Development of an efficient large-scale synthesis for a 4 H-imidazo[5,1- c ][1,4]benzoxazine-3-carboxamide derivative for depression and anxiety

Article abstract of DOI:10.1021/op100103v

The development and scale-up of an optimized synthesis for a novel drug candidate for depression and anxiety is presented. The updated synthesis represents a convergent and efficient four-stage approach to the API, overcoming high cost of goods (COG), gen

Full text of DOI:10.1021/op100103v

Organic Process Research & Development 2010, 14, 859–867
Development of an Efficient Large-Scale Synthesis for a 4H-imidazo[5,1-c][1,4]
benzoxazine-3-carboxamide Derivative for Depression and Anxiety
Nicola Giubellina,* Paolo Stabile, Gilles Laval, Alcide D. Perboni, Zadeo Cimarosti, Pieter Westerduin, and Jason W. B. Cooke
Chemical DeVelopment, GlaxoSmithKline Medicine Research Centre, Via Fleming 4, 37135 Verona, Italy
Abstract:
inhibitor 5-HT1 receptor antagonist selected for the cure of
depression and anxiety that reached early phase II trials.5
The original medicinal chemistry synthesis and the subse-
quent process development campaigns followed a linear syn-
thetic strategy (Scheme 1). Process development of the me-
dicinal chemistry route (10 steps) followed a fit for purpose
(design for delivery) strategy to afford substantial improvements,
resulting in a more scalable process. This was run successfully
for the preparation of kilogram quantities of active pharmaceuti-
cal ingredient (API) to cover phase I studies. While this process
was suitable for the preparation of the initial supplies of API,
it was found to be inefficient for further scale-up. The route
suffers from low yields over a linear sequence, used toxic
reagents (e.g., osmium tetroxide), and high loading of metal
catalysts, and introduced an expensive intermediate, 5-(1-
piperazinyl)quinaldine 5, in the early steps. The promising
results of the first dose in man have led the Chemical
Development team to look into the discovery of an alternative
route to supply material for phase II.
The development and scale-up of an optimized synthesis for a novel
drug candidate for depression and anxiety is presented. The
updated synthesis represents a convergent and efficient four-stage
approach to the API, overcoming high cost of goods (COG),
general lack of convergence, and low yield of previous routes. A
lower cost of goods resulted from using 3-nitrosalicylaldehyde as
a starting material and introducing the expensive side chain (2-
methyl-5-(piperazin-1-yl)quinoline) at a later stage. Green chem-
istry principles were applied when a direct amidation enabled a
straightforward conversion of the 4H-imidazo[5,1-c][1,4]benzoxazine-
3-carboxylate to the corresponding amide in the last step. In
addition, the total number of stages was reduced from seven to
four, and solvent usage was greatly minimized. The modified
synthesis was demonstrated on a kilogram pilot scale, allowing
the isolation of the API in 17% overall yield with the required
purity.
Introduction
Results and Discussion
Neuroscience drugs will be playing a paramount role in the
treatment of depression in the coming years.1 Depression is a
chronic, recurring, and potentially life-threatening illness that
affects up to 20% of the population across the globe.2 It is one
of the top 10 causes of morbidity and mortality worldwide.3
Although today’s treatments for depression are generally safe
and effective, they are far from ideal, in addition to the need to
administer the drugs for weeks or months to observe clinical
benefit.3
Selective serotonin reuptake inhibitors (SSRIs) have a
widespread utility in the treatment of depression and other
mental illnesses, but their therapeutic use showed latency in
the onset of clinically meaningful effects. It was identified that
approaches based on inhibition of serotonin reuptake in
conjunction with antagonism of 5-HT1 autoreceptors offer
advantages over the current antidepressants in terms of a faster
onset of therapeutic effect and improved efficacy.4 Candidate
6-{2-[4-(2-methyl-5-quinolinyl)-1-piperazinyl]ethyl}-4H-imi-
dazo[5,1-c][1,4]benzoxazine-3-carboxamide 11 is a presynaptic
The alternative route selection was tackled by synthetic
chemists that envisioned several synthetic strategies toward the
API, out of which the route with the use of 2-hydroxy-3-
nitrobenzaldehyde 17 as a starting material was evaluated in
full (Scheme 2). 17 gives the possibility of modulating the
nitrophenol into a benzoxazine-3-one in a few steps that can
eventually provide the imidazolylcarboxylate ring.5,6 In addition,
homologation of the aldehyde 16 would easily give access to
the piperazinylquinoline side of the molecule. The last stage
encompasses the hydrolysis of the carboxylic ester 13, followed
by a coupling reaction to the desired tricyclic carboxamide 11.
