Production of (R)-aminoglutethimide: A new route from 1 -chloro-4-nitrobenzene

Article abstract of DOI:10.1021/op9900075

The development of a short, safe and enantioselective route for the preparation of (R)-aminoglutethimide is described. The process was designed for economic large-scale manufacture of the bulk drug substance to acceptable quality standards, to allow clinical evaluation of the single enantiomer over the existing racemate. (R)-Aminoglutethimide was prepared from 1-chloro-4-nitrobenzene using a six-stage synthetic sequence, via chemoresolution of key intermediate racemic 4-cyano-4-(4-nitrophenyl)hexanoic acid using (-)-cinchonidine. The process allowed for preparation of several kilograms of the precursor (R)-nitroglutethimide, to cGMP at pilot-plant scale, along with demonstration of the final hydrogenation step to (R)-aminoglutethimide in the laboratory. This route avoids the problems of hazardous nitration technology, and therefore regio-isomer contamination of the product, associated with other procedures. The resolution chemistry described represents an improvement on literature procedures. Optimisation of the asymmetric Michael addition offers an attractive alternative approach.

Full text of DOI:10.1021/op9900075

Organic Process Research & Development 1999, 3, 442−450
Production of (R)-Aminoglutethimide: A New Route from
1-Chloro-4-nitrobenzene
Michael J. Bunegar,‡ Ulrich C. Dyer,† Graham R. Evans,† Richard P. Hewitt,‡ Stephen W. Jones,*,† Neil Henderson,†
Christopher J. Richards, Sivadasan Sivaprasad,† Benjamin M. Skead,† Mark A. Stark,§ and Eric Teale‡
Celltech-Chiroscience p.l.c., Cambridge Science Park, Milton Road, Cambridge, CB4 0WE, UK, Hickson and Welch Ltd.,
Wheldon Road, Castleford, West Yorkshire, WF10 2JT, UK, and Department of Chemistry, UniVersity of Wales,
Cardiff, P.O. Box 912, Cardiff, CF1 3TB, Wales
Abstract:
However, it was unknown if these debilitating side effects
could be reduced or even eliminated by treatment with the
(R)-enantiomer as opposed to the racemate. If this were the
case then the (R)-enantiomer would offer both increased
efficacy and reduced side effects over the existing racemate.
Therefore, at Celltech-Chiroscience we established a drug
development programme to establish proof of principle for
the use of (R)-aminoglutethimide, with an improved thera-
peutic index over the racemate. Within the Process R & D
department we were responsible for material supply for the
programme. Initial supplies of the bulk drug were supplied
by classical resolution of racemic aminoglutethimide with
L-tartaric acid. This approach provided rapid access to
material for clinical trials. However, for long-term supply
this resolution was inefficient, requiring multiple crystalli-
sations to ensure adequate optical purity. Furthermore,
recycle of the unwanted enantiomer would be difficult.
Therefore, we required a short, safe enantioselective route
to (R)-aminoglutethimide, which could be developed for the
economic large scale manufacture of the bulk single enan-
tiomer drug substance.
The development of a short, safe and enantioselective route for
the preparation of (R)-aminoglutethimide is described. The
process was designed for economic large-scale manufacture of
the bulk drug substance to acceptable quality standards, to
allow clinical evaluation of the single enantiomer over the
existing racemate. (R)-Aminoglutethimide was prepared from
1-chloro-4-nitrobenzene using a six-stage synthetic sequence,
via chemoresolution of key intermediate racemic 4-cyano-4-(4-
nitrophenyl)hexanoic acid using (-)-cinchonidine. The process
allowed for preparation of several kilograms of the precursor
(R)-nitroglutethimide, to cGMP at pilot-plant scale, along with
demonstration of the final hydrogenation step to (R)-amino-
glutethimide in the laboratory. This route avoids the problems
of hazardous nitration technology, and therefore regio-isomer
contamination of the product, associated with other procedures.
The resolution chemistry described represents an improvement
on literature procedures. Optimisation of the asymmetric
Michael addition offers an attractive alternative approach.
Introduction
Racemates of 3,3-di-substituted glutarimides such as
aminoglutethimide (3-ethyl-3-(4-aminophenyl)-piperidine-
2,6-dione) (1) and rogletimide (3-ethyl-3-(4-pyridyl)-piperi-
dine-2,6-dione) (2) have been shown to be effective for the
treatment of hormone-dependent breast cancer.1,2 The mode
of action of these compounds is considered to be by
inhibition of the aromatase enzyme that catalyses the
formation of oestrogens from androgens. Therefore, the
compounds inhibit tumours whose growth is promoted by
oestrogens.3 Further studies have shown that the (R)-
enantiomers of these compounds are much more potent
aromatase inhibitors than their corresponding (S)-enanti-
omers.4 In addition, medical evidence has also shown that
some patients suffer acute CNS side effects during treatment,
including lethargy, dizziness, and an all-over-body rash.
Synthetic Strategy
Previous preparations of (R)-aminoglutethimide have been
reported utilising both biochemical5 and chemical6,7 resolu-
tion technology as well as synthetic approaches to heteroaryl8
and fluoroaryl analogues.9 The key issues to be addressed
in the prior art were:
1. Introduction of the amino group via hydrogenation of
a nitro function, incorporated using nitration methodology
(5) Fogliato, G.; Fronza, G.; Fuganti, C.; Grasselli, P.; Servi, S. J. Org. Chem.
1995, 60, 5693.
(6) Finch, N.; Dziemian, R.; Cohen, J.; Steinetz, B. G. Experientia 1975, 31,
1002; U.S. Patent No. 3,944,671.
† Celltech-Chiroscience p.l.c.
‡ Hickson and Welch Ltd.
§ University of Wales.
(1) Smith, I. E.; Fitzharris, B. M.; McKinna, J. A.; Fahmy, D. R.; Nash, A. G.;
Neville, A. M.; Gazet, J.-C.; Ford, H. T.; Powles, T. J. Lancet 1978, 2,
646.
(2) Foster, A. B.; Jarman, M.; Leung, C.; Rowlands, M. G.; Taylor, G. N.;
Plevey, G. R.; Sampson, P. J. Med. Chem. 1985, 28, 200.
(3) Ogbunude, P. O. J.; Aboul-Enein, H. Y. Chirality 1994, 6, 623.
