); 7.15 (s, 1H); 6.31 (s,
1H); 6.25 (d, 1H); 5.50 (s, 2H); 3.80 (s, 3H).
4.5. CarbamicAcid,N-[[3-[[[[3-Methoxy-4-(5-oxazolyl)phen-
yl]amino]carbonyl]amino]phenyl]methyl]-(3S)-tetrahydro-
3-furanyl Ester (8). 5 (34.09 kg, 95.7 mol) and 7 (19.50 kg,
102.5 mol) were charged to a glass-lined reactor. EtOAc (461
L) was charged, followed by diisopropylethylamine (12.3 kg,
95.2 mol). The lines were rinsed with EtOAc (50 L). The
reaction was stirred at reflux (76 °C ( 2.5 °C) for 13.5 h. The
resulting suspension was cooled to 20 °C ( 5 °C and stirred
for 1 h at this temperature. The product was filtered, washed
with EtOAc three times (133 L each), and dried under reduced
pressure (<20 mbar) at 50 °C ( 5.0 °C. The crude title
compound (40.2 kg, 91.4%) was isolated as an off-white solid
(98.7% purity, 98.3% w/w assay). This material was purified
as follows: Crude 8 (28.00 kg, 61.9 mol) was charged to a glass-
lined reactor, followed by NMP (147 L). The reaction was
stirred at 20 °C ( 5.0 °C for 0.5 h to attain full dissolution.
This solution was polish filtered through a 1.0 µm filter to a
second glass-lined reactor. This solution was heated to 51 °C
( 5.0 °C. MeOH (99.7 kg) was charged to the solution
maintaining a temperature of 51 °C ( 5.0 °C. The temperature
was reduced to 48 °C ( 2.0 °C. The reaction was seeded with
a slurry of crude 8 (280 g in 1300 g MeOH), followed by a
rinse with MeOH (700 g). This slurry was stirred at 48 °C (
2.0 °C for 15 min. The slurry was cooled to 45 °C ( 5.0 °C,
and MeOH (130.7 kg) charged over 1.5 h. The resultant
suspension was cooled linearly to 0 °C ( 5.0 °C over 5 h and
held at this temperature for 1 h. The product was filtered,
washed with MeOH two times (280 L each), and dried under
reduced pressure (<20 mbar) at 50 °C ( 5.0 °C. The
recrystallized title compound (25.6 kg, 92.2%) was isolated as
a white solid (100.0% purity, 99.8% w/w assay): 1H NMR (d6-
DMSO): 8.90 (br, 1H); 8.75 (br, 1H); 8.35 (s, 1H); 7.75 (t,
1H); 7.60 (d, 1H); 7.50 (d, 1H); 7.41 (s, 1H); 7.38 (m, 1H);
7.33 (m, 1H); 7.25 (t, 1H); 7.05 (d, 1H); 6.85 (d, 1H); 5.15 (m,
1H); 4.15 (d, 2H); 3.90 (s, 3H); 3.77 (m, 2H); 3.70 (m, 2H);
2.10 (m, 1H); 1.90 (m, 1H).
Acknowledgment
We acknowledge and thank Dave Turnquist, Chris Sander-
son, and Amy Zhang currently, and formerly of Vertex
Pharmaceuticals, Inc. for critical analytical support and guid-
ance. We also thank Sara Dilliplane for her help in preparing
the manuscript.
Received for review March 15, 2008.
OP800060H
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
673