Process research and development of a dihydropyrimidine dehydrogenase inactivator: Large-scale preparation of eniluracil using a sonogashira coupling

Article abstract of DOI:10.1021/op0100100

Eniluracil (5-ethynyluracil) is a potent inactivator of the enzyme dihydropyrimidine dehydrogenase, which is the rate-limiting enzyme in the metabolism of 5-fluorouracil, a widely used anti-cancer drug. The process research and development of a three-stag

Full text of DOI:10.1021/op0100100

Organic Process Research & Development 2001, 5, 383−386
Process Research and Development of a Dihydropyrimidine Dehydrogenase
Inactivator: Large-Scale Preparation of Eniluracil Using a Sonogashira
Coupling
Jason W. B. Cooke,* Robert Bright, Mark J. Coleman, and Kevin P. Jenkins
Chemical DeVelopment DiVision, GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road,
SteVenage, Hertfordshire. SG1 2NY, UK
Scheme 1. Route to eniluracil
Abstract:
Eniluracil (5-ethynyluracil) is a potent inactivator of the enzyme
dihydropyrimidine dehydrogenase, which is the rate-limiting
enzyme in the metabolism of 5-fluorouracil, a widely used anti-
cancer drug. The process research and development of a three-
stage route to eniluracil is described. A Sonogashira coupling
between 5-iodouracil and trimethylsilylacetylene was used to
synthesise 5-(2-trimethylsilylethynyl)uracil on a >60 kg scale.
Sodium hydroxide deprotection and acidification with acetic
acid completed the synthesis of eniluracil in high yield and
quality. The optimisation of this process is described with
particular attention paid to minimising the input of palladium
and copper catalysts and ensuring that the copper catalyst is
well suspended in the reaction mixture.
The Sonogashira reaction6 is well-known in the literature
and has achieved widespread use as a mild method for
introducing an acetylene into a molecule via coupling with
a vinyl or aryl halide. The use of trimethylsilylacetylene7
(TMSA) was introduced in 1980, and this provides a mild,
safe and high-yielding method for the introduction of an
unsubstituted ethyne fragment. Later papers describe Sono-
gashira-type coupling of acetylenes with 5-iodouracil8 and
close analogues.9 It is the development and scale-up of a
Sonogashira coupling as the foundation of a route to
eniluracil that forms the subject of this paper.
Introduction
5-Fluorouracil (5-FU) is one of the most widely used anti-
cancer drugs for solid tumours. 5-FU is administered by slow
intravenous infusion owing to low and variable oral bio-
availability due to metabolism by the enzyme dihydropyri-
midine dehydrogenase1 (DPD, also known as uracil reduc-
tase), and it is this enzyme that is the target of eniluracil 1
(5-ethynyluracil). Eniluracil is a potent, suicide substrate for
DPD and becomes covalently linked to the enzyme,2 thereby
inactivating it. Co-administration of eniluracil with 5-FU
leads to higher and more prolonged serum levels3 of 5-FU
and allows oral administration of 5-FU. The synthesis of
eniluracil is reported in the literature4 from the 1970s. Patents
were filed5 by the Wellcome Foundation covering the use
of eniluracil as a DPD inactivator to improve therapy with
5-FU.
Issues with the Initial Process
When the project was transferred to Stevenage, a three-
stage pilot-plant process10 had already been devised and
operated (Scheme 1). Stage 1 consisted of a Sonogashira
coupling of TMSA with 5-iodouracil 2 (5-IU) catalyzed by
1.5 mol % PdCl2 and 10 mol % CuI. This was followed by
two charcoal treatments (2 × 0.9 weights of charcoal) and
recrystallisations, which were required to control the heavy
metal contamination of 5-(2-trimethylsilylethynyl)uracil 3
(stage 2). Finally, stage 3, a deprotection with aqueous
sodium hydroxide, was used to remove the trimethylsilyl
group.
The route was considered suitable for the routine long-
term manufacture of eniluracil; however, a number of key
(4) (a) Perman, J.; Sharma, R. A.; Bobek, M. Tetrahedron Lett. 1976, 51, 2427.
(b) Barr, P. J.; Jones, A. S.; Walker, R. T. Nucleic Acids Res. 1976, 3,
2845.
(5) Spector, T.; Porter, D. J. T.; Rahim, S. G. WO 9201452 and WO 9204901.
(6) (a) Sonogashira, K.; Tohda, Y.; Hagihara, N. Tetrahedron Lett. 1975, 50,
4467. (b) For a review see: Campbell, I. B. The Sonogashira Cu-Pd-
Catalyzed Alkyne Coupling Reaction. Organocopper Reagents; Taylor, R.
J. K., Ed.; IRL Press: Oxford, UK, 1994; pp 217-35.
(7) Takahashi, S.; Kuroyama, Y.; Sonogashira, K.; Hagihara, N. Synthesis 1980,
627.
* Author for correspondence. Telephone: +44 1438 764407. Fax: +44 1438
764414. E-mail: jwbc16317@gsk.com.
(1) Heggie, D.; Sommadossi, J.-P.; Cross, D. S.; Huster, W. J.; Diasio, R. Cancer
Res. 1987, 47, 2203.
(2) Porter, D. J. T.; Chestnut, W. G.; Merrill, B. M.; Spector, T. J. Biol. Chem.
1992, 267 (8), 5236.
(3) For a review of eniluracil see: Spector, T.; Porter, D. J. T.; Nelson, D. J.;
Baccanari, D. P.; Davis, S. T.; Almond, M. R.; Khor, S. P.; Amyx, H.;
Cao, S.; Rustum, Y. M. Drugs Future 1994, 19 (6), 565.
(8) Imamura, K.; Yamamoto, Y.; Bioorg and Med. Chem. Lett. 1996, 6 (15),
1855-1858.
(9) (a) Robins, M. J.; Barr, P. J. Tetrahedron Lett. 1981, 22, 421-424. (b) De
Clercq, E.; Descamps, J.; Balzarini, J.; Giziewicz, J.; Barr, P. J.; Robins,
M. J. J. Med. Chem. 1983, 26, 661-666.
(10) See the Supporting Information for details of this process.
10.1021/op0100100 CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 05/23/2001
Vol. 5, No. 4, 2001 / Organic Process Research & Development
•
383