(4) Graves, P. E.; Salhanick, H. A. Endocrinology 1979, 105, 52.
(7) Achmatowicz, O.; Malinowska, I.; Szechner, B.; Maurin, J. K. Tetrahedron
1997, 53, 7917.
(8) McCague, R.; Jarman, M.; Rowlands, M. G.; Mann, J.; Thickett, C. P.;
Clissold, D. W.; Neidle, S.; Webster, G. J. Chem. Soc., Perkin Trans. 1
1989, 196.
(9) Hammond, G. B.; Plevey, R. G.; Sampson, P.; Tatlow, J. C. J. Fluorine
Chem. 1988, 40, 81.
442
•
Vol. 3, No. 6, 1999 / Organic Process Research & Development
10.1021/op9900075 CCC: $18.00 © 1999 American Chemical Society and The Royal Society of Chemistry
Published on Web 10/28/1999

Scheme 1
which would be both nonregioselective and hazardous to
operate at scale.5
2. Enantioselection via low-yielding resolution technology.
3. Lengthy synthetic sequences not suitable for operation
at scale.
Therefore, we required a synthetic strategy which incor-
porated para selectivity along with an efficient asymmetric
step, in a short sequence suitable for scale-up, from low-
cost starting materials. This would be required as a credible
material supply route upon completion of the pivotal clinical
trials, which imposed challenging time constraints on the
development programme.
After a retrosynthetic analysis we adopted a strategy
aimed at developing a cost-effective racemic synthesis,
incorporating a convenient resolution step. Our approach was
developed using 1-chloro-4-nitrobenzene as a convenient
starting material. This was available at low cost with less
than 2% regioisomer contamination. Therefore the correct
regiochemistry was established at the outset, and potentially
hazardous nitration chemistry was avoided. Thus, (R)-
aminoglutethimide (1) was prepared from 1-chloro-4-ni-
trobenzene using the six-stage synthetic sequence shown in
Scheme 1. Aromatic nucleophilic substitution with methyl
cyanoacetate, followed by ethylation, Michael addition and
hydrolysis gave the key intermediate racemic 4-cyano-4-(4-
nitrophenyl)hexanoic acid (7). A screen of suitable resolving
agents identified an efficient classical resolution and subse-
quent cyclisation and reduction gave (R)-aminoglutethimide
(1). The route utilised low-cost readily available materials
and allowed for safe, manageable processing. The final
Vol. 3, No. 6, 1999 / Organic Process Research & Development
•
443

product and all isolated intermediates are crystalline solids,
and stages 1-5 of the process were successfully scaled up
in the pilot plant.
Stage 2. Alkylation of Methyl 2-(4-nitrophenyl)cy-
anoacetate (3). It had been reported that alkylation of
analogues of methyl 2-(4-nitrophenyl)cyanoacetate (3) could
be achieved with diethyl sulphate and potassium carbonate
in refluxing acetone.10 Early laboratory investigations using
diethyl sulphate and potassium carbonate in 10 vol 2-bu-
tanone (MEK) at reflux (80 °C) allowed for isolation of
methyl 2-cyano-2-(4-nitrophenyl)butyrate (4) as a crystalline
solid. The product was isolated by an extractive work-up
followed by evaporation to dryness under vacuum. However,
the work-up was made difficult due to intensely dark-
coloured organic/aqueous interfaces. Despite this the process
was rapidly scaled to a 20 L working volume and gave yields
of 41-55% and crude purities of 88-90% by HPLC (PAR),
which increased to 97% after recrystallisation from 2-pro-
panol.
Before the process could be operated in the pilot plant,
product isolation issues had to be addressed. Therefore, a
modified process was developed. Potassium carbonate was
replaced with triethylamine which was added to a preheated
two-phase mixture of the precursor methyl 2-(4-nitrophenyl)-
cyanoacetate (3), diethyl sulphate and 100-120 °C petroleum
ether (SPB-3), at 90-95 °C. The reverse addition of
triethylamine was beneficial, as it did not impact on reaction
processing or product quality, and handling of the toxic
alkylating agent was reduced to a minimum. The petroleum
ether did not solubilise the reagents and was effectively used
as an inert diluent to increase the mobility of the reaction
mixture. This ensured efficient heat transfer and therefore
safe operation of the process, as a hazard assessment had
revealed thermal instability of both starting material and
product.13 Importantly, the lengthy aqueous work-up was also
avoided by isolation of the product directly from the reaction
mixture. Initially, excess diethyl sulphate was destroyed by
the addition of water, followed by addition of 2-propanol
and crystallisation of the product on cooling. Two 82.6 kg
batches of methyl 2-(4-nitrophenyl)cyanoacetate (3) were
processed. The optimisations resulted in improved yields of
59 and 73% of methyl 2-cyano-2-(4-nitrophenyl)butyrate (4),
purity 94% by HPLC (PAR), with no detrimental effect on
impurity profile.
Results and Discussion
Stage 1. Nucleophilic Substitution of 1-Chloro-4-
nitrobenzene. Nucleophilic substitutions of substituted aryl
nitro compounds with carbanions have been reported and
product yields maximized when alkyl cyanoacetates were
used in polar solvents in the presence of strong base.9-11
Initial investigations into the nucleophilic substitution of
1-chloro-4-nitrobenzene revealed optimum conditions in-
volved the use of 2 equivalents (equiv) of methyl cyanoac-
etate/2 equiv potassium carbonate/10 volumes (vol) DMF
at elevated temperatures (>100 °C). However, product
isolation was one of the major challenges. Recovered yields
were good (80-90%) when the extractive work-up was
replaced by an aqueous precipitation of the crystalline
product, methyl 2-(4-nitrophenyl)cyanoacetate (3). This
circumvented isolation problems caused by the use of ethyl
cyanoacetate in early investigations, which gave the ethyl
analogue as a difficult to handle semisolid product.
Further improvements to the work-up were required as
neutralisation at ambient temperature with concentrated HCl
caused extensive foaming due to carbon dioxide liberation,
together with the formation of the byproduct, 4-nitrobenz-
ylcyanide (15%), produced by decarboxylation. Both of these
problems were overcome by slow neutralisation of the
reaction mixture with dilute HCl at 10 °C. These combined
process improvements allowed successful scale-up into 20
L laboratory glassware, based on an 800 g input of 1-chloro-
4-nitrobenzene, to give methyl 2-(4-nitrophenyl)cyanoacetate
(3) in 83-92% yield, 97-99% HPLC purity by peak area
ratio (PAR).
