Development of a practical synthesis of a pyrazolopyridinone-based p38 MAP kinase inhibitor

Article abstract of DOI:10.1021/op100205s

A practical synthesis of the pyrazolopyridinone-based p38 MAP kinase inhibitor (4) was required for an ongoing program. The synthesis of a key pyrazolopyridinone building block was refined and optimized to provide kilogram quantities of 10 without chromat

Full text of DOI:10.1021/op100205s

Organic Process Research & Development 2011, 15, 31–43
Development of a Practical Synthesis of a Pyrazolopyridinone-Based p38 MAP
Kinase Inhibitor
Robert R. Milburn,* Oliver R. Thiel, Michal Achmatowicz, Xiang Wang, Jamie Zigterman, Charles Bernard, John T. Colyer,
Evan DiVirgilio, Rich Crockett, Tiffany L. Correll, Karthik Nagapudi, Kumar Ranganathan, Simon J. Hedley, Alan Allgeier,
and Robert D. Larsen
Amgen Inc., One Amgen Center DriVe, Thousand Oaks, California, 91362, United States
Abstract:
consideration at the outset of process development.4 Herein we
describe the route scouting and development that led to a
suitable synthesis for their large-scale preparation.
A practical synthesis of the pyrazolopyridinone-based p38 MAP
kinase inhibitor (4) was required for an ongoing program. The
synthesis of a key pyrazolopyridinone building block was refined
and optimized to provide kilogram quantities of 10 without
chromatography or extractive workups. An efficient building-block
strategy was employed to give optimal control of the key quality
attributes, and in situ Raman spectroscopy was used to monitor
and understand the complex solid-state properties of 4.
The initial route to inhibitors 3 and 4 is shown in Scheme
2. The strategy to assemble the pyrazolopyridinone core 10
entailed fusing a pyridinone ring onto a pyrazoloaminoaldehyde
(7) using a Knoevenagel-type transformation, with requisite 7
accessed Via condensation of arylhydrazine 5 with 2-ethoxym-
ethylene malononitrile. Fragment assembly involved coupling
of the aryl boronate 11 with 10 under Suzuki coupling
conditions. The established chemistry was efficient, robust, and
well suited to further development. The end-game, however,
was not scalable as Pd-mediated steps present considerable
challenges in the form of Pd removal and impurity rejection
and are best carried out earlier in the sequence to allow
additional isolation points for impurity control.
An effective solution to the problems associated with cross-
coupling reactions was the building-block approach shown in
Scheme 3.5 While inclusion of the amidation step as the final
chemical transformation decreases the overall convergence, the
amidation reaction does not introduce significant process related
impurities and its incorporation provides an additional control
point to deal with the cross-coupling derived impurities. To deal
with the challenging physical properties of 4, an additional step
was included to control the final form of the drug substance.
The Suzuki coupling partners 10 and 13 were designated as
the key intermediates, leveraging existing technology to access
boronic acid 13 on scale.5a Although the known synthetic
sequence to 10 was lengthy, it was adopted for scale-up as
exploratory studies failed to yield a shorter, more convergent
route. Along with gaining an in-depth understanding and
optimization of each step, some noteworthy challenges presented
by the existing route to 10 include the following: the large
solvent volumes required for work-ups due to poorly soluble
intermediates, an insufficient understanding of chemoselectivity
during hydrogenation of nitrile 6, the use of excess brominating
reagent required to access 9, and the use of sodium hydride
combined with an undesirable solvent mixture. The optimized
process route to 10 is shown in Scheme 4.
Introduction
p38 Mitogen-activated protein (MAP) kinases are intracel-
lular serine/threonine kinases that positively regulate the produc-
tion and action of several pro-inflammatory mediators, specif-
ically the release of tumor necrosis factor-R (TNF-R) and
interleukin-1 (IL-1),1 and are involved in disease states such as
rheumatoid arthritis (RA), Crohn’s disease (inflammatory bowel
disease), and psoriasis. Biological agents that sequester TNF-
R, such as Etanercept, Infliximab, and Adalimumab, show
impressive clinical efficacy in the treatment of these diseases.
The potential for small-molecule inhibitors of p38 MAP kinase
as alternatives for these biological agents has spawned signifi-
cant activity in the area, resulting in the discovery of a number
of selective inhibitors of p38 that have shown great promise in
early clinical trials.2 As part of our ongoing effort in this area,
the Medicinal Chemistry team has been pursuing inhibitors of
p38 MAP kinase, manifesting in a number of suitable phthala-
zine-based candidates 1 and 2 (Scheme 1).3 Most recently, p38
inhibitors based on a pyrazolopyridinone scaffold have been
investigated, and compounds 3 and 4 were identified as
candidates for further development. While compound 4 was
ultimately selected for development, both 3 and 4 were in
* Author for correspondence. Telephone: (805) 313 5282. Fax: 805-375-4531.
E-mail: rmilburn@amgen.com.
(1) (a) Chen, Z.; Gibson, T. B.; Robinson, F.; Silvestro, L.; Pearson, G.;
Xu, B.; Wright, A.; Vanderbilt, C.; Cobb, M. H. Chem. ReV. 2001,
101, 2449. (b) Raingeaud, J.; Gupta, S.; Rogers, J. S.; Dickens, M.;
Han, J.; Ulevitch, R. J.; Davis, R. J. J. Biol. Chem. 1995, 270, 7420.
(2) (a) Gaestal, M.; Mengel, A.; Bothe, U.; Asadullah, K. Curr. Med.
Chem. 2007, 14, 2214. (b) Peifer, C.; Wagner, G.; Laufer, S. A. Curr.
Top. Med. Chem. 2006, 6, 113. (c) Margutti, S.; Laufer, S. A.
ChemMedChem 2007, 2, 1116. (d) Schett, G.; Zwerina, J.; Firestein,
G. Ann. Rheum. Dis. 2008, 67, 909.
(4) Pettus, L. H.; Wurz, R. P.; Xu, S.; Herberich, B.; Henkle, B.; Liu, Q.;
McBride, H. J.; Mu, S.; Plant, M. H.; Saris, C. J. M.; Sherman, L.;
Wong, L. M.; Chmait, S.; Lee, M. R.; Mohr, C.; Hsieh, F.; Tasker,
A. S. J. Med. Chem. 2010, 53, 2973.
(3) Herberich, B.; Cao, G.-Q.; Chakrabarti, P. P.; Falsey, J. R.; Pettus,
L.; Rzasa, R. M.; Reed, A. B.; Reichelt, A.; Sham, K.; Thaman, M.;
Wurz, R. P.; Xu, S.; Zhang, D.; Hsieh, F.; Lee, M. R.; Syed, R.; Li,
V.; Grosfeld, D.; Plant, M. H.; Henkle, B.; Sherman, L.; Middleton,
S.; Wong, L. M.; Tasker, A. S. J. Med. Chem. 2008, 51, 6271.
(5) (a) Achmatowicz, M.; Thiel, O. R.; Wheeler, P.; Bernard, C.; Huang,
J.; Larsen, R. D.; Faul, M. M. J. Org. Chem. 2009, 74, 795. (b) Thiel,
O. R.; Achmatowicz, M.; Bernard, C.; Wheeler, P.; Savarin, C.;
Correll, T.; Kasparian, A.; Allgeier, A.; Bartberger, M. D.; Tan, H.;
Larsen, R. D. Org. Process Res. DeV. 2009, 13, 230.
10.1021/op100205s  2011 American Chemical Society
Published on Web 11/16/2010
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
31

Scheme 1. Selected p38 MAP kinase inhibitors from Amgen
Scheme 2. Initial route to 3 and 4
Results and Discussion
could be observed by LC/MS. A solvent screen showed that
the problems associated with the addition rate of triethylamine
could be mitigated by performing the reaction in acetic acid.
Workup of the reaction was also facilitated by using acetic acid,
as neutralization with aqueous sodium hydroxide to pH 6-7
induced crystallization of 6a as a filterable solid. Application
of this protocol on kilogram scale afforded 6a as a pale yellow
solid in 90% yield (96 wt %).
Expectations that development work carried out on the 2,4-
difluorophenyl system could be directly applied to the 2,6-
difluorophenyl system were quickly proven naive. The first
indication of this came during sourcing of 5b, which was
generated via a nonscalable route that lead to long delays in
sourcing kilogram quantities. The stability of 5a and 5b also
contrasted sharply; whereas 5a was a bench-stable solid over
weeks of storage at ambient temperature, 5b deliquesced and
became highly colored within days. An explanation for the
differing stabilities of 5a/b was not investigated further;
however, N2 elimination accompanied by the loss of HF has
been reported for related systems (Scheme 7).8
Until intermediate 9, the development of both regioisomers
was pursued. Development was initiated on the 2,4-series, and
the resulting conditions were ported to the 2,6-series with further
optimization. When 4 was identified as the final candidate,
development efforts focused exclusively on the 2,6-difluoro
series.