Process Development for Stage One: Synthesis of 1,4-
Benzoxazine-3-one 16. O-Alkylation of 3-nitrosalicylaldehyde
17 with ethyl bromoacetate proceeded smoothly in tetrahydro-
furan at reflux.7
Initially, the reaction conditions included tetrahydrofuran and
diisopropylethylamine (DIPEA) for the first step. Accordingly,
(5) (a) Bergauer, M.; Bertani, B.; Biagetti, M.; Bromidge, S. M.; Falchi,
A.; Leslie, C. P.; Merlo, G.; Pizzi, D. A.; Rinaldi, M.; Stasi, L. P.;
Tibasco, J.; Vong, A. K.; Ward, S. E. Quinoline and quinazoline
derivatives having affinity for 5-HT1-type receptors. WO/2005/014552,
2005. (b) Bentley, J.; Bergauer, M.; Bertani, B.; Biagetti, M.; Borriello,
M.; Bromidge, S. M.; Gianotti, M.; Granci, E.; Leslie, C. P.;
Pasquarello, A.; Zucchelli, V. Fused tricyclic derivatives for the
treatment of psycotic disorders. WO/2006/024517, 2006. (c) Serafi-
nowska, H. T.; Blaney, F. E.; Lovell, P. J.; Merlo, G.; Scott, C. M.;
Smith, P. W.; Starr, K. R.; Watson, J. M. Bioorg. Med. Chem. Lett.
2008, 5581.
* Author to whom correspondence may be sent. Telephone: +39 045 8218053.
Fax: +39 045 8218117. E-mail: Nicola.2.Giubellina@gsk.com.
(1) Pardridge, W. M. Drug DiscoVery Today 2007, 12, 54.
(2) (a) Manji, H. K.; Drevets, W. C.; Charney, D. S. Nature Med. 2001,
7, 541. (b) Charney, D. S. Am. J. Psychiatry 2004, 161, 195.
(3) Berton, O.; Nestler, E. J. Nat. ReV. Neurosci. 2006, 7, 137.
(4) (a) Oficialdegui, A. M.; Martinez, J.; Perez, S.; Hears, B.; Irurzun,
M.; Palop, J. A.; Tordera, R.; Lasheras, B.; del Rio, J.; Monge, A.
Farmaco 2000, 55, 345. (b) Evrard, D. A.; Zhou, P.; Yi, S.; Zhow,
D.; Smith, D.; Sullivan, K. M.; Hornby, G. A.; Scheckter, L. E.;
Andree, T. H.; Mewshaw, R. E. Bioorg. Med. Chem. Lett. 2005, 15,
911.
(6) Watjen, F.; Baker, R.; Engelstoff, M.; Herbert, R.; MacLeod, A.;
Knight, A.; Merchant, K.; Moseley, J.; Saunders, J.; Swain, C. J.;
Wong, S. E.; Springer, J. P. J. Med. Chem. 1989, 32, 2282.
10.1021/op100103v  2010 American Chemical Society
Published on Web 06/07/2010
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
859

Scheme 1. First scale-up route of the API 6-{2-[4-(2-methyl-5-quinolinyl)-1-piperazinyl]ethyl}-
4H-imidazo[5,1-c][1,4]benzoxazine-3-carboxamide 12
Scheme 2. Retrosynthetic approach toward the intermediate grade 11
860
•
Vol. 14, No. 4, 2010 / Organic Process Research & Development

Scheme 3. Alkylation reaction of 3-nitrosalicylaldehyde 17 with ethylbromoacetate, and formation of benzofuran 197
Scheme 4. Stage one: telescoped procedure for 3-oxo-1,4-benzoxazine-8-carbaldehyde 16
the selected alkylation conditions were run on a small scale
and then were reproduced in a 2-L glass reactor with complete
conversion to the ester intermediate 18 in high yields (Scheme
3). Further process development included the replacement of
tetrahydrofuran with 2-methyltetrahydrofuran (MeTHF) with
no impact on the reaction profile and yields.8 While the initial
process included partition between hydrochloric acid and
ethylacetate, MeTHF was used during the aqueous washes
because it is only partially soluble in water, thus simplifying
the process. Fine-tuning of the reaction temperature, equivalents
of reagents, and amount of solvents led to the isolation of the
phenoxy derivative 18 in 95% a/a (area/area, HPLC) or higher,
with no need for further purification.
agents were screened, including palladium on carbon,10,11 tin(II)
chloride,12 and iron.13
Catalytic hydrogenation reactions were carried out in metha-
nol with various amounts of palladium catalyst. Unfortunately,
they furnished the 4-hydroxybenzoxazine-3-one 21 as the major
reaction product through intramolecular oxime cyclization with
the acetate moiety. Tin(II) chloride reductions12 were strongly
dependent on the acid concentration, leading to mixtures of
acetic acid derivative 20, benzoxazine-3-one 16, and compound
21 in various ratios according to the reaction conditions. As an
example, the amount of the undesired cyclized derivative 21
increased in tin(II) chloride dihydrate with concentrated hy-
drochloric acid in THF, and N-hydroxybenzoxazine 21 was
isolated in 76% yield (Figure 1).
Assuming that basic conditions and protic solvents might
allow the undesired cyclization of the nitrophenoxyacetate 18
to the benzofuran 19 (Scheme 3),7 the O-acetylation reaction
conditions were studied in non-protic media (THF, MeTHF)
and a few bases (DIPEA, Et3N) in order to evaluate the
formation of this side product 19. Under no circumstances did
the reaction give the undesired benzofuran side product, 19.