Upon transfer to the pilot plant both the methyl cyano-
acetate and potassium carbonate charge were reduced to 1.75
equiv, and the DMF charge was reduced to 4 vol. This
improved the volume efficiency, and therefore batch size,
aided the aqueous precipitation of product from DMF, and
led to a further reduction in foaming of the reaction mixture
on neutralisation. Dust explosion tests carried out on the
product revealed it to be a class 2 dust hazard (medium risk)
which required nitrogen purging, whilst discharging from
the drier, and charging the product into anti-static bags.12
Three consecutive 48 kg batches of 1-chloro-4-nitrobenzene
were converted in 63, 73, and 94% yields, and 89-93%
HPLC purity (PAR). The presence of unreacted 1-chloro-
4-nitrobenzene (up to 3.4% by HPLC) caused excessive
drying times, as it sublimed into the line connecting the filter/
dryer to the vacuum system partially blocking it.
Stage 3. Conversion of Methyl 2-Cyano-2-(4-nitro-
phenyl)butyrate (4) to 4-Cyano-4-(4-nitrophenyl)hexanoic
Acid (7). The hydrolysis and decarboxylation of methyl
2-cyano-2-(4-nitrophenyl)butyrate (4) to give 2-(4-nitrophe-
nyl)butyronitrile (5) was extremely facile. This gave a
product superior to that prepared by an alternative approach
involving ethylation of 4-nitrobenzylcyanide. The alternative
approach was attractive as it obviated stages 1 and 2.
However 4-nitrobenzylcyanide was not commercially avail-
able and gave 5 contaminated with di-alkylated material.
Michael addition of ethyl acrylate to 2-(4-nitrophenyl)-
butyronitrile (5) gave ethyl 4-cyano-4-(4-nitrophenyl)hex-
anoate (10). However, both products 5 and 10 were isolated
(10) Maeda, R.; Ohsugi, E.; Fujioka, T.; Hirose, K. Chem. Pharm. Bull. 1983,
31, 3424.
(11) Makosza, M.; Podraza, R.; Kwast, A. J. Org. Chem. 1994, 59, 6796.
(12) Dust explosion tests were carried out on a sieved sample of 3, particle size
<150 µm, in a Hartmann tube. Worst case results were maximum
overpressure ) 8.56 bar, maximum rate of pressure rise ) 7128 bar sec-1
,
lower explosive limit ) 413 gm-3, kst (dust explosion class standard) )
234 barmsec-1, kst values of between 200 and 300 barmsec-1 are categorised
as class 2, strongly explosible, by Lunn, G. Guide to Dust Explosion
PreVention and Protection Part 1 - Venting, 2nd ed.; Gulf Publishing
Company, 1992; Vol. 1. We thank Miss Sarah Birkby of Hickson and Welch
for these results.
(13) DSC analysis (Mettler Toledo DSC 12E) of 3 revealed exothermic
decomposition, onset 210 °C, -46 J/g, and also above 282 °C. DSC analysis
of 4 revealed exothermic decomposition, onset 207 °C, -12 J/g, and also
above 283 °C. We thank Dr. John O’Rourke of Chiroscience for these
results.
444
•
Vol. 3, No. 6, 1999 / Organic Process Research & Development

Scheme 2
Scheme 3
as oils which would not crystallise, and differential scanning
calorimitry (DSC) had revealed decomposition on heating.14
Therefore the materials were unsuitable for isolation or
purification by large scale batch distillation. Hence a strategy
of hydrolysis/decarboxylation of methyl 2-cyano-2-(4-nitro-
phenyl)butyrate (4), followed by in situ Michael reaction and
hydrolysis, was investigated. This gave 4-cyano-4-(4-nitro-
phenyl)hexanoic acid (7) as a granular crystalline solid in a
single stage, avoiding isolation of 2-(4-nitrophenyl)butyro-
nitrile (5) and ethyl 4-cyano-4-(4-nitrophenyl)hexanoate (10)
(Scheme 2).
After considerable development the hydrolysis/decar-
boxylation reaction was performed in aqueous methanolic
potassium carbonate solution over 10 min at 45 °C. The
preferred Michael reagent methyl acrylate was then added
directly to the butyronitrile intermediate, and the resultant
reaction was complete after 5 h at 45 °C. A solution of
sodium hydroxide was then added to promote hydrolysis of
the intermediate methyl ester (6) to the sodium salt of the
racemic acid (7). A dichloromethane wash of the basic
solution removed any non-acidic organic impurities. After
acidification, followed by extraction into hot toluene, the
racemic acid (7) was crystallised directly. The three-step
process was scaled into 20 L glassware, and crystallised
product yields were good, at 59-63%, with HPLC purities
of 96-99% (PAR).
As a further improvement to the procedure the quantity
of potassium carbonate was reduced, which significantly
improved the impurity profile; however, this was at the
expense of reaction time, which now took in excess of 15 h
to reach completion. Therefore, a mixed base system
involving potassium carbonate and triethylamine was devel-
oped which not only reduced reaction times to more workable
levels ( ∼10 h) but also resulted in increased yields and
purity. This combined three-step process was transferred to
the pilot plant with typical yields in the region of 65-70%
and product purity >98% by HPLC (PAR). A total of 71.5
kg of racemic acid (7) was manufactured from two 50.8 kg
batches of methyl 2-cyano-2-(4-nitrophenyl)butyrate (4)
using this procedure.
Asymmetric Michael Addition. In an alternative approach
an asymmetric Michael addition was proposed which would
avoid the need for resolution-based technology. A series of
cationic 2,6-bis(2-oxazolinyl)phenylpalladium(II) complexes
had been developed by a group at Cardiff University that
effectively catalysed the Michael addition of R-cyano car-
boxylates to methyl vinyl ketone.15 Preliminary studies
encouragingly revealed an ee of 30% was achieved in the
addition of 2-(4-nitrophenyl)butyronitrile (5) to acrylonitrile,
using the cationic (S,S)-2,6-bis(4′-methylenecyclohexyl-2′-
oxazolinyl)phenyl-aquopalladium(II) hexafluoroantimonate
complex (12), to give 2-ethyl-2-(4-nitrophenyl)pentanedini-
trile (11) (Scheme 3).16 However, time constraints prevented
further development of this approach.