Synthesis of Building Block 10. The first step in the
synthesis of 10a involved condensation of 5a with 2-ethoxym-
ethylene malononitrile (Scheme 5).6 The reaction occurs in two
stages, a rapid condensation of the hydrazine and 2-ethoxym-
ethylene malononitrile7 followed by a much slower cyclization
of 18 to the aminopyrazole. The initial approach to this reaction
involved addition of triethylamine to a suspension of 2-ethoxym-
ethylene malononitrile and 5a in ethanol. A detailed study of
the reaction revealed it to be extremely sensitive to the rate of
triethylamine addition, whereas slow triethylamine addition (>30
min) gave a bright-yellow mixture from which 6a crystallized,
rapid triethylamine addition (<1 min) gave a dark-brown slurry
from which no product could be isolated, and 19 (Scheme 6)
(6) Cheng, C.; Robins, R. J. Org. Chem. 1956, 21, 1240.
(7) 18 may be observed Via HPLC. Isolated 18 is a crystalline solid that
slowly cyclizes to 6a in the solid state.
(8) (a) Collins, I.; Roberts, S. M.; Suschitzky, H. J. Chem. Soc. (C) 1971,
167. (b) Holland, D. G.; Moore, G. J.; Tamborski, C. J. Org. Chem.
1964, 29, 3042.
32
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Scheme 3. End-game strategy to pyrazolopyridinone-based p38 inhibitors
Scheme 4. Process route to pyrazolopyridinone core
Scheme 5. Condensation of 5a with 2-ethoxymethylene
malononitrile
Scheme 7. Potential decomposition pathway of 5b
Scheme 6. Major impurities during formation of
aminopyrazoles (6a/b)
2-ethoxymethylene malononitrile 5b in acetic acid (5 L/kg) with
aqueous sodium hydroxide, affording 5-aminopyazole 6b in
72% adjusted yield (83 wt %) as a filterable solid upon
crystallization induced by the addition of aqueous sodium
hydroxide.
As 5b was not a suitable raw material for future campaigns,
an alternate approach to 6b was required. Cognizant of the
stabilization imparted to 5b upon N-acylation, we pursued
trapping of the hydrazine as a more stable adduct. Ascorbic
acid has been reported as an organic reducing agent that yields
the oxalate adduct (i.e., 26) directly upon reaction with the
corresponding diazonium salts (Scheme 8).10,11 The reduction
of 25 with ascorbic acid cleanly afforded 26, which upon
The instability of 5b also led to its poor performance under
the optimized conditions established for the synthesis of 6a.
Significantly, precipitation of the 6b did not occur upon
application of the established workup.9 Replacing triethylamine
with aqueous sodium hydroxide led to a much cleaner reaction,
reduced the amount of N-acylation, and facilitated isolation via
antisolvent induced crystallization. The optimized reaction
conditions for the formation of 6b involved treatment of
(9) N-Acylation was also observed under the reaction conditions. N-
Acylation of 5b imparted significant stabilization.
(10) Weidenhagen, R.; Wegner, H.; Lung, K. H.; Nordstrom, L. Chem.Ber.
1939, 72B, 2010.
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
33

Scheme 8. Telescoped synthesis of aminopyrazole
Table 1. Catalyst loading studies
yield (%)
29a
entry
catalyst loading (equiv)
7a
1
2
3
0.6
59
71
93
14
11
2
0.3
0.15
cantly cleaner reaction profile was obtained when the H2
pressure was reduced to 15 psig.14 At this H2 pressure, the
established super stoichiometric amounts of catalyst was
incrementally decreased to 0.15 equivalents (Table 1), which
served to further improve both the purity profile and reaction
yield without significantly increasing the reaction time.15
The originally described isolation conditions involving
distillation of acetic acid under reduced pressure at 60-70 °C
lead to slow decomposition of 7a during the solvent removal.
To avoid both the distillation and the voluminous extractive
workup, a crystallization directly from acetic acid was pursued.
This could be achieved upon addition of water to the reaction
mixture; however, liquor losses were significant due to the high
solubility of 7a in acetic acid. Use of aqueous sodium hydroxide
as the antisolvent remedied this by consuming acetic acid and
minimizing the total liquor volume. The optimized procedure
for the isolation of 7a involved removal of the catalyst by
filtration, crystallization Via addition of aqueous sodium to a
pH of 4-5, and then isolation Via vacuum filtration. On
kilogram scale (3:1 mixture of acetic acid/water, 6 L/kg of 6a),
7a was isolated in 70-79% yield (95 wt %).
treatment with methanol gave 27 as a nonhygroscopic, bench
stable crystalline solid. Treatment of 27 with sodium hydroxide
revealed the free hydrazine, which was immediately condensed
with 2-ethoxymethylene malononitrile. This approach was
demonstrated on gram scale to access 6b in moderate yield
(54%). A telescoped procedure was also developed from the
diazonium salt 2512 through to the aminopyrazole 6b. In this
case, the solution of 26 was solvent switched into methanol to
afford the transesterified product 27. Without isolation, this
mixture was hydrolyzed under basic conditions, and then treated
with 2-ethoxymethylene malononitrile to give 6b. From dia-
zonium 25, this sequence could be carried out in 48% yield
(93 wt %) as demonstrated on 50 g scale.
The second step in the synthesis of building block 10 was
the hydrogenation of nitrile 6. The reaction proceeds through
the intermediacy of imine 28 (Scheme 9), which must be
hydrolyzed before it has the opportunity to be further reduced
to the corresponding aminomethylene compound (29). Other
possible competing reaction pathways include the reduction of
7 to the corresponding hydroxymethylene (30), and the forma-
tion of products arising from intermolecular condensation and
reduction (i.e., 31). The established conditions for this trans-
formation used an acetic acid/water mixture (3:1, 15 L/kg of
6a) under RaNi catalysis (1.15 equiv) and 100 psig of
hydrogen.13 The workup involved removal of the acetic acid
Via distillation, an extractive workup into ethyl acetate and
crystallization upon addition of heptane. While this protocol
afforded pure material, the yield was highly variable and the
low solubility of 7a in ethyl acetate (56 g/L) made the amount
of solvent required for an extractive workup prohibitive.
A closer look at the reaction under the initially reported
conditions showed the observed side products to be associated
with over-reduction of 28. Attention was therefore focused on
maintaining a low steady state concentration of 28 by decreasing
rate of reduction relative to hydrolysis. As expected, a signifi-
Application of this protocol to the reduction of 6b required
only optimization of the reaction and workup volumes. On
kilogram scale, using 5 wt % RaNi 2400 (0.15 equiv) in acetic
acid/water (3:1, 5 L/kg of 6b) under 15 psig H2, 7b was isolated
in 87% yield (100 wt %).
The third step in the synthesis entailed annulation of the
pyridinone ring onto aminopyrazole 7. This was initially
achieved Via condensation of diethylmalonate with 7 under
iminium ion catalysis.16 The product (8) was then isolated as a
piperidine complex Via antisolvent (water) induced crystalliza-
tion. Under the reaction conditions, pyridinone formation was
essentially quantitative; however, partial hydrolysis of 8 to the
corresponding acid (14) significantly impacted the isolated yield
of 8. A telescoped Knoevenagel/hydrolysis protocol was
therefore pursued. The low solubility of 8 in aqueous solvent
mixtures (Table 2) was the major challenge in the formation
of 14, which was prohibitively slow under both acidic and basic
conditions. Mindful of the potential reactivity of the difluoro-
phenyl system to nucleophilic fluoride displacement reactions,
hydrolysis under acidic conditions was favored. Attempts to
increase the rate of hydrolysis with additives (NaI, I2, ZnCl2,
MgCl2), or cosolvents (DMSO, NMP, THF) were largely
ineffective; however, a slight increase in rate was observed using
acetic acid which was attributed to the high acetic acid solubilty
of 8a. Instead of adding copious amounts af acetic acid to
overcome the low solubility of 8a in aqueous ethanol, an
“anhydrous” hydrolysis protocol in acetic acid/ethanol was
(11) During preparation of this manuscript, Norris et al. from Pfizer Process
R&D reported a related application of the use of ascorbic acid in the
reduction of diazonium salts: Norris, T.; Bezze, C.; Franz, S. Z.;
Stivanello, M. Org. Process Res. DeV. 2009, 13, 354.
(12) Safety studies were carried out and showed 25 to be thermally stable
under 160 °C.
(14) The reaction at 100 psig H2 afforded 25% of 29a, 15 psig H2 afforded
4% of 29a.
(15) All reactions achieved complete conversion with 19 h.
(16) Ahluwalia, V. K.; Goyal, B. Synth. Commun. 1996, 26, 1341.