The best reaction conditions included N,N-diisopropylethy-
lamine (1.9 equiv) as a base and THF or MeTHF (3 vol) as a
solvent at 70 °C for 30 min. Those conditions gave the
phenoxyacetate 18 in 98% yield and 99% a/a (HPLC).
The nitrophenoxyacetate 18 was submitted to reduction to
give 1,4-benzoxazine-3-one 16 that results from the amino group
formation followed by in situ lactamization (Scheme 4).9 In
order to gain access to the amino compound 22 a few reducing
Complete conversion of the nitrophenoxyacetate 18 to
benzoxazine-3-one 16 was observed with iron powder in acidic
media. When acetic acid was applied, the major product of the
synthesis was 16, which was isolated in poor yield because of
the rather difficult workup. It was found that ammonium
chloride was the best proton source because it prevented the
hydrolysis of the ester, allowing the ring closure in high yield.
Reduction of nitrobenzaldehyde 18 with iron powder in hot
DMF and aqueous ammonium chloride gave complete conver-
(10) (a) Koini, E. N.; Papazafiri, P.; Vassilopoulos, A.; Koufaki, M.; Horvath,
Z.; Koncz, I.; Virag, L.; Papp, G. J.; Varro, A.; Calogeropoulou, T. J. Med.
Chem. 2009, 52, 2328. (b) Breznik, M.; Grdadolnik, S. G.; Giester,
G.; Leban, I.; Kikelj, D. J. Org. Chem. 2001, 66, 7044. (c) Kajino,
M.; Shibouta, Y.; Nishikawa, K.; Meguro, K. Chem. Pharm. Bull.
1991, 39, 2896.
(11) Palladium on carbon powder (10%), moisture content (54.66%),
purchased from Engelhard.
(12) (a) Duerfeldt, A. S.; Brandt, G. E. L.; Blagg, B. S. J. Org. Lett. 2009,
11, 2353. (b) Anderluh, P. S.; Anderluh, M.; Ilas, J.; Mravljak, J.;
Dolenc, M. S.; Stegnar, M.; Kikelj, D. J. Med. Chem. 2005, 48, 3110.
(13) (a) Hamper, B. C.; Leschinsky, K. L.; Massey, S. S.; Bell, C. L.;
Brannigan, L. H.; Prosch, S. D. J. Agric. Food Chem. 1995, 43, 219.
(b) Pradhan, P. K.; Dey, S.; Jaisankar, P.; Giri, V. S. Synth. Commun.
2005, 35, 913.
(7) Nakai, H.; Konno, M.; Kosuge, S.; Sakuyama, S.; Toda, M.; Arai,
Y.; Obata, T.; Katsube, N.; Miyamoto, T.; Okegawa, T.; Kawasaki,
A. J. Med. Chem. 1988, 31, 84.
(8) Aycock, D. F. Org. Process Res. DeV. 2007, 11, 156.
(9) Pradhan, P. K.; Dey, S.; Jaisankar, P.; Giri, V. S. Synth. Commun.
2005, 35, 913.
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
861

mixture ready for step two (reduction). Indeed, safe control of
the process was obtained via implementation of a reverse
addition of the nitrobenzaldehyde 18 to the slurry of aqueous
ammonium chloride, Celite and iron in DMF at 90 °C.16
Experiments revealed that the method of addition of the nitro
derivative to the slurry of ammonium chloride in DMF was
critical for the good outcome of the reaction. In fact slow
addition via syringe pump failed to give complete conversion
of the nitro derivative to the amine, while portionwise addition
gave complete conversion. In addition, it was found that the
filtration of the slurry of step two ran smoothly when Celite17
(0.3 wt referred to 17) was added as a filter aid directly into
the reaction mixture.18 The volume of DMF used in the hot
filtration of the solid byproducts was lowered from 24 vol to 4
vol. Having implemented those changes on a 500-g scale-up
experiment resulted in the alkylated compound 18 in 96% a/a
(HPLC, 3% residual starting material), and the benzoxazine-
3-one 16 was isolated in 97% a/a (HPLC) and 69% yield
(Scheme 4).
Pilot Scale for Stage One. After having established a fairly
good process understanding of the critical parameters for stage
one (Scheme 4), a 10-kg-scale reaction was implemented at a
pilot scale. The telescoped alkylation procedure resulted in the
ester 18 in 98% a/a after workup, isolated as a DMF solution
(4 volumes). As part of the engineering controls for the nitro
group reduction, mechanical safeguards included the addition
of the previous DMF solution (4 volumes) through the head
tank, which was charged in four portions in order to prevent
full accidental mischarge of the starting nitro derivative 18
solution in DMF that could result in a hazardous situation. As
an additional precaution, during the addition of the nitro
derivative to the hot iron-Celite-DMF mixture, the reactor
temperature was set at 75 °C in jacket control (Figure 2). In
this way, the nitro derivative solution 18 was added in 10
portions through the reactor head tank, resulting in a controlled
process, and with a complete conversion to the desired ben-
zoxazine-3-one 16. Thus, step two gave benzoxazine-3-one 16
in 98% a/a (7.1 kg, 67% yield), as brown crystals. ICP for the
iron content was only 246 ppm, confirming that the iron content
on the isolated benzoxazine-3-one 16 was influenced by the
amount of DMF used in the waste cake washes, and that the
use of hydrochloric acid as antisolvent in the final recrystalli-
zation was beneficial to achieve this result.