4-Cyano-4-(4-nitrophenyl)pentanoic Acid (13) Contami-
nation. It was observed that the racemic acid (7) prepared
in the pilot plant contained a new impurity at a level of ∼2%
by HPLC (PAR). This was of particular concern as it carried
through subsequent processing to appear at ∼0.3% in the
(15) Stark, M. A., Richards, C. J. Tetrahedron Lett. 1997, 38, 5881. Stark, M.
A., Jones, G., Richards, C. J. Cationic 2,6-Bis(2-oxazolinyl)phenylpalladium-
(II) Complexes: Catalysts for the Asymmetric Michael Reaction, in press.
(16) Chiroscience funded LINK award to M. A. Stark, supervised by Dr. C. J.
Richards, Department of Chemistry, University of Wales, Cardiff, P.O. Box
912, Cardiff, CF1 3TB.
(14) DSC analysis of 5 (Mettler Toledo DSC 12E) revealed violent exothermic
decomposition, onset 258 °C. DSC analysis of 10 revealed slow exothermic
decomposition above 50 °C. We thank Dr. John O’Rourke of Chiroscience
for these results.
Vol. 3, No. 6, 1999 / Organic Process Research & Development
•
445

final product, (R)-aminoglutethimide (1), which was unac-
ceptable from a quality standpoint. Exhaustive preparative
HPLC effort, and high field NMR studies revealed the
unknown impurity as the methyl analogue of racemic acid
(7), 4-cyano-4-(4-nitrophenyl)pentanoic acid (13).17
yield was determined by isolation of the crystalline salt, and
optical purity by chiral HPLC analysis of a sample of acid
regenerated from the diastereomeric salt. Fortunately, a large
solubility differential was observed between the two dia-
stereomeric salts. The (R)-acid cinchonidine salt had a
solubility of 2.8 mg/mL in acetone at 15 °C, compared with
20.3 mg/mL for the diastereomer. Therefore, after investigat-
ing a variety of solvents the best results were achieved with
1 equiv of (-)-cinchonidine, and 30 vol of acetone with 1%
added water to achieve complete dissolution, which gave a
36% yield of diastereomeric salt (72% of available), and 90%
ee of regenerated acid. These results in acetone were
improved by reducing the (-)-cinchonidine charge to 0.6
equiv, which gave comparative yields ( ∼35%) and ee’s
( ∼90%) along with improved volume efficiency at 15 vol.
This resolution procedure was rapidly scaled up to 500 g
racemic acid input, in 10 L laboratory glassware. However,
the resolution did not perform satisfactorily on scale-up. The
crystallised mixtures were very thick and difficult to stir and
could only be removed from the vessel under pressure. The
ee of the regenerated samples of (R)-acid at this stage was
lower than expected at 70-80%. Fortunately this material
was optically enriched, by slurrying in acetone at ambient
temperature, to raise the ee to the required level of >94%.
The resolution was further modified, using 0.5 equiv of (-)-
cinchonidine in 20 vol of acetone. This typically gave
material with an ee > 90% (R) in >30% yield. The required
ee of >94% (R) was achieved with a further optical
enrichment in acetone at reflux followed by an overnight
stir, at 20 °C.
This procedure was used to resolve 52 kg of racemic acid
(7) provided by the pilot plant campaign. Unfortunately a
use-test gave poor optical quality material (40% ee regener-
ated (R)-acid), when compared with laboratory synthesised
racemic acid, and required two further optical enrichments
by slurry to produce acceptable quality material (97% ee
regenerated (R)-acid, 21% yield). However, the time available
necessitated that the racemic acid was resolved, using this
procedure, in three 17 kg batches in the pilot plant. This
gave a total of 40.6 kg of (R)-acid cinchonidine salt (8) in
three batches (48-68% ee regenerated (R)-acid, 35-42%
yield by loss on drying) which were then enriched in a single
slurry, with batch integrity, to give 27.9 kg (84-91% ee
regenerated (R)-acid, 24-27% yield). These three batches
were combined and given a second optical enrichment in
two batches, to give 23.2 kg of (R)-acid cinchonidine salt
(8) (97 and 98% ee regenerated (R)-acid, 83 and 86%
recovery).
It was initially thought that this impurity may have been
derived from the stage 2 alkylation, by contamination of
diethyl sulphate with dimethyl sulphate, or during the stage
1 alkylation by contamination of methyl cyanoacetate with
methyl 2-cyanopropionate. However, NMR and HPLC
analysis of these materials showed no contamination. We
believe that the methyl impurity originated during the stage
1 nucleophilic substitution of 1-chloro-4-nitrobenzene. How-
ever, isotopic labeling studies will be required to confirm
this. Removal of this impurity was achieved by crystallisation
of the final product from aqueous ethanol, as described in
the Stage 6 discussion.
Stage 4. Resolution of 4-Cyano-4-(4-nitrophenyl)-
hexanoic Acid (7). The resolution of 4-cyano-4-(4-nitro-
phenyl)hexanoic acid (7) was investigated with the use of a
variety of chiral amines. (+)-R-Methylbenzylamine was
found to give a crystalline product, but with a large variation
in ee of regenerated acid. This was characteristic of a
resolution which could only be controlled using kinetic
conditions (i.e., accurate temperature and concentration
control, controlled seeding with single diastereomer salt
crystals, mechanical as opposed to magnetic stirring, and
controlled crystallisation times). Therefore, we did not
anticipate that an R-methylbenzylamine-based approach
would provide a reproducible and scaleable resolution. This
was in direct contrast with results reported by the Polish
group.7 They had prepared both (R)- and (S)-aminoglute-
thimide, via R-methylbenzylamine resolution of 4-cyano-4-
(4-nitrophenyl)hexanoic acid (7) in acetonitrile, on a labo-
ratory scale.
The resolution was also investigated using the alkaloid
base (-)-quinine. This gave a crystalline salt with a high
diastereomeric excess but unfortunately with the unwanted
(S)-configuration. Therefore, the cinchona alkaloids quini-
dine, cinchonine and cinchonidine were also screened. It was
found that (-)-cinchonidine (14) gave a crystalline diaster-
eomeric salt, with the correct (R)-configuration. Therefore
it was anticipated that, since it is commercially available in
bulk quantities, a (-)-cinchonidine resolution would provide
a practical approach to the synthesis of (R)-aminoglutethim-
ide (1).