(13) Acevedo, O. L.; Krawczyk, S. H.; Townsend, L. B. J. Heterocycl.
Chem. 1985, 22, 349.
34
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Scheme 9. Nitrile reduction mediated by Raney nickel
designed. Carrying out the hydrolysis in acetic acid/ethanol with
sulfuric acid maintained the solubility of 8a while the water
required for hydrolysis was slowly generated Via condensation
of ethanol and acetic acid. In the experiment, acetic acid and
sulfuric acid were added to the ethanolic mixture of 8a from
the Knoevenagel condensation. The reaction was then heated
to reflux, and ethyl acetate was distillatively removed to drive
the equilibrium reaction to completion. Under these conditions,
hydrolysis of 8a was complete within 20 h.
Scheme 10. LiOAc-catalyzed Hunsdiecker reaction of 17a
On kilogram scale, the piperidine-mediated Knoevenagel
reaction between 7a (2.50 kg) and diethyl malonate was carried
out in refluxing ethanol (3 L/kg of 8a). The reaction reached
completion within 20 h, at which time acetic acid (6 L/kg of
8a) and concentrated sulfuric acid (2 equiv) were added to the
cooled reaction mixture. The hydrolysis was then carried out
at 80-100 °C with continuous distillation of ethyl acetate to
achieve complete conversion within 20 h. The mixture was then
cooled, and 14a was isolated in 96% yield (100 wt %) Via direct
filtration.
Transfer of this technology to the 2,6-difluoro-phenyl series
was straightforward and required only modest optimization of
the reaction and isolation volumes. On kilogram scale, 7b was
condensed with diethylmalonate in ethanol (5 L/kg of 7b)
followed by ester hydrolysis and isolation of 14b in 90% yield
(100 wt %).
The next step in the sequence involved a decarboxylative
bromination (Hunsdiecker-type)17 reaction (Scheme 10). In the
initial approach this step was telescoped with the hydrolysis of
ester 8. The sequence proceeded under relatively mild condi-
tions; however, the reaction required a significant excess of NBS
(2-3 equiv) to achieve full conversion. While the piperidine
from the isolated 8-piperidine complex accounted for some of
the required excess of NBS, experiments with pure 14 revealed
a background reaction with the solvent (MeCN, DMF, NMP,
acetic acid) was also occurring. Use of nonpolar solvents such
as THF eliminated the background reaction; however, very slow
reactions and low conversions resulted. Additives such as
iodosylbenzene,18 lithium acetate,19 and manganese acetate have
been shown to be beneficial to the reaction, with lithium acetate
standing out for its availability, cost, and safety. This proved
to be critical to the reaction, and the bromination of 14 mediated
by lithium acetate in a THF/water solvent mixture (6.5 L/kg of
8a) proceeded to completion with only a slight excess of NBS
(1.1 equiv) in the presence of catalytic amounts of lithium
acetate. Presumably, the reaction proceeds through ꢀ-lactone
intermediate 33 as evidenced by an unisolable transient inter-
mediate (Scheme 10).20 On kilogram scale, NBS was added to
the reaction mixture at 25 °C in portions, the mixture was aged
until most of the starting material was consumed, and then
heated to 50 °C to achieve complete conversion. Upon addition
of water, 9a crystallized and was isolated Via filtration in 93%
yield (99 wt %).
(17) (a) Johnson, R. G.; Ingham, R. K. Chem. ReV. 1956, 56, 219. (b)
Wilson, C. V. Org. React. 1972, 19, 279. (c) Crich, D. Hunsdiecker
and Related Reactions. In ComprehensiVe Organic Synthesis; Trost,
B. M., Fleming, I., Eds.; Pergamon Press: Oxford, 1991; Vol. 7, pp
717-734.
(18) Graven, A.; Jorgensen, K. A.; Dhal, S.; Stanczak, A. J. Org. Chem.
1994, 59, 3543.
Table 2. Solubility of 8a and 14a in water/ethanol mixtures
at 20 °C
entry
% EtOH in water
8a (g/L)
14a (g/L)
(19) (a) Chowhury, S.; Roy, S. J. Org. Chem. 1997, 62, 199. (b) Kuang,
C.; Yang, Q.; Senboku, H.; Tokuda, M. Synthesis 2005, 1319.
(20) Attempts to observe 33 Via react-IR failed due to its low
concentration. Ionization under standard LC/MS conditions was
also not successful. Related intermediates were previously observed
by NMR spectroscopy in the decarboxylative bromination of cis-
cinnamic acid. See ref 19b.
1
2
3
4
5
0
25
6.7
11.6
40.8
39.8
25.6
<0.1
0.3
1.1
2.1
2.6
50
75
100
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
35

Table 3. Screen of conditions for pyridine N-methylation of
9b
water-induced crystallization followed by filtration to give 10b
in 90% yield (99 wt %). Regioisomer (34b) levels in the isolated
10b were further reduced to 0.8 HPLC A%.
Introduction of iodomethane at such a late stage in the
sequence required close attention due to its genotoxicity.
Historical lots of 10b consistently contained <500 ppm of
residual iodomethane. Spiking studies in the subsequent Suzuki
cross-coupling reaction showed it to tolerate up to 5000 ppm
iodomethane without impacting catalyst performance or result-
ing in measurable iodomethane in the resulting 16b (<76 ppb).
The specification for iodomethane content in 10b was therefore
set at 5000 ppm, and no further treatment was necessary for
iodomethane control.
.
entry
conditions
10b:34b time
1
2
3
4
5
6
LiH, LiBr, MeI, 10:1 DME/DMF, rt
LiOtBu, TBAI, MeI, THF, 40 °C
LiOtBu, TBAI, Me2SO4, THF, rt
LiOtBu, LiI, MeI, THF, 40 °C
LiOtBu, MeI, THF, 40 °C
28:1
40:1
5.6:1
50:1
1.5 d
9 h
48 h
9 h
9 h
100:1
5.6:1
LiH, Me2SO4, THF, 70 °C
6 h
To summarize the process development towards building
block 10b, a kilogram-scale route was developed in 48% yield
over five steps without extractive workups or chromatography.
To achieve this, the isolation points were redesigned to
accommodate the reactivity and low solubility of the intermedi-
ates, a novel trans-esterification protocol was designed that
addressed aqueous solubility problems incurred in the Knoev-
enagel reaction of 7, catalytic conditions were applied to the
Hunsdiecker reaction of 14 to avoid the use of excess NBS,
and simplified alkylation conditions were developed to provide
maximal N- Vs O-selectivity and reaction rate during the
formation of 10.
Endgame for the Synthesis of 4. With access to both
boronic acid 135 and pyrazolopyridinone 10b, the remainder
of the synthesis consisted of a three-step protocol (Scheme 11)
that was used for the kilo-scale preparation of related p38 MAP
kinase inhibitors 1 and 2. The order of operations in this
sequence was designed to give maximal control of drug
substance quality attributes by effecting the cross-coupling
reaction first, followed by amidation to give the crude 4, and
then formation of 4-hydrate as a separate step to accommodate
its challenging physicochemical properties.
Up to intermediate 9, the compounds of the 2,4-difluorophe-
nyl series (5a-8a) were less soluble than the corresponding
2,6-difluorophenyl series (5b-8b). Conversely, 9b is signifi-
cantly less soluble than 9a causing its precipitation from the
reaction mixture prior to full conversion. Since slurry-to-slurry
transformations are difficult to control and often give incomplete
conversion,21 a higher dilution (7.5 L/kg of 8b) was used to
achieve a homogeneous end point. Solubility differences aside,
the reactivity and stability of the two series were similar and
9b was isolated in 91% yield (97 wt %) on kilogram scale.
The remaining development work towards 10 focused
exclusively on the 2,6-difluoro-phenyl series (10b). The final
step in the preparation of the 10b was N-methylation of 9b.
Since pyridinones are ambident nucleophiles, selectivity during
their alkylation can be a problem.22 The initial conditions to
effect this transformation involved deprotonation of 9b with
sodium hydride in a 10:1 DME/DMF mixture, followed by
treatment with iodomethane in the presence of excess LiBr (10
equiv).23 These conditions were not considered for scale-up due
to the inherent hazards of handling sodium hydride on scale,
the undesirable solvent mixture, and the large amounts of LiBr
that were required to achieve useful selectivity. Investigation
(Table 3) of cation effects showed that O-alkylation of the
pyridone system to be preferred with a soft counterion (Na, K,
Mg), and N-alkylation with hard counterions (Li). The leaving
group also played an important role, with iodomethane giving
much improved N-selectivity over dimethylsulfate (entries 2 and
3). Solvent influenced both the rate of reaction and N- Vs
O-selectivity. Highly polar solvents gave faster reaction rates
than nonpolar systems, but had a deleterious effect on selectivity.