Figure 1. Products of the reduction of the nitrophenoxyacetate 18.
sion to benzoxazine-3-one 16 (HPLC analysis). However the
isolated yield was not satisfactory, mainly due to low solubility
of benzoxazine-3-one 16 in DMF at room temperature. Indeed,
the iron reduction is producing large amounts of inorganic
byproducts, i.e., iron and ammonium salts, which may initiate
the crystallization of the benzoxazine-3-one 16. Thus, the
inorganic byproducts were removed by filtration, and the waste
cake was washed with hot DMF to ensure a good recovery of
the aldehyde. The filtration and waste cake washing were
performed at temperatures above 60 °C to prevent loss of the
aldehyde by crystallization, leading to a satisfactory mass
balance of the reaction. Then, an experiment was run on a 100-g
scale to evaluate the scalability of the two-step procedure
(Scheme 4). At this time the ester 18 was isolated and used as
crude for the synthesis of benzoxazinone 16. The iron was added
to the hot mixture of 18 in DMF and ammonium chloride, and
then brought to 90 °C. After complete conversion in 30 min
and filtration, DMF washes (24 volumes) allowed the isolation
of benzoxazinone 16 as a green solid, which we speculated to
have fairly high residual iron content. This solid was suspended
in DMF (8 vol) and hydrochloric acid 37 wt %/wt (2 vol) and
dissolved at 65 °C, kept for 2 h, and recrystallized at room
temperature. Upon filtration, the crystals were analyzed by
means of inductively coupled plasma (ICP) to determine the
residual iron content as 5297 ppm (0.74 wt %/wt).
This first attempt to scale up these reactions in a batch unit
highlighted the known concerns about the difficult and very
exothermic nitro-group reductions. In fact, there was a thermal
runaway of the reduction step that is only controlled by the
reflux of the solvent mixture (undetected in experiments on
small scale). On the basis of previous observations and the fact
that reduction of nitroaromatic compounds is recognized to be
exothermic,14 it was found that a chemical reaction hazard exists
for this reaction step and the reaction should not be scaled up
without further investigation. This prompted us to look into the
current synthesis conditions using reaction calorimetry.15 In
addition, large volumes of DMF, slow cake filtration, and a
high level of residual iron in the isolated dry crystals 16 were
still problems.
Process Development for Stage Two: Synthesis of Imi-
dazo[5,1-c][1,4]benzoxazine-3-carboxylate 9. Stage two gave
access to the tricyclic ester imidazo[5,1-c][1,4]benzoxazine-3-
carboxylate 9, including a Wittig reaction, an esterification, an
imidazole formation, a hydrolysis, and a reductive amination
(Scheme 5). Those reactions were at first conducted as separated
steps, and the products 26, 27, and 9 were isolated (Scheme
5). Similar results were obtained at laboratory scale while
telescoping the previously mentioned steps, and tricyclic ester
9 was isolated with comparable purity and yield. Initial attempts
to synthesize the vinylether 25 Via the Horner-Wadsworth-
A process improvement allowed the telescoping of steps one
and two (Scheme 4), namely the alkylation and reduction
reactions, isolating the aldehyde 16 with yield and purity
comparable to that obtained in the two-stage process with the
isolation of the intermediate 18: 68% overall yield and 98%
a/a as compared to 67% yield and 98% a/a in the telescoped
process. As a result, the first step was performed with MeTHF
and 18 was not isolated. Solvent swap to DMF (3 vol) gave a
(16) Grimm, J.; Liu, F.; Stefanick, S.; Sorgi, K. L.; Maryanoff, C. A. Org.
Process Res. DeV. 2003, 7, 1067.
(14) Duggan, P. J. Hazards XIII: Process Safety: The Future; Institute of
Chemical Engineers Symposium Series 141; Zeneca Specialties:
Manchester, U.K., 1997; pp 285-291.
(17) Celite 545 (particle size 0.02-0.1 mm), purchased from Merck.
(18) Anderson, N. G. Practical Process Research & DeVelopment;
Academic Press: San Diego, 2000; p 216.
(15) Hazard data will follow as a separate paper in due course.
862
•
Vol. 14, No. 4, 2010 / Organic Process Research & Development

Figure 2. Reactor temperature profiles for the reduction of the nitrobenzaldehyde 18 to benzoxazine-3-one 16 (stage one).
Scheme 5. Initial conditions for stage two for the synthesis of imidazo[5,1-c][1,4]benzoxazine-3-carboxylate 9
Scheme 6. Horner-Wadsworth-Emmons reaction of diethylphosphonate 23 and arylaldehyde 16
Emmons reaction failed to give the desired compound 25
(Scheme 6). As an example, the diethyl(ethoxymethyl)phos-
phonate 23, generated via an Arbuzov reaction of the trialky-
lphosphite and (bromoethoxy)methane, was reacted in THF in
various bases (lithium hexamethyldisilazane, n-butyllithium,
potassium tert-butoxide) at low temperature leading to complex
reaction mixtures out of which the vinylether 25 was isolated
in poor yields or not recovered.