The (-)-cinchonidine resolution of racemic acid (7) was
investigated on a small scale, using jacketed vessels for
reliable temperature control, and external seeding. Chemical
(17) We thank Mr. Iain Reid of Chiroscience for the isolation of this impurity
by preparative HPLC.
446
•
Vol. 3, No. 6, 1999 / Organic Process Research & Development

Stage 5. Preparation of (R)-Nitroglutethimide (9) from
(R)-Acid Cinchonidine Salt (8). Treatment of (R)-4-cyano-
4-(4-nitrophenyl)hexanoic acid cinchonidine salt (8) with 1
M Hydrochloric acid and toluene at ambient temperature
gave a toluene solution of (R)-4-cyano-4-(4-nitrophenyl)-
hexanoic acid. This was azeotropically dried and cyclised
in situ at reflux with catalytic sulphuric acid. The (R)-acid
regeneration was facile, with (-)-cinchonidine hydrochloride
removed in the aqueous acidic layer and aqueous washes,
which were retained for regeneration of the resolving agent.
The toluene solution was then azeotropically dried and
cyclisation to (R)-nitroglutethimide (9) effected in 1-2 h
using 0.2 equiv concentrated sulphuric acid. The azeotropic
drying step was necessary, as in its absence the reaction took
5-6 h to complete, and resulted in partial decomposition of
both substrate and product. The product was crystallised
directly from the toluene solution as a pale yellow granular
crystalline solid in 80-90% yield, with an ee value typically
>98% compared with 95-98% ee of regenerated (R)-acid
from input cinchonidine salt (8). The achiral purity was
>99.5%.
This process was operated successfully in the laboratory
up to 250 g (R)-acid cinchonidine salt and taken into the
pilot plant. Two batches of 11.4 and 11.7 Kg were processed
giving 3.5 and 3.3 kg (65% and 60% yields) of (R)-
nitroglutethimide, both 99.9% ee, and >99.7% achiral purity.
The reaction had scaled up satisfactorily both in terms of
optical and chemical purity. However, the yields of product
isolated from both batches were below that obtained from
laboratory batches, due to losses in the toluene mother liquors
of recrystallisation.
purities, but required a higher catalyst loading, equivalent
to 10% w/w catalyst/substrate.
The previously described methyl analogue impurity was
present at a level of 0.3% by HPLC (PAR) in (R)-
nitroglutethimide produced in the pilot-plant batches, and
remained at this level after hydrogenation, in laboratory use-
tests. Attempts to remove the impurity by crystallisation/
trituration indicated that alcoholic solvents were superior,
however, only a modest purification was achieved. This led
to the reinvestigation of crystallisation from alcohol/water
mixtures. Problems were encountered with both methanol/
water and propan-2-ol/water systems at elevated tempera-
tures, as the product oiled out. Superior results were obtained
with methanol/water, at ambient temperature, which gave
optimum impurity removal and crystalline product. The
current process would therefore require modification, involv-
ing a solvent exchange from ethyl acetate to methanol, or a
suitable alcohol. The azeotrope of ethyl acetate/methanol is
rather poor at 51:49; however, for ethyl acetate/ethanol it is
considerably better at 74:26. Hence, ethanol was the solvent
of choice and this process improvement was introduced.
Laboratory use-tests of (R)-nitroglutethimide (9) produced
in the pilot-plant, using the optimum process, gave crystalline
(R)-aminoglutethimide (1) in 77% yield, overall purity 99.8%
by HPLC (PAR), methyl analogue impurity level 0.11% and
ee > 99.9%. A second methanol/water crystallisation reduced
the impurity level to <0.1%, with 82% recovery, thereby
giving an overall yield of 63% from (R)-nitroglutethimide.
In summary, this hydrogenation procedure was shown to be
versatile, effective, and scaleable in the laboratory, with the
product isolated as a crystalline solid, in moderate yield, with
individual impurity levels within specification.
Stage 6. Preparation of (R)-Aminoglutethimide (1) by
Hydrogenation of (R)-Nitroglutethimide (9). Initial small-
scale catalytic hydrogenation studies using dry palladium on
carbon proved problematic due to the limited solubility of
(R)-nitroglutethimide (9) in several common organic solvents.
Methanol was investigated as hydrogenation solvent; how-
ever, the substrate was only sparingly soluble even in 15
vol. Dissolution was achieved by addition of a cosolvent such
as tetrahydrofuran, or by performing the hydrogenation at
35-45 °C. However, reaction rates were modest at ambient
temperature, and impurity profiles were unacceptable at
elevated temperatures. Typically, product isolated by crys-
tallisation from methanol/water contained 3-4% unidentified
impurities, with the major one at a level of 0.7%.