Fortunately, the selectivity was not temperature dependent, and
both a suitable reaction rate and high selectivity could be
achieved by running the reaction in nonpolar aprotic solvent at
high concentration and at elevated temperature. The optimal
reaction conditions consisted of lithium tert-butoxide, io-
domethane in THF at 40 °C (entry 5). On scale, complete
conversion was reached within 18 h with an N- Vs O-selectivity
of 62:1 (1.5 HPLC A% of 34). Isolation was carried out Via
In developing the cross-coupling reaction, it was desirable
to avoid class 2 solvents and bases that could cause off-gassing
during the quench, to minimize catalyst loading, and to control
the formation of reaction-related impurities. An initial screen
of conditions revealed the K3PO424 in isopropanol/water to
be optimal. A range of Pd(OAc)2/ligand combinations
along with a number of preformed catalysts were then
screened and the three most promising catalyst systems
were compared in detail (Table 4).25 Both Pd(OAc)2/S-
Phos26 (entry 4) and Pd(A-Phos)2Cl2 (entry 11) gave
27
excellent reaction profiles that minimized the formation
of byproducts, with Pd(A-Phos)2Cl2 being the ultimate
(24) K3PO4 may be considered a privileged base for large-scale Suzuki
couplings of this type as it is applicable to a wide range of substrates
and does not present off-gassing issues upon neutralization.
(25) The major byproducts, 36 and 37 derived from 13 Via deboronation
and homocoupling, respectively, and 35 derived from 10b Via
dehalogenation, were fully characterized and their identities confirmed
by synthesis.
(21) For poorly soluble products, starting material may become entrained
in the product, making it unavailable for reaction.
(26) (a) Barder, T. E.; Walker, S. D.; Martinelli, J. R.; Buchwald, S. L.
J. Am. Chem. Soc. 2005, 127, 4685. (b) Walker, S. D.; Barder, T. E.;
Martinelli, J. R.; Buchwald, S. L. Angew. Chem., Int. Ed. 2004, 43,
1871.
(22) (a) Kornblum, N.; Smiley, R. A.; Blackwood, P. K.; Iffland, D. C.
J. Am. Chem. Soc. 1955, 77, 6269. (b) Hopkins, G. G.; Jonak, J. P.;
Minnemeyer, H. J.; Tieckelmann, H. J. Org. Chem. 1967, 32, 4040.
(c) Chung, N. M.; Tieckelmann, H. J. Org. Chem. 1970, 35, 2517.
(23) LiBr was critical to improve the N- Vs O-selectivity.
(27) Guram, A. S.; Wang, X.; Bunel, E. E.; Faul, M. M.; Larsen, R. D.;
Martinelli, M. J. J. Org. Chem. 2007, 72, 5104.
36
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Scheme 11. Process route to 4-hydrate
Table 4. Optimization of cross-coupling conditions
entry
solvent
catalyst
12b (% yld)a
35 (HPLC A%)
36 (HPLC A%)
37 (HPLC A%)
1
H2O
Pd(PPh3)Cl2
15
89
46
91
13
91
96
87
83
80
98
-
-
-
-
2.5
-
-
5
-
-
-
4.7
-
-
-
6
-
-
1.4
-
-
-
1.3
2
EtOH
toluene
iPrOH
iPrOH
iPrOH
iPrOH
iPrOH
iPrOH
iPrOH
iPrOH
Pd(PPh3)Cl2
-
3
Pd(PPh3)Cl2
Pd(PPh3)Cl2
-
4
2.9
-
5
Pd(OAc)2/Xantphos
Pd(OAc)2/X-phos
Pd(OAc)2/S-phos
Pd(OAc)2/PCy3
Pd(OAc)2/DPE-phos
Pd(OAc)2/DPPF
Pd(A-Phos)2Cl2
6
-
7
1.5
-
8
9
-
10
11
-
0.6
a HPLC assay yields.
choice due to its stability and superior turnover number.28
The optimized reaction conditions were 0.1 mol % Pd(A-
Phos)2Cl2, K3PO4 in IPA/water at 80 °C.
adjustment at 70 °C followed by slow cooling to 20 °C resulted
in >100-fold increase in particle size. On kilogram scale, the
coupling of 10b and 13 (1.05 equiv) catalyzed by 0.1 mol %
Pd(A-Phos)2Cl2 in a 1:2 mixture of isopropanol/water (7.5 L/kg
10b) at 70 °C was complete within 2 h. Isolation was carried
out as described above to give 12b in 93% yield (100 wt %).
Pd levels in the isolated 12b were consistently low (<10 ppm),
coupling-related impurities 35, 36, and 37 (Table 4) were all
below 0.05 HPLC A%, and the major carryover impurity (38)
was reduced 10-fold. Noteworthy was the efficiency with which
palladium was rejected in the workup, which may be attributed
to the stability of the catalyst along with its high solubility in
toluene.
The amidation of 12b (Scheme 12) was again based on
technology used to deliver kilogram quantities of related
cyclopropyl amides 1 and 2 (Scheme 1). During the process
development of compound 1, attempts to activate the acid as
its acid chloride were unsuccessful. Conversely, activation using
CDI was effective and offered several advantages over use of
Upon completion of the reaction, the mixture was cooled to
25 °C to give a biphasic system (isopropanol/water)29 in which
the potassium salt of 12b was contained in the isopropanol layer.
Workup of the reaction consisted of separation of the layers,
dilution of the isopropanol layer with deionized water, and then
washing this with toluene to remove the un-ionized organic
impurities (35 and Pd(A-Phos)2Cl2). From the resulting aqueous
layer, 12b was selectively crystallized upon addition of aqueous
hydrochloric acid to a pH of 4. Filtration of the product was
initially a problem as a nucleation-dominated crystallization
caused small particle size (Figure 1). Fortunately, the influence
of temperature on the crystal growth was dramatic, and pH
(28) Loading studies demonstrating that 0.1 mol % catalyst was sufficient
to obtain both full conversion and acceptable reaction kinetics (<4
h).
(29) Isopropanol and water are typically miscible in all proportions;
however, aqueous salt solutions are often immiscible with isopropanol.
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
37

Figure 1. Microscopy of 16b isolated at 25 and 70 °C.
Scheme 12. CDI activation and amidation of 12b with
cyclopropylamine (CPA)
Figure 2. Solubility of 4 in EtOH/water mixtures.
Neither the O-Me isomer of 4 nor the cross-coupling-derived
impurities from 35, 36, or 37 could be detected in the isolated
product.
The final isolation of 4 was complicated by the availability
of a number of distinct solvates, along with a hydrate and an
anhydrous form. Solvates are typically undesirable as the final
form for a drug substance due to limitations on residual solvents
in the active pharmaceutical ingredient (API). Between the
hydrate and the anhydrous form, the hydrate was ultimately
selected for the final form as it was thermodynamically more
stable in aqueous media (preclinical formulations). An ethanol/
water system was selected to generate 4-hydrate on the basis
of the solubility and solid-phase data (Figure 2). In order to
gain further insight into the pseudo-polymorphic behavior of
4, Raman spectroscopy was used to monitor the crystallization.
Characteristic spectroscopic features were identified for each
solvate (ethanol, hydrate) and the anhydrous form (Figure 3),
and these were monitored during cooling/turnover experiments
in a range of ethanol/water mixtures. In general, rapid cooling
afforded the ethanol solvate in most solvent compositions, which
turned over to the hydrate upon aging in water/ethanol mixtures
greater than 10% water. Slow cooling inconsistently gave either
the ethanolate or the hydrate, but seeding with the hydrate
afforded the hydrate exclusively. In addition, the ethanol solvate
was rapidly converted to the hydrate upon slurrying in 1:1 water/
ethanol without seeding.
With the assistance of modeling software32 and the above
data, the following protocol was designed and implemented for
the isolation of 4-hydrate: 4-IPAc was dissolved in 90:10
ethanol/water (32 L/kg of 4) at 40 °C, the solution was polish
filtered, and the vessel/filter was rinsed with 3 volumes of
preheated 90:10 ethanol/water to adjust the concentration to ∼30
mg/mL. Water (1.2 L/kg of 4) was charged into the reactor,33
and atmospheric distillation was carried out to remove 25
volumes of distillate and to adjust the concentration to ∼90
mg/mL. The batch was then cooled to 70 °C, and 0.5 wt % of
Table 5. Solubility of 4 in common solvents at 20 °C
solvent
solubility (mg/mL) solvent solubility (mg/mL)
methanol
ethanol
1-propanol
2-butanol
19.9
10.3
6.9
IPAc
1.3
11.9
0.7
acetone
toluene
THF
1.4
7.4
water
<0.1
the acid chloride including avoidance of HCl gas evolution
along with the comparatively facile handling and safety of CDI.