The Wittig reaction was superior in terms of reaction profile
and conversion of the arylaldehyde 16 to the methylether 26
(Scheme 5). The previously illustrated arylaldehyde 16 was
submitted to Wittig reaction with (methoxymethyl)(triph-
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
863

solution of potassium tert-butoxide in THF. As a result, the
tert-butyl derivative 29 was only detected in trace quantities in
the reaction mixture. The crude reaction mixture resulting from
step two/three (Scheme 7) was treated with aqueous NaHCO3
and extracted with MeTHF. After two distillations, the solution
of methylether 27 in MeTHF was directly treated with aqueous
2 N hydrochloric acid at 60 °C to afford arylacetaldehyde 28.
This avoided the solvent swap from ethylacetate to acetone,
and the overall process was simplified by using only 78 vol vs
134 vol and affording tricyclic ethyl ester 9 (96% a/a by HPLC)
in 34% yield on 5-g scale. The sequence was then scaled up in
the Verona kilo-lab facility on a 200-g scale and the API
precursor 9 was obtained in 35% yield with good HPLC purity
(95.4% a/a).
The synthesis described in Scheme 7 was finally scaled up
in the Verona pilot plant to prepare 5.75 kg of ethyl 6-{2-[4-
(2-methylquinolin-5-yl)piperazin-1-yl)]ethyl}-4H-imidazo[5,1-c]-
[1,4]benzoxazine-3-carboxylate 9. Despite the fact that the yield
(29%) was lower than the one obtained in the kilo-lab exp-
eriment, the quality of the material (94% a/a by HPLC) was in
the expected range. The quality of the 21 wt % sodium ethoxide
in ethanol was suspected to be the principal cause of the lower
yield: tarry material was in fact present in the pilot batch.
Unfortunately, the base was released for use on the basis of its
certificate of analysis, and no use test was performed. In order
to have a scalable process, stage two was modified as follows:
the Wittig reaction was carried out at 20 °C instead of -10
°C; 1.6 M potassium tert-butoxide solution in THF was used
in step one instead of 1 M potassium tert-butoxide in THF;
MeTHF replaced THF as a reaction solvent in step one (Scheme
7), and it was used during the workup of step three instead of
ethylacetate and for step four (hydrolysis); 21 wt %/wt sodium
ethoxide in ethanol replaced potassium tert-butoxide in THF
in step three to avoid the formation of the tert-butyl ester side
product 29; the total amount of solvents was reduced from 134
vol to 78 vol; the extraction in acidic water of step four with
removal of the TPPO was studied to give a higher recovery of
material by working at pH 2 or lower.
Figure 3. tert-Butyl 6-[2-methoxyethenyl]-4H-imidazo[5,
1-c][1,4]benzoxazine-3-carboxylate 29, major side product of
stage two.
enyl)phosphonium chloride and 2 equiv of potassium tert-
butoxide in THF to give the vinylether 26 as (3:1) E/Z isomer
mixture. Unfortunately, the triphenylphosphine oxide (TPPO),
side product of the Wittig step, could not be removed at this
point and was only separated during the last recrystallization
of tricyclic compound 9 in acetone, thus making the analy-
sis of the process with HPLC (at 220 nm) more difficult because
of the high absorbance of the triphenylphosphine oxide.
Chlorodiethylphosphate, ethylisocyanoacetate, and an additional
equivalent of potassium tert-butoxide were added at low
temperature (0 to -10 °C) to the previously isolated vinylether
26 to facilitate the synthesis of the imidazobenzoxazine-3-
carboxylate intermediate 27, according to known literature
conditions.19 After workup, the solution of 27 was treated with
2N hydrochloric acid to reveal the aldehyde 28, final reductive
amination gave piperazinylquinaldine derivative 9 in 22%
overall yield (92% a/a by HPLC) after recrystallization from
acetone. As highlighted before, the triphenylphosphine oxide
was partially removed during the latter acetone recrystallization
of the imidazo[5,1-c][1,4]benzoxazine-3-carboxylate deriva-
tive 9.
First attempts at scale-up of the stage two process depicted
in Scheme 5 gave a poorly reproducible process and overall
yields were ranging from 13% to 28% yield. One of the main
drawbacks of the synthesis shown in Scheme 5 was the very
large volume of solvents and reagents used (134 vol in relation
to the compound 16). Ethyl acetate (bp 76.5 °C) was used during
the workup for the isolation of methylether 27, followed by a
tedious solvent swap to run the next hydrolysis step in acetone
(bp 56 °C). Finally, variable amounts of the tert-butyl ester
derivative 29 (Figure 3) were formed along with the desired
methylether derivative 27 in the imidazole formation step. Thus,
the development focused on the consistency and robustness of
stage two (Scheme 5).