Greater success was achieved when methanol was re-
placed by ethyl acetate as reaction solvent, and this process
was rapidly scaled up to provide material which was urgently
required for formulation studies. A total of 150 g of product
was isolated in four separate batches in the laboratory. The
reaction was complete within 2 h, and the product was readily
precipitated from heptane. Yields were ∼93%, and the
impurity profile was considerably improved compared to
material prepared using methanol, with no individual impu-
rity at a level >0.1%. However, to allow safe scale-up to
the pilot plant, the use of wet palladium on carbon was
investigated (5% w/w Pd/C containing 50% water). This gave
∼85% yield, comparable reaction times, and acceptable
Conclusions
(R)-Aminoglutethimide (1) was prepared from 1-chloro-
4-nitrobenzene using a six-stage synthetic sequence, which
avoids the problems of hazardous nitration technology, and
therefore regio-isomer contamination of final product. The
process utilised low-cost, commercially available reagents
and exploited the crystallinity of intermediates to overcome
isolation difficulties. This, along with a number of key
process improvements developed within the constraints
imposed by time schedules, allowed safe and rapid scale-up
from laboratory to pilot plant over stages 1-5. A total of 52
kg of the key intermediate racemic 4-cyano-4-(4-nitro-
phenyl)hexanoic acid (7) was produced from 144 Kg of
1-chloro-4-nitrobenzene. This material was prepared on a
1000 L scale over three stages in several batches and
represents a 22% overall yield. A total of 6.8 kg of the stage
five product (R)-nitroglutethimide (9) was prepared to cGMP
on a 450 L scale, in a yield of 14%, from 50.3 kg of racemic
acid (7). A scaleable procedure for the Stage 6 hydrogenation
to produce (R)-aminoglutethimide (1) was demonstrated in
the laboratory. Acceptable yields were achieved on all stages
except the resolution of racemic acid (7). Therefore, future
development work should be focused on optimising this key
stage. Efforts to evaluate alternative resolving agents and
solvent systems should be most productive. In addition the
proposed synthetic approach via an asymmetric Michael
Vol. 3, No. 6, 1999 / Organic Process Research & Development
•
447

addition was attractive. This would avoid resolution based
technology and preliminary results were encouraging. Early
clinical trial results in an oestrogen clearance study in
postmenopausal breast cancer patients revealed both (R)-
aminoglutethimide and the racemate suppressed all oestro-
gens measured to a similar extent and did not suggest any
major clinical advantage from eliminating the (S)-enanti-
omer.18
i-PrOH (2 × 49.7 L, 2 × 0.6 vol), before the product was
dried under vacuum at 45-55 °C. Yield of 4 ) 65.5 kg,
73%, mp 92 °C, (94% assay by HPLC). 1H NMR δ 1.10 (t,
3H), 2.22 (m, 1H), 2.50 (m, 1H), 3.85 (s,3H), 7.78 (m, 2H),
and 8.30 (m, 2H).
4-Cyano-4-(4-nitrophenyl)hexanoic Acid (7). The 1000
L glass-lined steel reaction vessel was charged with 4 (50.9
kg, 193 mol, 1 equiv) and MeOH (407.2 L, 8 vol), and
stirring commenced. To this suspension was added a solution
of K2CO3 (28.0 kg, 203 mol, 1.05 equiv) in water (61.1 L,
1.2 vol). The contents were heated to 30-35 °C, and held
for 30 min, until HPLC analysis indicated complete formation
of 5. NEt3 (29.3 kg, 290 mol, 1.5 equiv) was added to the
solution, which was heated to 50-55 °C. Methyl acrylate
(18.3 kg, 212 mol, 1.1 equiv) was added over 1 h,
maintaining the temperature at 50-55 °C. After stirring for
3 h, further methyl acrylate (5.0 kg, 58 mol, 0.3 equiv) was
added, followed by a further 10 h stir to complete the
reaction. The reactor contents were cooled to 10 °C, a
solution of NaOH (12.4 kg, 309 mol, 1.6 equiv) in water
(127.3 L, 2.5 vol) added, and the mixture heated to 40-45
°C and maintained at this temperature for 3 h to hydrolyze
the intermediate 6. MeOH was removed under vacuum and
replaced with water to maintain the volume. The aqueous
suspension was filtered to remove insoluble impurities and
washed with CH2Cl2 (3 × 102 L, 2 vol) to remove any non-
acidic organic impurities. The aqueous basic layer was
acidified to pH 2.5 with 25% w/w H2SO4 to precipitate the
product. Toluene (142.5 L, 2.8 vol) was added and the two-
phase mixture heated to 70 °C to dissolve the product. The
lower, aqueous layer was separated and the hot toluene layer
washed with water (142.5 L, 2.8 vol) and separated. The
toluene solution was then cooled to 0-5 °C to crystallise
the product, before filtration, washied with cold toluene, and
dried under vacuum at 40 °C. Yield of 7 ) 43.7 kg, 84%,
Experimental Section
1
General. H NMR spectra were recorded at 400 MHz
on a Bruker AM 400 instrument, in CDCl3 solution, using
SiMe4 as internal standard, unless otherwise stated. Melting
points were recorded on a Mettler Toledo DSC 12E and are
uncorrected. The IR spectrum was determined on a Perkin-
Elmer 1600 FTIR spectrometer. The mass spectrum was
determined on a VG Platform II quadrupole spectrometer.
The accurate mass was determined on a VG ZAB magnetic
sector spectrometer.
Methyl 2-(4-Nitrophenyl)cyanoacetate (3). 1-Chloro-
4-nitrobenzene (48 kg, 305 mol, 1 equiv), K2CO3 (73.7 kg,
533 mol, 1.75 equiv) and DMF (206 L, 4.3 vol) were charged
to the 1000 L glass-lined steel reaction vessel. The mixture
was stirred and heated to 120 °C, at which temperature
methyl cyanoacetate (52.8 kg, 533 mol, 1.75 equiv) was
added over 3 h. The temperature was controlled at 118-
125 °C during the methyl cyanoacetate feed, and during the
subsequent 8 h stir. After the mixture was monitored for
complete reaction by HPLC (<2% residual 1-chloro-4-
nitrobenzene), it was cooled to 10-20 °C, and water ( ∼144
L, 3 vol) was added. The pH was adjusted to 2 with 25%
H2SO4. After acidification, further water ( ∼398 L, 8.3 vol)
was added to the mixture, and the slurry was stirred at 10
°C for 1 h to ensure complete precipitation of product. The
mixture was filtered and the filter cake washed with water
(2 × 180 L, 3.75 vol) and dried under vacuum at 55 °C.
Yield of 3 ) 45.5-70.3 kg, 63-93%, mp 95 °C, (89-93%
1
mp 125 °C, (>99.9% assay by HPLC). H NMR δ 0.96 (t,
3H), 2.05 (m, 1H), 2.18 (m, 2H), 2.30 (m, 1H), 2.42 (m,
1H), 2.56 (m, 1H), 7.64 (m, 2H), 8.28 (m, 2H), and 10.32
(br s, 1H).
1
assay by HPLC), 3 batches. H NMR δ 3.85 (s, 3H), 4.90
(s, 1H), 7.70 (m, 2H), and 8.30 (m, 2H).