The neutral byproducts from CDI-mediated amidations also
allow them to be carried out in almost any solvent, THF being
particularly well suited for the preparation of 1 and 2. The
protocol developed for these previous campaigns involved
treatment of the acid/THF mixture with 1.2 equiv of CDI in
portions at 25 °C to give the imidazolide,30 which was then
treated with cyclopropylamine. The primary challenge in the
application of this protocol to 12b (Scheme 12) was the low
solubility of 4 (Table 5), which led to cocrystallization of 39
with the product inasmuch as 5-10%.31 Fortunately, the rapid
reaction of 39 with cyclopropylamine, the mild reaction
energetics (adiabatic temperature rise of 12 °C) and slow
kinetics of crystallization allowed complete conversion to be
achieved Via rapid addition of the cyclopropylamine. Isolation
was accomplished Via solvent switch to isopropyl acetate and
then filtration to give 4 as its isopropyl acetate solvate. On
kilogram scale, the reaction afforded 94% yield (93.2 wt %, 6
wt % IPAc), the major impurity being the unreacted 12b.
(32) Modeling of the distillation was carried out using Dynochem modeling
software.
(33) This was determined by setting the target after distillation to 10
volumes of solvent composed of 4:1 ethanol/water. To reach this
endpoint, Dynochem modeling of the azeotropic system showed a
starting volume of 36.2 volumes composed of 6.7:1 ethanol/water to
be required.
(30) The imidazolide intermediate was unstable under the HPLC conditions
(0.1% TFA in acetonitrile/water); however, it could be quenched with
benzylamine and then analyzed as the corresponding benzamide.
(31) The entrained imidazolide remained unreacted in isolated 4; however,
it readily reacted with the solvent upon dissolution in the subsequent
step.
38
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

Figure 3. Raman data for the 4 hydrate, ethanolate, and anhydrous forms. (All spectra are normalized.)
seed was added. (See Supporting Information for overlay plots
of the Raman spectra recorded during the crystallization.) Slow
cooling over 4.75 h afforded a solution concentration of 10 mg/
mL which was reduced to <1.5 mg/mL upon addition of water
(6 L/kg of 4) and then aged for a further 2 h. The mixture was
then filtered, and the cake was washed with water and then
dried under a stream of nitrogen to give 4-hydrate in 96% yield,
(96 wt %, 4 wt % water). Using this procedure, kilogram
amounts of the drug substance were delivered with the ap-
propriate critical quality attributes, with all impurities and heavy
metals controlled and the residual solvent levels well within
the ICH guidelines.
deliver the prescribed 4-hydrate. Thus, the drug substance was
obtained in eight steps and 38% overall yield for the delivery
of kilogram quantities of the highly pure drug substance.
Experimental Section
5-Amino-1-(2,4-difluoro-phenyl)-1H-pyrazole-4-carboni-
trile (6a). A dry 60-L jacketed reactor equipped with a nitrogen
inlet, a temperature probe, a glycol-cooled condenser, and a
mechanical stirrer was charged with 2,4-difluororphenylhydra-
zine hydrochloride (5a) (2.0 kg, 10.85 mol) and sodium acetate
(445 g, 5.43 mol). Acetic acid (10.0 L) was charged to the
reactor, stirring was initiated, and the internal temperature was
adjusted to 15 °C. Triethylamine (1.65 kg, 16.28 mol) was
charged, while keeping the internal temperature below 30 °C,
and then a solution of 2-(ethoxymethylene)malononitrile (1.4
kg, 11.4 mol) in acetic acid (4.0 L) was added to the reaction
mixture. The mixture was stirred at 20 °C for 1.5 h (<1 A% 5a
by HPLC), the internal temperature was increased to 60 °C,
and the reaction mixture was stirred for 2 h (<1 A% adduct by
HPLC). The batch temperature was adjusted to 20 °C, a solution
of 5 N aqueous NaOH (10.0 L) was added over 30 min while
maintaining the batch temperature below 27 °C, and then the
mixture was aged at 20 °C for 8 h. The mixture was then
filtered, the cake was washed with water (2 × 4 L) and dried
under a stream of nitrogen to give 6a as a pale-orange solid
(2.2 kg, 89.8% yield). The crude 6a (2.3 kg, 97 wt %) was
dissolved in ethyl acetate (23 L) at 75 °C and then cooled to
25 °C over 2 h. Toluene (11.5 L) was charged to the reactor,
the batch volume was reduced in Vacuo (23 L collected), and
additional toluene (13 L) was charged to the reactor. The
mixture was then filtered, and the cake was washed with toluene
Summary and Conclusions
A number of challenges were overcome during the develop-
ment of a kilogram-scale synthesis of p38 MAP kinase inhibitor
4. Foremost was the instability and limited availability of the
2,6-difluorophenylhydrazine (5b), which was solved Via de-
velopment of a telescoped sequence starting from stable,
crystalline, and readily available 2,6-difluorobenzenediazonium
tetrafluoroborate (25). During the synthesis of building block
10b, a methodical understanding of each step led to highly
optimized reaction conditions and designation of more ap-
propriate isolation points. Furthermore, consideration of the
isolation during reaction optimization facilitated direct isolation
of the intermediates as crystalline solids with minimal manipu-
lation during workup. The endgame strategy of this sequence
consisted of a three-step protocol designed to give maximal
control of process-related, genotoxic, and heavy-metal impuri-
ties. Finally, the complicated physicochemical nature of 4 was
navigated with the assistance of process analytics to reliably
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
39

(2 × 4.5 L) and dried under a stream of nitrogen to give 6a as
a pale-beige solid (2.15 kg, 98% yield). Mp 190-191 °C. IR:
3472, 3321, 3234, 3202, 3083, 2230, 1649, 1543, 1511, 1113,
collected as the batch temperature gradually climbed to 100-105
°C over 22 h (<1 A% 8a by HPa LC). The resulting slurry was
cooled to 20 °C over 1 h and aged at until e5 mg/mL of 14a
remained in solution. The mixture was then filtered, the filter-
cake was washed with acetic acid (1 × 3.8 L) followed by water
(2 × 5 L), and then the solids were dried under a stream of
nitrogen to afford 14a as a colorless solid (3.142 kg, 96.0%
yield). Mp 307-308 °C (acetic acid). IR: 2694, 1742, 1624,
1594, 1565, 1512, 1464, 1434, 1411, 1392, 1309, 1275, 1257,
1197, 1177, 1150, 1112, 1036, 991, 975, 961, 879, 858, 827,
1
963, 858, 817, 543 cm-1. H NMR (400 MHz, DMSO-d6) δ
) 7.78 (s, 1H), 7.67-7.49 (m, 2H), 7.25 (t, J ) 8.6 Hz, 1H),
6.79 (br s, 2H). 13C NMR (100.6 MHz, DMSO-d6) δ ) 167.5
(dd, J ) 248.8, 11.3 Hz), 163.0 (dd, J ) 281, 13.9 Hz), 158.2,
147.5, 136.4 (d, J ) 10.4 Hz), 126.7 (dd, J ) 12.1, 3.4 Hz),
120.0, 117.7 (dd, J ) 22.5, 3.4 Hz), 110.8 (app t, J ) 23.4),
77.2. HRMS m/e: Calcd for C10H7F2N4+ (M + H): 221.0633,
Found: 221.0639.
1
811, 736, 691, 677 cm-1. H NMR (400 MHz, DMSO-d6) δ
5-Amino-1-(2,4-difluoro-phenyl)-1H-pyrazole-4-carbalde-
hyde (7a). A 10-L Buchi autoclave equipped with a mechanical
stirrer was charged with 5-amino-1-(2,4-difluoro-phenyl)-1H-
pyrazole-4-carbonitrile (6a) (1.02 kg, 4.63 mol). Acetic acid
(4.5 L) and a slurry of Raney nickel 2400 (45.g) in water (0.25
L), followed by a water (1.25 L) rinse were sequentially charged
to the autoclave. The autoclave was purged with 15 psig
hydrogen, then evacuated (five cycles), and then pressurized to
15 psig hydrogen. Agitation was initiated (800 rpm), and the
mixture was stirred for 9 h (<1 A% 6a by HPLC). The hydrogen
was then purged from the reactor with argon, and additional
acetic acid (1.5 L) was charged to the reactor. The Raney nickel
was then removed Via filtration (1 and 10 µm bag filter), and
the reactor/filter was rinsed with acetic acid (0.5 L). The
resulting solution was stable at 0 °C and stored while a second
batch of the same scale was carried out. The two batches were
charged to a 60-L jacketed reactor followed by an acetic acid
(1.0 L) rinse. The batch temperature was adjusted to 6 °C, 5 N
aqueous NaOH (6 L) was added over 13 min and the resulting
cloudy mixture was aged for 1 h. Aqueous NaOH (5 N, 15 L)
was then added over 1.5 h, and the contents of the reactor were
aged at 20 °C for 75 min (supernatant concentration of 7a )
4.5 mg/mL). The resulting slurry was filtered, the cake was
washed with water (2 × 4.0 L) and dried under a stream of
nitrogen to afford 7a as yellow powder (1.56 kg, 70.8% yield).