The previously mentioned route (Scheme 1) allowed the
synthesis of the intermediate grade 11 from the ester 9 in a
two-step process (Scheme 8). On a plant scale the ethyl ester 9
was converted in the presence of potassium hydroxide to the
corresponding potassium carboxylate 10, and the latter was then
reacted with hexamethyldisilazane (HMDS) and O-(benzotria-
zol-1-yl)-N,N,N′N′-tetramethyluronium tetrafluoroborate (TBTU)
to give 11 (IG) in high overall yield (∼98 wt %/wt).
It was found that the first step could be carried out at room
temperature (instead of 0 °C) in MeTHF as solvent. It was also
demonstrated that MeTHF could be used as both extraction
solvent for the workup of the second step and solvent in the
biphasic acidic hydrolysis of the vinylether 27 (step two/three).
Potassium tert-butoxide (1.6 M) was also used instead of 1 M
solution in THF in the first step, thus lowering the total volumes
of the process.
Step two was carried out at -10 °C, which simplified the
operations and reduced the processing time required in the plant.
To lower the total volumes of the process and to avoid the
formation of side product 29 (Figure 3), 21 wt % sodium
ethoxide in ethanol was introduced as a base in place of 1 M
The main drawback of the approach shown in Scheme 8
was the high value of residue on ignition (∼17 wt %/wt) in the
intermediate grade 11 due to the coprecipitation of the potassium
tetrafluoroborate side product along with the tricyclic amide 11.
After a tedious and poorly efficient recrystallization to remove
the inorganic impurity, the carbamide 11 was recovered in
modest yield (∼57 wt %/wt). Moreover, hydrolysis and amide
coupling were required to isolate intermediate 11. Therefore, a
new and straightforward synthesis of 11 starting from 9 was
investigated. First attempts to directly transform the ethyl ester
9 into 11 were carried out on a laboratory scale with 7 N
ammonia in methanol under microwave irradiation (Scheme
(19) Erker, T.; Handler, N. J. Heterocycl. Chem. 2002, 39, 645.
864
•
Vol. 14, No. 4, 2010 / Organic Process Research & Development

Scheme 7. Pilot scale-up of stage two, five-step synthesis of imidazo[5,1-c][1,4]benzoxazine-3-carboxylate 9
Scheme 8. Synthesis of 11 via TBTU coupling
Scheme 9. Attempted direct amidation of ester 9 with
as the amidation reagent. In search of a straightforward synthesis
of the primary amide 11 from the ethyl ester 9, the latter direct
transformation looked extremely attractive for its simplicity and
convenience, since atmospheric pressure and low temperatures
are employed. The formamide-induced amidation of 9 was
initially carried out in the presence of two equiv of 21 wt %
sodium ethoxide in ethanol as the base. Formamide (10 vol)
was used as the solvent of the reaction. The mixture was
magnetically stirred for 23 h at 60 °C, cooled to room
temperature, and diluted with methanol (10 vol), and the desired
product was directly collected by filtration in 73% yield.
A variety of alkoxide bases was screened: powder potassium
tert-butoxide, 30 wt % sodium methoxide in methanol, 21 wt
% sodium ethoxide in ethanol, and 1 M potassium tert-butoxide
in THF. All the bases were effective, affording 11 in high yield
and purity. Sodium methoxide (30 wt %) in methanol was
finally chosen as the base, to effect volume reduction of the
reaction mixture. Amide 11 was collected by filtration at room
temperature after dilution of the reaction mixture with methanol.
The resulting sluggish filtration was overcome by the use of a
mixture of formamide/methanol (Table 1). Two equivalents of
30 wt % sodium methoxide in methanol was used in this series
of experiments. In addition, when using methanol as a cosolvent,
reactions were more than 7 times faster (3 h instead of 22-30
h). Moreover, the filtration times were significantly lower, as
expected.
ammonia in methanol
9). Pure 11 was recovered in only modest yield (38%). On the
other hand, the reaction was extremely slow under conventional
heating (less than 2% a/a of 11 by HPLC at 150 °C after 1.5 h,
while increasing the temperature to 175 and 190 °C under
pressure led only to complex reaction mixtures).
High temperature and high pressure made the latter approach
(Scheme 9) not easily applicable for the scale-up. Because of that,
a facile and extremely convenient methodology for the synthesis
of 11 was found to overcome the issues described above.
Direct amidation of esters with amides in the presence of
methoxide ion is a pretty well-known transformation.20 The
preparation of primary amides requires the use of formamide
(20) Allred, E. L.; Hurwitz, M. D. J. Org. Chem. 1965, 30, 2376.
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
865

Table 1. Effect of the presence of different amounts of
methanol in the reaction mixture on the filtration time; filter
porosity was 20 µm (2-g scale of 9)
electrospray ion mode. Commercially available reagents and
solvents were purchased from ordinary chemical suppliers and
used without purification.