Methyl 2-Cyano-2-(4-nitrophenyl)butyrate (4). The
1000 L glass-lined steel reaction vessel was charged with 3
(82.85 kg, 340 mol, 1 equiv), SPB-3 (equivalent to 100-
120 °C boiling petroleum ether) (99.4 L, 1.2 vol), and diethyl
sulphate (57.7 kg, 374 mol, 1.1 equiv). The mixture was
stirred and heated to 90-95 °C. NEt3 (35.8 kg, 354 mol,
1.04 equiv) was added over 2 h, maintaining the reaction
temperature at 90-95 °C. After the NEt3 addition the reaction
was stirred for 1 h at 90-95 °C and then cooled to 50-55
°C. Water (14.1 L, 0.17 vol) was added and the mixture
heated to 70-75 °C for 1 h to destroy any residual diethyl
sulphate. The reaction was then cooled to 50-55 °C, and
i-PrOH (198.8 L, 2.4 vol) was added. The reactor contents
were heated to 70-75 °C to ensure all the product was
dissolved and cooled to 0-5 °C. To ensure complete
crystallisation of the product the mixture was held at 0-5
°C for 2 h. The slurry was filtered and washed with cold
HPLC conditions for above procedures: column, Zorbax
Silica, 250 mm × 4.6 mm; mobile phase, heptane 95, i-PrOH
5, TFA 0.1; wavelength, 254 nm; injection volume, 20 µL;
flow rate, 1 mL/min; run time, 20 min. Retention times: 3,
15.9 min; 4, 4.8 min; 5, 7.6 min; 6, 8.7 min; 7, 11.5 min.
2-Ethyl-2-(4-nitrophenyl)pentanedinitrile (11). To a
suspension of (S,S)-2,6-bis(4′-methylenecyclohexyl-2′-ox-
azolinyl)phenyl-aquopalladium(II) hexafluoroantimonate (12)
(0.0074 g, 9.6 µmol) in toluene (3 mL) under a nitrogen
atmosphere was added 5 (0.14 mL, 0.96 mmol) and acry-
lonitrile (0.09 mL, 1.37 mmol). The resulting solution was
cooled to -10 °C. N-Ethyldiisopropylamine (0.017 mL, 98
µmol) was added and the resulting mixture maintained at
-10 °C for 24 h. The reaction mixture was warmed to room
temperature and filtered through a short plug of neutral
alumina (Grade IV) and washed through with CH2Cl2. The
solvent was removed in vacuo and the residue purified by
bulb-to-bulb distillation to give 11 as a brown oil. The
instability of the product precluded further distillation. Yield
(18) Geisler, J.; Lundgren, S.; Berntsen, H.; Greaves, J. L.; Lonning, P. E. J.
Clin. Endocrinol. Metab. 1998, 83, 2687.
448
•
Vol. 3, No. 6, 1999 / Organic Process Research & Development

of 11 ) 0.173 g, 74%, 30.6% ee by HPLC, 93.5% achiral
purity by HPLC. HPLC conditions: achiral chromatography:
column, Phenomenex Luna 2 C18, 150 mm × 4.6 mm;
mobile phase, solvent A, 95:5, 20 mM KH2PO4 pH ) 7.0:
MeCN; solvent B, 25:75, 20 mM KH2PO4 pH ) 7.0:MeCN;
0 min 100% A, 8.00 min 0% A, 11.50 min 0% A, 11.55
min 100% A, 14 min 100% A; wavelength, 210 nm; injection
volume, 20 µL; flow rate, 2 mL/min; run time, 14 min.
Retention time: 11, 7.5 min: chiral chromatography: column,
Chiralcel OD-H, 250 mm × 4.6 mm; mobile phase, heptane
70, i-PrOH 30; wavelength, 210 nm; injection volume, 20
µL; flow rate, 1 mL/min; run time, 20 min. Retention times:
14.5 min, 16.6 min. 1H NMR (Bruker RPX 400, 400 MHz,
CDCl3) δ 0.90 (t, 3H), 2.00 (m, 1H), 2.10 (m, 2H), 2.25
(m,1H), 2.45 (m, 2H), 7.55 (m, 2H), and 8.25 (m, 2H). 13C
NMR (Bruker AMX 360) δ 10.0, 14.2, 34.5, 36.4, 49.1,
117.9, 120.1, 125.1, 127.6, 143.4, 148.4. IR (liquid film) νmax
3055, 2985, 2306, 2253, 1608, 1527, 1422, 1348, 1265, 738
cm.-1 m/z (EI) 243 (M+, 11%), 215 (22), 203 (19), 189 (44),
175 (100), 69 (41). Found: M+, m/z (EI) 243.1009, C13H13N3O2
requires 243.1008.
(R)-4-Cyano-4-(4-nitrophenyl)hexanoic Acid Cinchoni-
dine Salt (8). Acetone (340 L, 20 vol) and (7) (17.0 kg,
64.8 mol, 1 equiv) were charged to the 450 L glass-lined
steel reaction vessel with stirring, and the cloudy solution
was filtered through a 5 µm filter. (-)-Cinchonidine (9.54
kg, 32.4 mol, 0.5 equiv) was added, the mixture heated to
reflux (57 °C), and reflux maintained for 30-35 min,
ensuring complete dissolution. The solution was cooled to
40 °C, seeded with (8) (17 g, 0.1% w/w), and allowed to
crystallise at 40 °C for 4-5 h. The mixture was cooled to
30 °C over 1-2 h and held at this temperature for 15 h. The
mixture was then cooled to 20 °C over 1-2 h, held at this
temperature for 5 h, and isolated by centrifugation to give
the product as an acetone damp cake (15.25 kg, 42% yield
by loss on drying, 48% ee of regenerated acid). Acetone (340
L, 20 vol) and the acetone damp product from above were
charged to the vessel, and the mixture was taken to reflux
(57 °C). The reflux was maintained for 30-35 min. The
mixture was cooled to 20 °C over 2-3 h and stirred out for
15-18 h. The product was isolated by centrifugation, washed
with acetone (170 L, 10 vol), and dried under vacuum at 40
°C, to give 8 (9.8 kg, 27% yield, 84% ee of regenerated
acid). Acetone (235 L, 17 vol) and dry enriched 8 (13.8 kg)
from above were charged to the vessel, the mixture was taken
to reflux (57 °C), and reflux was maintained for 30-35 min.