Mp 144-146 °C. IR: 3382, 3272, 3124, 2835, 2160, 2030,
1644, 1623, 1531, 1514, 1433, 1412, 1364, 1301, 1272, 1253,
1217, 1198, 1140, 1107, 1102, 1021, 961, 932, 886, 845, 818,
808, 734 cm-1. 1H NMR (400 MHz, chloroform-d) δ 6.95 (s,
2 H) 7.25 (t, J ) 8.61 Hz, 1 H) 7.50 - 7.65 (m, 2 H) 7.86 (s,
1 H) 9.61 (s, 1 H). 13C NMR (100.1 MHz, chloroform-d) δ
105.6, 112.4, 121.3, 130.9, 150.9, 156.2, 158.6, 161.1, 163.6,
182.6. HRMS m/e Calcd for C10H8F2N3O (M + H): 224.0635.
Found: 224.0638.
8.89-8.84 (m, 1 H), 8.41-8.36 (m, 1 H), 7.78 (td, J ) 8.75,
5.97 Hz, 1 H), 7.63 (ddd, J ) 10.56, 9.10, 2.84 Hz, 1 H),
7.38-7.31 (m, 1 H). 13C NMR (100.6 MHz, DMSO-d6) δ
167.6, 162.3 (JCF ) 249, 11 Hz), 157.0 (JCF ) 253, 14 Hz),
147.0, 139.0, 138.1, 130.5 (JCF )10 Hz), 121.4 (JCF ) 12, 4
Hz), 112.3 (JCF ) 24, 4 Hz), 108.9, 108.6, 105.4 (JCF ) 27, 23
Hz). HRMS m/e Calcd for C13H8F2N3O3 (M + H): 292.0534;
Found 292.0523.
5-Bromo-1-(2,4-difluoro-phenyl)-1,7-dihydro-pyrazolo[3,4-
b]pyridin-6-one (9a). A dry 60-L jacketed reactor equipped
with nitrogen inlet, temperature probe, glycol-cooled condenser
and mechanical stirrer was charged with 14a (3.32 kg, 11.39
mol), lithium acetate (75.1 g, 1.14 mol), tetrahydrofuran (20.0
L) and water (1.66 L). Efficient stirring was established, and
then N-bromosuccinimide (2.23 kg, 12.53 mol, 1.1 equiv) was
added at 20 °C in four portions over 3 h during which time the
reaction mixture turned from a slurry into a yellow solution
that was accompanied by gentle off-gassing. The reaction was
stirred at 20 °C for 2 h, heated to 50 °C for 1 h to achieve
complete conversion (0.1 PA% 14a by HPLC) and then cooled
to 20 °C. Water (33.16 L) was added over 70 min and then the
mixture was aged for 2 h at 20 °C. The resulting slurry was
filtered, the reactor/filter cake was washed with water (3 × 6.63
L), and the filter cake was dried under a stream of nitrogen to
afford 9a as a pale-yellow solid (3.30 kg, 89% yield). Mp
223-225 °C. IR: 1638, 1601, 1573, 1530, 1506, 1439, 1296,
1276, 1250, 1190, 1144, 1104, 1063, 1003, 961, 917, 844, 812,
776, 750, 735, 704, 683 cm-1. 1H NMR (400 MHz, CDCl3) δ
8.13 (s, 1H), 7.86 (s, 1H), 7.60 - 7.52 (m, 1H), 7.34 - 7.07
(m, 2H). 13C NMR (100 MHz, DMSO-d6) δ 163.3 (JCF ) 252,
11 Hz), 160.2, 157.3 (JCF ) 254, 13 Hz), 141.3, 136.5, 135.6,
129.4 (JCF ) 10 Hz), 120.6 (JCF ) 13 Hz, 3 Hz), 112.5 (JCF
)
23, 3 Hz), 109.2, 106.9, 106.2 (JCF ) 26, 22 Hz). HRMS m/e
Calcd for C12H7N3OF2Br (M + H): 325.9735. Found: 325.9732.
5-Amino-1-(2,6-difluoro-phenyl)-1H-pyrazole-4-carboni-
trile (6b). A 60-L jacketed reactor equipped with a mechanical
stirrer, a reflux condenser, and a nitrogen inlet was charged with
2,6-difluorophenylhydrazine hydrochloride (5b) (3.2 kg, 17.72
mol), 2-(ethoxymethylene)malononitrile (2.32 kg, 18.61 mol)
and acetic acid (16.0 L). The jacket temperature was set to 0
°C, and 10 N aqueous NaOH (2.66 L, 26.58 mol) was added
over 9 min. The batch was aged for 1.75 h at 25 °C to achieve
complete adduct formation (>1 A% 5b by HPLC), at which
time the mixture was heated to 60 °C for 2.3 h (<0.5 A% of
the adduct remained). The mixture was then cooled to 0 °C,
the pH was adjusted to 5.25 with 5 N aqueous NaOH (34 L) at
such a rate as to maintain the batch temperature <25 °C, and
1-(2,4-Difluoro-phenyl)-6-oxo-6,7-dihydro-1H-pyrazolo[3,4-
b]pyridine-5-carboxylic Acid (14a). A 60-L jacketed reactor
equipped with a nitrogen inlet, a temperature probe, a glycol-
cooled condenser, and a mechanical stirrer was charged was
charged with 5-amino-1-(2,4-difluoro-phenyl)-1H-pyrazole-4-
carbaldehyde (7a) (2.5 kg, 11.2 mol), ethanol (7.5 L), followed
by dimethyl malonate (2.04 L, 13.44 mol) and piperidine (1.16
L, 11.76 mol). The resulting suspension was heated, becoming
homogeneous at 61-62 °C, and stirred for 20 h (<1 A% 7a by
HPLC). The mixture was cooled to 70 °C, ethyl acetate (5.0
L), acetic acid (15 L), and then concentrated sulfuric acid (1.19
L, 22.4 mol) were charged to the reactor. The resulting
suspension was heated to reflux (79 °C), and the distillates were
40
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

then aged at 20 °C for 16 h (supernatant concentration of 6b )
3.2 mg/mL). The mixture was then filtered and washed with
water (3 × 6.4 L) followed by toluene (3 × 3.6 L), and then
the cake was dried under a nitrogen stream to give 6b as an
orange solid (3.36 kg, 71% yield).
pressure of 15 psig, agitation initiated (800 rpm) and the mixture
was stirred for 6.5 h (<1 A% 6b by HPLC). The hydrogen was
then purged from the autoclave with argon, the Raney nickel
was removed Via filtration (1 and 10 µm bag filter), and the
reactor was washed with acetic acid (1.5 L). Aqueous 5 N
NaOH was added to the reaction mixture over 2.5 h to adjust
the pH to 5.23 (supernatant assay 7b ) 1.8 mg/mL), and the
resulting slurry was aged for 1 h at 20 °C. The mixture was
filtered, and the cake was washed with water (2 × 10.5 L) and
dried under a stream of nitrogen to give 7b as a yellow solid
(1.25 kg, 87.4% yield). Mp 169.7-170.5 °C. IR: 3396.5, 3269,
3120, 1644, 1625, 1597, 1545, 1474, 1405, 1355, 1288.5,
1244.0, 1197.9, 1000, 933, 882, 814, 780, 763.7, 735, 717,
632.2, 593, 531, 498, 456 cm-1. 1H NMR (400 MHz, CDCl3)
δ 9.73 (s, 1 H) 7.89 (s, 1 H) 7.45 - 7.58 (m, 1H) 7.15 (t, J )
8.03 Hz, 2H) 5.68 (s, 2H). 13C NMR (100.1 MHz, CDCl3) δ
184.0, 158.6 (JCF ) 257, 4 Hz), 150.6, 142.9, 131.9 (JCF ) 10
Hz), 113.6, 112.9 (JCF ) 19, 4 Hz), 106.5. HRMS m/e Calcd
for C10H8F2N3O (M + H): 224.0635. Found: 224.0626.
1-(2,6-Difluoro-phenyl)-6-oxo-6,7-dihydro-1H-pyrazolo[3,4-
b]pyridine-5-carboxylic Acid (14b). To a 60-L jacketed
reactor equipped with a nitrogen inlet, a temperature probe,
a glycol-cooled condenser and a mechanical stirrer, was
charged 7b (2.37 kg, 10.63 mol) and ethanol (6.0 L).