Observed retention times were as follows: 3-nitrosalicylal-
dehyde 17 (2.99 min), nitrobenzaldehyde 18 (4.28 min),
benzoxazine-8-carbaldehyde 16 (2.63 min), imidazo[5,1-c]-
[1,4]benzoxazine-3-carboxylate 9 (3.02 min), imidazo[5,1-c]-
[1,4]benzoxazine-3-carboxamide 11 (2.67 min).
formamide/
MeOH (v/v)
filtration
time (min)
11, yield
(%)
HPLC
(%a/a)
entry
1
2
3
4
10/0
4/1
5/5
1/4
22
10
3.5
4
80.1
69.1a
82.0
84.2
96.5
96.2
97.0
97.2
3-Oxo-3,4-dihydro-2H-1,4-benzoxazine-8-carbaldehyde
(16). Step One. To a suspension of 3-nitrosalicylaldehyde (10
kg, 1 wt, 1 equiv) in MeTHF (3 vol) at 20 °C, DIPEA (2 vol,
1.92 equiv) is added followed by ethylbromoacetate (0.8 vol,
1.2 equiv). The mixture is stirred at reflux (ca. 80 °C, internal
temperature) for 30 min. The reaction mixture is cooled to room
temperature in 20 min, MeTHF (2 vol) is added followed by
water (5 vol). After 5 min at room temperature, the organic
phase is collected and the aqueous phase is back-extracted with
MeTHF (2 vol). The combined organic phases are washed with
2 M hydrochloric acid (5 vol) and water (5 vol). The organic
phase is concentrated under vacuum to 2.5 vol and DMF (3
vol) is added. The resulting mixture is concentrated under
vacuum to 4 vol and is called Solution 1, this is used for the
next step without further purification.
a Lower yield was obtained due to loss of material passing through the filter
during the filtration step.
Scheme 10. Direct conversion of the ester 9 into amide 11
Step Two. Celite (0.3 wt) and iron powder-325mesh (0.8
wt, 2.4 equiv) is successively added to a vigorously stirred
solution of DMF (3 vol) and aqueous ammonium chloride (1
vol) at 20 °C. The resulting suspension is warmed to 85 °C
(internal temperature) over 40 min. Solution 1 is added
portionwise over 20-30 min (jacket controlled, jacket temp
75 °C) maintaining the internal temperature below 100 °C. At
the end of the addition the reaction mixture is stirred at 85 °C
for 30 min. DMF (10 vol) is added at 80 °C. The suspension is
filtered at 80 °C and the cake is washed with DMF (4 vol) at
80 °C. The filtrate is concentrated under vacuum to 3 vol and
then 2 M aqueous hydrochloric acid (5 vol) is added at 40 °C.
The resulting suspension is stirred at 40 °C for 1 h then diluted
with water (5 vol) and stirred at 5 °C for 5 h. The resulting
suspension is filtered and the cake is washed with water (3× 3
vol). The cake is dried on the filter for 1 h and then in the
vacuum oven at 50 °C for 20 h to give 3-oxo-3,4-dihydro-2H-
1,4-benzoxazine-8-carbaldehyde 16 (7.1 kg, 67%th yield, 71
wt %/wt) as a brown solid. 1H NMR (600 MHz, d6-DMSO) δ
4.75 (s, 2H), 7.07 (t, 1H), 7.17 (dd, 1H), 7.33 (dd, 1H), 10.28
(s, 1H), 10.94 (s, 1H).
Ethyl 6-{2-[4-(2-methylquinolin-5-yl)piperazin-1-yl)]ethyl}-
4H-imidazo[5,1-c][1,4]benzoxazine-3-carboxylate (9). 3-Oxo-
3,4-dihydro-2H-1,4-benzoxazine-8-carbaldehyde 16 (7.0 kg, 1
wt) and methoxymethyl-triphenylphosphonium chloride (1.1
equiv, 2.13 wt) are suspended at 18 °C in MeTHF (4 vol).
Potassium tert-butoxide (1.6 N) in THF (2.4 equiv, 8.4 vol) is
added dropwise over 30 min, keeping the temperature below
30 °C during the addition. The reaction mixture is stirred for
30 min at 18 °C, and then it is cooled to -10 °C. Diethylchlo-
rophosphate (1.4 equiv, 1.14 vol) is added in 10 min, maintain-
ing the temperature below -3 °C. The reaction mixture is stirred
at -10 °C for 15 min, and then ethylisocyanoacetate (1.1 equiv,
0.68 vol) is added. Sodium ethoxide (21 wt %) in EtOH (1.2
equiv, 2.5 vol) is added dropwise in 20 min, keeping the
A second series of experiments was then carried out to
investigate the effect of different amounts of base (30 wt %
sodium methoxide in methanol) on the efficiency of the reaction.
A mixture of formamide/methanol (1:4) was used as the solvent.
An excess of base is required to maximize the recovery of 11.
The optimized reaction conditions were then scaled up on
800-g scale. Therefore, tricyclic ester 9 was reacted at 60 °C in
a mixture formamide/methanol (1:4) (10 vol) in the presence
of two equiv of 30 wt % sodium methoxide in methanol. After
diluting with methanol and cooling to room temperature, the
desired intermediate grade 11 was isolated in good yield (86%)
and high purity (98% a/a by HPLC).