The mixture was cooled to 20 °C over 2-3 h and then stirred
for 15-18 h. The product was isolated by centrifugation,
washed with acetone (117 L, 8.5 vol), and then dried under
vacuum at 40 °C, to give 8 (11.48 kg, 83% recovery, 97%
ee of regenerated acid). HPLC conditions: achiral chroma-
tography: as for preparation of 7. Chiral chromatography:
column, Chiralpak AD, 250 mm × 4.6 mm; mobile phase,
heptane 90, i-PrOH 10, TFA 0.1; wavelength, 254 nm;
injection volume, 20 µL; flow rate, 1 mL/min; run time, 40
min. Retention times: (R)-(7), 18.0 min; (S)-(7), 26.5 min.
(R)-Nitroglutethimide (9). Toluene (69.2 kg, 7 vol) was
charged to the 450 L glass-lined steel reaction vessel,
followed by 8 (11.43 kg, 20.5 mol). A solution of 36% HCl
(5.2 kg) in water (60 L, 5.2 vol) was added, and the contents
were adjusted to 20 °C and stirred until the solids dissolved.
The lower, aqueous acidic layer was separated and extracted
with toluene (30 L, 2.6 vol). The combined toluene extracts
were washed with water (57.2 L, 5 vol) and separated. The
toluene layer was cooled to 0-5 °C, activated charcoal (286
g, 2.5% w/w) added, and the mixture stirred at this
temperature for 1 h. The suspension was filtered, returned
to the vessel, and dried by azeotropic distillation. The solution
was heated to reflux (111-112 °C), and 96% H2SO4 (419
g, 0.2 equiv) was added. Reflux was maintained until all of
the (R)-(7) was consumed by HPLC. The contents were
cooled to 50-60 °C and, whilst this temperature was
maintained, were washed with water (35.2 L, 3 vol), a
solution of NaHCO3 (1825 g, 1.06 equiv) in water (21.9 L,
1.9 vol), and further water (35.2 L, 3 vol). The toluene
solution was hot-filtered through a 1 µm filter to remove
particulates and concentrated to 9-10 vol, with respect to
expected product, by distillation at reflux. The solution was
cooled to 40 °C and seeded with (9) (5.7 g). The crystallising
mixture was then cooled to 0-5 °C and stirred at this
temperature for 1 h. The product was isolated by centrifuga-
tion, washed with cold toluene (9.3 L), and dried under
vacuum at 55 °C. Yield of 9 ) 3.5 kg, 65% yield, mp 136
°C, 99.9% ee by HPLC, 99.7% achiral purity by HPLC.
HPLC conditions: achiral chromatography: column, Zorbax
Silica, 250 mm × 4.6 mm; mobile phase, heptane 85, i-PrOH
15, TFA 0.1; wavelength, 254 nm; injection volume, 20 µL;
flow rate, 1 mL/min; run time, 20 min. Retention times: 7,
5.0 min; 9, 10.0 min; chiral chromatography: column,
Chiralpak AD, 250 mm × 4.6 mm; mobile phase, heptane
80, i-PrOH 20, TFA 0.1; wavelength, 254 nm; injection
volume, 20 µL; flow rate, 1 mL/min; run time, 40 min.
Retention times: (S)-Nitroglutethimide, 29.0 min; 9, 31.6 min.
1H NMR δ 0.92 (t, 3H), 1.98 (m, 1H), 2.10 (m, 1H), 2.30-
2.48 (m, 3H), 2.70 (m, 1H), 7.52 (m, 2H), 8.24 (m, 2H),
and 8.80 (br s, 1H).
(R)-Aminoglutethimide (1). To a solution of 9 (20.0 g,
76 mmol) in EtOAc (300 mL, 15 vol) in a 2-L Parr pressure
vessel was added 5% Pd/C containing approximately 50%
water (4.7 g). The system was stirred under an atmosphere
of hydrogen until complete removal of the substrate was
observed by HPLC. The catalyst was filtered off and washed
with EtOAc (40 mL, 2 vol). The EtOAc solution was
concentrated to 5 vol by distillation at atmospheric pressure
and the EtOAc replaced by successive additions of EtOH
(124 mL, 6.2 vol), followed by concentration to 5 vol each
time, until EtOAc was removed. The solution was cooled to
25-30 °C, water (140 mL, 7 vol) added until the cloud point
was reached, and the mixture seeded with 1 (0.2 g, 0.1%
w/w). After crystallisation, water (60.0 mL, 3 vol) was added
and the mixture cooled to 0-5 °C for 1 h. The product was
filtered, washed with cold water (100 mL, 5 vol), and pulled
dry on the filter. The product was then dissolved in MeOH
(80 mL, 5 vol). Water (65 mL, 4 vol) was added to this
solution, maintaining the temperature between 20 and 25 °C,
and the solution was seeded with 1 (0.2 g, 0.1% w/w). The
Vol. 3, No. 6, 1999 / Organic Process Research & Development
•
449

crystallising mixture was stirred for 30 min, water (160 mL,
10 vol) added, and the mixture cooled to 0-5 °C. The
mixture was stirred at 0-5 °C for 1 h, and the product was
filtered, washed with cold water (80.0 mL, 5 vol), and dried
under vacuum. Yield of 1 ) 33.6 g, 82% yield, mp 111 °C,
>99.9% ee by HPLC, 99.9% achiral purity by HPLC. HPLC
conditions: achiral chromatography: column, Nova-Pak C18,
150 mm × 3.9 mm; mobile phase, MeOH 25%, pH 5
Phosphate Buffer 75%; wavelength, 254 nm; injection
volume, 20 µL; flow rate, 1 mL/min; temperature, 40 °C;
run time, 45 min. Retention times: 9, 29.0 min; 1, 7.5 min;
chiral chromatography: column, Chiralcel OJ, 50 mm × 4.6
mm; mobile phase, EtOH 25%, hexane 75%; wavelength,
242 nm; injection volume, 20 µL; flow rate, 1 mL/min;
temperature, 40 °C; run time, 12 min. Retention times: 1,
1
3.0 min; (S)-aminoglutethimide, 6.0 min. H NMR δ 0.86
(t, 3H), 1.88 (m, 1H), 1.98 (m, 1H), 2.16 (m, 1H), 2.30 (m,
1H), 2.44 (m, 1H), 2.58 (m, 1H),3.72 (br s, 2H), 6.66 (m,
2H), 7.04 (m, 2H), and 8.18 (br s, 1H).
Received for review January 19, 1999.
OP9900075
450
•
Vol. 3, No. 6, 1999 / Organic Process Research & Development