Agitation was initiated, diethylmalonate (1.94 L, 12.76
mol) and piperidine (1.10 L; 11.16 mol) were charged and
the reaction mixture was heated at reflux (80 °C) for 7 h
(96.7 A% 8b, and 2.3 A% 18b by HPLC). The batch was
then cooled to 30 °C, acetic acid (18.9 L) was charged to
the reaction mixture over 20 min, followed by sulfuric
acid (1.11 L). The mixture was heated to reflux (batch
temperature ) 81 °C), and the distillate was collected until
the batch reached a final temperature of 108 °C (<1 A%
8b by HPLC). The reaction mixture was then cooled to
20 °C and held for 1 h. The solids were isolated by
vacuum filtration, washed with a 1:3 acetic acid/water
mixture (2 × 7 L), water (1 × 7 L) and dried under a
stream of nitrogen to afford 14b as a pale yellow solid
(2.8 kg, 90.1% yield). Mp 304-306 °C. IR: 3030, 2951,
2846, 2688, 1718, 1640, 1615, 1599, 1503, 1478, 1464,
1393, 1291, 1248, 1214, 1196, 1076, 1037, 1001, 876,
To a dry 2.0-L jacketed reactor equipped with a nitrogen
inlet, temperature probe, glycol-cooled condenser, and mechan-
ical stirring, was charged with 2,6-difluorobenzenediazonium
tetrafluoroborate (25) (65.2 g, 219.4 mmol), L-ascorbic acid
(39.8 g, 226 mmol), and acetonitrile (250 mL). Water (19.8 g,
1.09 mol) was added to the reaction mixture, stirring was
initiated, the jacket was set to 20 °C, and the mixture was stirred
until complete consumption of the 2,6-phenyldiazonium tet-
rafluoroborate (typically 30 min). Methanesulfonic acid (28.5
mL, 438.8 mmol) was added over 15 min followed by methanol
(50 mL, 1.23 mol); the internal temperature was then increased
to 60 °C and held for >5 h (>95% hydrazine by HPLC). The
mixture was then cooled to 20 °C, and 10 N NaOH (83 mL,
830 mmol) was added over 20 min until the pH ) 6.
2-Ethoxymethylene malonitrile (26.8 g, 219.4 mmol) was then
added to the biphasic mixture, and the temperature was
increased to 80 °C and the mixture stirred for >4 h (>95%
product by HPLC). The solvent was then switched to isopropyl
acetate Via distillation under vacuum (e3% by GC, ∼5 V IPAc
final), and 5 N NaOH (250 mL, 1.25 mol) was added. The
resulting mixture was stirred for 30 min and then filtered to
remove brown solids. The reactor and wetcake were washed
with isopropyl acetate (150 mL, 3.0 V), and the biphasic mixture
was charged back into the 2.0-L reactor. The layers were
separated, and the organic layer was washed with a 1:1 brine/
water (250 mL) mixture. The solvent of the organic layer was
switched to 2-BuOH (e3% IPAC by GC, ∼2 V 2-BuOH final)
Via vacuum distillation, the mixture was heated at 50 °C until
homogeneous, and then heptane (150 mL, 3 V) was added
slowly over 1 h. The solution was then cooled to 30 °C over
1 h, and held at 30 °C for >30 min. A heat cycle was then
applied (60 °C for >30 min, cool 20 °C over >1 h) and then
held until the mother liquors concentration of 6b was <25 mg/
mL. The resulting slurry was then filtered, and the cake was
washed with toluene (1 × 150 mL) and heptane (1 × 150 mL)
and dried to constant mass under vacuum/N2 at room temper-
ature to give 6b as a tan solid (24.9 g, 48% yield). Mp 130-131
°C. IR: 3324, 3228, 3193, 2225, 1645, 1600, 1568, 1539, 1501,
1476,1246, 1005, 950, 967, 750, 703, 722, 669 cm-1. 1H NMR
(400 MHz, DMSO-d6) δ 7.83 (s, 1 H), 7.66 (m, 1 H), 7.35 (t,
J ) 8.3 Hz, 2 H), 6.95 (s, 2 H). 13C NMR (100.6 MHz, DMSO-
d6) δ 158.5 (JCF ) 253, 3 Hz), 153.7, 142.9, 132.4(JCF ) 10
Hz), 114.6, 114.0 (JCF ) 16 Hz), 112.8 (JCF ) 23, 14 Hz),
71.7. HRMS m/e Calcd for C10H7F2N4+ (M + H): 221.0633;
Found 221.0628.
1
788, 780, 750, 693, 635, 583, 572, 549 cm-1. H NMR
(600 MHz, DMSO-d6) δ 7.44 (t, J ) 8.4 Hz, 2 H) 7.73
(m, 1 H) 8.43 (s, 1 H) 8.87 (s, 1 H), 13 (bs, 1 H). 13C
NMR (150 MHz, DMSO-d6) δ 168.0, 165.00, 158.1 (JCF
) 253.1, 3.2 Hz), 148.39, 138.8, 138.7, 132.5 (JCF ) 10
Hz) 114.1 (JCF ) 16 Hz) 112.8 (JCF ) 19, 3 Hz) 108.9,
108.6. HRMS m/e Calcd for C13H8F2N3O3 (M + H):
292.0534. Found: 292.05284.
5-Bromo-1-(2,6-difluoro-phenyl)-1,7-dihydro-pyrazolo[3,4-
b]pyridin-6-one (9b). A dry 60-L jacketed reactor equipped
with a nitrogen inlet, a temperature probe, a glycol-cooled
condenser, and a mechanical stirrer was charged with 14b (2.76
kg, 9.48 mol), lithium acetate (63.5 g, 0.95 mol), tetrahydrofuran
(19.3 L) and water (1.38 L). Efficient stirring was established,
and N-bromosuccinimide (1.86 kg, 10.4 mol) was added at 20
°C in four portions over 3 h, during which the reaction mixture
turned from a slurry into a yellow solution which was
5-Amino-1-(2,6-difluoro-phenyl)-1H-pyrazole-4-carbalde-
hyde (7b). A 10 L Buchi autoclave equipped with a mechanical
stirrer was charged with a solution of 5-amino-1-(2,6-difluoro-
phenyl)-1H-pyrazole-4-carbonitrile (6b) (1.74 kg 6.4 mol) in
acetic acid (5.7 L). Raney nickel 2400 (83.54 g) in water (0.9
L) was then charged followed by a water rinse (1.0 L). The
autoclave was purged with 15 psig hydrogen and then evacuated
(5 cycles). Hydrogen was charged to the autoclave to a final
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
41

accompanied by gentle off-gassing. The mixture was stirred at
20 °C for 2 h, and then heated to 50 °C for 1 h (0.1 PA% 18b
remained by HPLC). The reaction mixture was cooled to 20
°C, water (37.3 L) was added over 70 min, and the batch was
aged for 2 h at 20 °C. The resulting slurry was then filtered,
the cake was washed with a 2:1 mixture of tetrahydrofuran:
water (2 × 8.28 L) and dried under a stream of nitrogen to
afford 9b as a pale yellow solid (2.8 kg, 91% yield). Mp
223-225 °C. IR: 1628, 1600, 1572, 1530, 1501, 1484, 1394,
1280, 1250, 1188, 1143, 1103, 1059, 1004, 915, 844, 794, 777,
752, 724, 704 cm-1. 1H NMR (400 MHz, DMSO-d6) δ 12.80
(s, 1H), 8.52 (s, 1H), 8.23 (s, 1H), 7.80-7.64 (m, 1H),
7.48-7.40 (m, 2H). 13C NMR (100 MHz, DMSO-d6) δ 159.6,
polish filtered and then diluted with isopropanol (8.0 L). The
colorless solution was then heated to 70 °C, and 2 N aqueous
HCl (4 L, 8 mol) was added slowly over 45 min to adjust the
pH to 3. The resulting mixture was then cooled to 20 °C over
1 h and aged for a further 12 h (supernatant concentration of
12b ) 1 mg/mL). The mixture was then filtered, washed with
a 1:2 isopropanol/water mixture (2 × 6.0 L), and dried under
a stream of nitrogen to afford 12b as a colorless solid (2.05 kg,
93.4% yield). Pd: <0.5. Mp 272-273 °C. IR: 2945, 1701, 1623,
1598, 1552, 1538, 1507, 1475, 1213, 1189, 783, 772, 740, 651
1
cm-1. H NMR (400 MHz, DMSO-d6) δ 12.85 (s, 1H), 8.20
(s, 1H), 7.91 (s, 1H), 7.87 (dd, J ) 2.4, 8.0 Hz, 1H), 7.75-7.82
(m, 2H), 7.49 (t, J ) 8.0 Hz, 2H), 7.10 (d, J ) 8.0 Hz, 1H),
3.28 (s, 3H), 2.24 (s, 3H). 13C NMR (100 MHz, DMSO-d6) δ
167.1, 160.1, 158.2 (JCF ) 251 Hz, 2 Hz), 156.9, 143.4, 142.3,
138.3, 137.6, 133.1 (JC F ) 10 Hz), 132.4. 131.1, 130.0, 128.6,
128.2, 116.3 (JCF ) 17 Hz), 112.7 (JCF ) 15, 3 Hz), 104.9,
29.0, 19.8.. HRMS m/e Calcd for C21H16F2N3O3 (M + H):
396.1154. Found: 396.1160.
158.3 (JCF ) 253 Hz, 3 Hz), 147.9, 136.9, 136.8, 132.6 (JCF
)
10 Hz), 114.8 (JCF ) 17 Hz), 112.7 (JCF ) 22 Hz), 109.8, 103.4.
HRMS m/e Calcd for C12H7N3OF2Br (M + H): 325.9735.