In conclusion, process development and scale-up synthesis
of 11 was accomplished (Scheme 11) by a new route that
represents an efficient and economical process starting from
the easily accessible 3-nitrosalicylaldehyde. The overall yield
was improved from 8% (Scheme 1) to 17% (Scheme 11), and
the number of stages to the API was lowered by almost half
(from 7 to 4), thus reducing number and quantity of solvents
and the cycle times. Finally, the route performed well at a pilot
scale, and no major issues were identified, confirming that the
novel approach is feasible for future deliveries and can be
considered for manufacturing.
General Remarks
NMR spectra were recorded on 600 MHz Varian Inova 600
for 1H NMR. HPLC analysis of the intermediates and reaction
monitoring was carried out on Agilent Series 1200 (Agilent).
Generic acidic HPLC method used: column type Luna C18;
mobile phase A: 0.05% TFA/water and B: 0.05% TFA/
acetonitrile; gradient: 0 min 100% A to 8 min 95% B; flow 1
mL/min; column temperature 40 °C; detector UV DAD @220
nm. Mass spectra analyses were performed on an Agilent 1100
LC/MS, with the mass spectrometer operating in positive
866
•
Vol. 14, No. 4, 2010 / Organic Process Research & Development

Scheme 11. Optimized synthesis of 11
temperature below -3 °C. The reaction mixture is stirred at
-10 °C for 30 min. Sodium hydrogencarbonate (5 wt %/wt)
(5 vol) is added, and the mixture is concentrated to 11 vol. The
residue is diluted with MeTHF (5 vol) and concentrated to 11
vol. The residue is diluted in MeTHF (5 vol) and concentrated
to 11 vol. The crude is diluted with 5 wt %/wt NaHCO3 (5
vol), and the separated aqueous phase is back-extracted with
MeTHF (5 vol). The organics are combined. Hydrochloric acid
(2 N, 5 vol) is added, and the mixture is stirred at 60 °C (internal
temperature) for 1 h. The mixture is cooled down to 18 °C.
Toluene (5 vol) is added. The organic phase is extracted with
2 N hydrochloric acid (5 vol). The combined aqueous phases
are diluted with 13% Na2CO3 (6 vol) (target: pH 1.5 to 2) and
extracted with dichloromethane (2 × 5 vol). The combined
organics are washed with water (3 vol). The resulting dichlo-
romethane solution is called Solution A.
Piperazinylquinaldine 5 (6.3 kg, 0.7 equiv, 0.9 wt) and
sodium triacetoxyborohydride (1 equiv, 1.2 wt) are suspended
in dichloromethane (6 vol). The mixture is allowed to stir at
room temperature for 10 min. Solution A is added dropwise
over 60 min, and the resulting mixture is stirred for 20 min at
room temperature. NaHCO3 (5%, 5 vol) is added, and the
resulting mixture is stirred for 30 min at room temperature. The
dichloromethane solution is evaporated to 5 vol. Acetone (15
vol) is added and the mixture evaporated to 10 vol. The
suspension is diluted with acetone (10 vol) and the mixture
evaporated to 10 vol. The suspension is stirred at room
temperature for 15 h. The solid is collected by filtration, washed
with acetone (2 vol), and dried at 40 °C under reduced pressure
(5.75 kg, 29%th, 82 wt %/wt). 1H NMR (600 MHz, d6-DMSO)
δ 1.31 (t, 3H), 2.63 (s, 3H), 2.64 (m, 2H), 2.75 (m, 4H), 2.87
(t, 2H), 3.03 (m, 4H), 4.27 (q, 2H), 5.55 (s, 2H), 7.10 (m, 2H),
7.24 (dd, 1H), 7.38 (d, 1H), 7.59 (m, 2H), 7.77 (dd, 1H), 8.35
(d, 1H), 8.60 (s, 1H).
6-{2-[4-(2-Methylquinolin-5-yl)piperazin-1-yl)]ethyl}-4H-
imidazo[5,1-c][1,4]benzoxazine-3-carboxamide (11). Sodium
methoxide in methanol (30 wt %) (2 equiv, 0.77 vol, 0.73 wt)
was added at room temperature to a vigorously stirred mixture
of 9 (800 g, 1 wt) in formamide/methanol 1:4 (10 vol). The
reaction mixture was heated up to 60 °C and stirred for 3 h.
Methanol (5 vol) was added, and the resulting mixture was
stirred for 1 h at 60 °C and then cooled down to room
temperature over 1 h and stirred overnight. The mixture is
filtered, and the cake is washed with methanol (3 × 2.5 vol)
and dried at 40 °C under reduced pressure (644 g, 86%th, 80
1
wt %/wt). H NMR (600 MHz, CD3COOD) δ 2.98 (s, 3H),
3.29 (dd, 2H), 3.54 (m, 4H), 3.55 (m, 6H), 5.66 (s, 2H), 7.14
(t, 1H), 7.30 (dd, 1H), 7.50 (d, 1H), 7.66 (dd, 1H), 7.79 (d,
1H), 7.97 (t, 1H), 8.00 (d, 1H), 8.39 (s, 1H), 9.06 (d, 1H).
Acknowledgment
We are grateful to Stefano Provera and Lucilla Turco for
the execution and assignments of NMR spectra.
Received for review April 15, 2010.
OP100103V
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
867