Found: 325.9737.
5-Bromo-1-(2,4-difluoro-phenyl)-7-methyl-1,7-dihydro-
pyrazolo[3,4-b]pyridin-6-one (10b). A 60-L jacketed reactor
equipped with a nitrogen inlet, a temperature probe, a glycol-
cooled condenser, and a mechanical stirrer was charged with
9b (2.78 kg, 8.55 mol) followed by tetrahydrofuran (14 L).
Agitation was initiated, and lithium tert-butoxide (0.75 kg, 9.4
mol) was added portion-wise, maintaining the internal temper-
ature below 30 °C. The mixture was stirred at 20 °C for 30
min during which time the mixture became homogeneous.
Iodomethane (2.43 kg, 17.1 mol) was then added, the jacket
was set to 40 °C, and the mixture was stirred for 18 h (>1.5
A% 9b by HPLC). Water (42 L) was then added over 2 h, and
the mixture was aged for an additional 2 h. The slurry was then
filtered, and the filter cake was washed with water (2 × 8.5 L)
and dried under a stream of nitrogen to afford 10b of as a
colorless solid (2.65 kg, 90% yield). Mp 188-189 °C; IR: 3090,
N-Cyclopropyl-3-[1-(2,6-difluoro-phenyl)-7-methyl-6-oxo-
6,7-dihydro-1H-pyrazolo[3,4-b]pyridin-5-yl]-4-methyl-ben-
zamide Hydrate (4-Hydrate). A dry 60-L jacketed reactor
equipped with a nitrogen inlet, a temperature probe, a
glycol-cooled condenser, and a mechanical stirrer was
charged with 12b (1.8 kg, 4.56 mol) and tetrahydrofuran
(27.4 L). The jacket temperature was set to 20 °C,
carbonyldiimidazole (0.96 kg, 5.9 mol) was charged in
four equal portions, and the mixture was aged for 1.5 h
(<0.1 A% of 16b by HPLC). The reaction was then heated
to 30 °C, and cyclopropylamine (1.120 L, 9.2 mol) was
charged over 2 min (max. batch temperature ) 42.5 °C).
The mixture was aged at 50 °C for 16 h, and then 9.75 L
of solvent was removed in Vacuo. Isopropyl acetate (18.0
L) was charged to the reactor,,,,, and a further 18.0 L of
solvent was removed in Vacuo. Additional isopropyl
acetate (7.2 L) was charged, the mixture was cooled to
20 °C and then stirred for 1 h at 20 °C (supernatant
concentration of 4 ) 3.0 mg/mL). The reaction mixture
was then filtered, the cake was washed with isopropyl
acetate (2 × 7.2 L) and dried under a stream of nitrogen
to afford 4-IPAc as a colorless solid (1.99 kg, 94.2%
yield). 4-IPAc (1.95 kg, 4.48 mol) followed by a 9:1
mixture of EtOH:water (62.4 L) was charged to a 100-L
jacketed reactor equipped with a nitrogen inlet, a tem-
perature probe, a glycol-cooled condenser and a mechan-
ical stirrer. The mixture was heated to 41 °C, stirred until
homogeneous, and then polish filtered into a 100-L
jacketed reactor equipped with a nitrogen inlet, a tem-
perature probe, a glycol-cooled condenser, and a mechan-
ical stirrer. The lines, reactor, and filter were washed with
warm 9:1 EtOH/water (3 × 1.2 L), and the resulting spec-
free solution was distilled at atmospheric pressure to
remove 48 L of solvent mixture. The batch temperature
was adjusted to 70 °C, seed (9.75 g, 0.5 wt %) was added
as a slurry in 1:1 EtOH/water (200 mL), and the mixture
was cooled to 20 °C over 4.75 h (supernatant concentration
of 4 ) 10 mg/mL). Water (11.7 L) was added over 24
min, and the batch was aged for an additional 2 h
1
1645, 1667, 1472, 1002, 988, 773, 744 cm-1. H NMR (400
MHz, DMSO-d6) δ 8.45 (s, 1H), 8.16 (s, 1H), 7.83-7.76 (m,
1H), 7.48 (t, J ) 9.0 Hz, 2H), 3.27 (s, 3H). 13C NMR (100
MHz, DMSO-d6) δ 157.5 (JCF ) 253.2, 2.6 Hz), 157.2, 143.0,
137.5, 134.9, 133.2 (JCF ) 10 Hz), 115.9 (JCF ) 17 Hz), 112.6
(JCF ) 19, 4 Hz), 109.2, 105.4, 30.1. HRMS m/e Calcd for
C13H9BrF2NO3 (M + H): 339.9892, Found: 339.9884.
3-[1-(2,6-Difluoro-phenyl)-7-methyl-6-oxo-6,7-dihydro-
1H-pyrazolo[3,4-b]pyridin-5-yl]-4-methyl-benzoic Acid (12b).
A 100-L jacketed reactor equipped with a mechanical stirrer, a
condenser, and a nitrogen inlet was charged with 5-bromo-1-
(2,6-difluorophenyl)-7-methyl-1H-pyrazolo[3,4-b]pyridin-6(7H)-
one (10b) (2.0 kg, 93.8 wt %, 5.52 mol), 3-borono-4-
methylbenzoic acid (13) (1.33 kg, 87.2 wt %, 6.45 mol) and
isopropanol (5.0 L). Agitation was initiated and a solution of
K3PO4 (3.52 kg, 16.6 mol) in water (10.0 L) was charged to
the reactor followed by a slurry of palladium bis-(4-dimethy-
laminophenyl-di-tert-butylphosphine) dichloride (4.2 g, 6 mmol)
in isopropanol (20 mL). The atmosphere was inerted, the jacket
temperature was set to 85 °C, and the reaction mixture was
stirred for 1 h (<1 A% 10b by HPLC). The reaction mixture
then was cooled to 20 °C, and the layers of the resulting biphasic
mixture were separated. The top organic layer was retained,
water (1.5 L) was added to this, and the resulting solution was
washed with toluene (2.5 L). The lower aqueous layer was
42
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

(supernatant concentration of 4 ) 1.34 mg/mL). The
resulting slurry was filtered, the cake was washed with
water (2 × 5.9 L) and then dried under nitrogen to give
4-hydrate as a colorless solid (1.88 kg, 96.4% yield). Mp
(hydrate) 240-241 °C. IR: 3520, 3290, 1622, 1608, 1556,
1536, 1530, 1477, 1006, 991, 793, 786, 773, 761, 748,
726 cm-1. 1H NMR (400 MHz, DMSO-d6) δ 8.34 (d, J )
3.5 Hz, 1H), 8.20 (s, 1H), 7.89 (s, 1H), 7.85-7.72 (m,
2H), 7.68 (s, 1H), 7.49 (t, J ) 8.6 Hz, 2H), 7.33 (d, J )
7.5 Hz, 1H), 3.26 (s, 3H), 2.90-2.82 (m, 1H), 2.19 (s,
3H), 0.73-0.65 (m, 2H), 0.59-0.52 (m, 2H). 13C NMR
(100 MHz, DMSO-d6) δ 166.9, 160.1, 158.2 (JCF ) 252,
3 Hz), 143.3, 140.2, 138.2, 137.2, 133.1 (JCF ) 10 Hz),
135.3, 131.6, 129.5, 128.9, 126.5, 125.8, 116.3 (JCF ) 16
Hz), 112.7 (JCF ) 19, 3 Hz), 29.0, 23.0, 19.5, 5.6. HRMS
m/e Calcd for C24H21F2N4O2 (M + H): 435.1627. Found:
435.1615.
Acknowledgment
We thank Dr. M. M. Faul, Dr. A. S. Tasker, Dr. L. Pettus,
and Dr. R. P. Wurz for valuable discussions of the synthesis of
the target compound. Dr. K. Turney, J. Chen, C. Scardino, and
B. Shaw are thanked for analytical support. J. Tvetan, J.
Tomaskevitch, B. Southern, S. Huggins, and G. Sukay are
thanked for support and valuable input during scale-up.
Supporting Information Available
Raman spectra from in-situ monitoring of the cooling
crystallization and NMR spectra of all intermediates and final
products. This material is available free of charge via the Internet
at http://pubs.acs.org.
Received for review July 26, 2010.
OP100205S
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
43