issues with the process were identified: high levels of copper
and palladium in the drug substance, the colour of the drug
substance, less than optimal yields, and complex work-up
and isolation procedures. The factors which were selected
as most relevant in addressing these issues were: the
quantities of palladium and copper catalysts used, the reaction
temperature, the formation of the palladium catalyst11
((Ph3P)2PdCl2) in situ, and the work-up procedure. It was
apparent that at the original reaction temperature of refluxing
ethyl acetate (ca. 75 °C) the reaction was complete very
quickly (<10min) and that a significant amount of catalyst
decomposition occurred as indicated by precipitation of a
black solid. Furthermore, the boiling point of the acetylene
component is only 53 °C, and considerable losses could be
expected at 75 °C.
drying to give the intermediate grade (IG) 3. This material
was off-white and typically had metal levels of: Pd ca. 600
ppm and Cu ca. 100 ppm. The earlier process provided dark
grey-coloured material with metal levels of: Pd >5000 ppm
and Cu >13000 ppm.
The much reduced heavy metal contamination of the IG
3 allowed a single, much reduced charge of charcoal to be
used in stage 2; typically 0.4-0.6 wt. This led to improved
yields and alleviated the issues of filtration of charcoal on
scale-up. Stage 2 consisted of treating a solution of 3 with
charcoal (0.4 wt) in a solvent mixture of boiling THF/MeOH
1:1 (this gave better solubility than either solvent alone) and
filtration of the hot mixture. The charcoal was collected in
a pressure filter jacketed with hot water to avoid crystalli-
sation in the filter. A vacuum distillation of solvent from
the filtrate was found to be beneficial to minimise the
formation of impurities. Addition of water as an anti-solvent
led to isolation of purified 3 as a white crystalline solid with
very low levels of palladium (typically <2 ppm) and copper
(typically <1 ppm).
Stage 3 was relatively unaltered except for the omission
of all charcoal and filter aid. The purified 3 was dissolved
in 1 M sodium hydroxide and when the reaction was
complete, the hexamethyldisiloxane (HMDSO) by-product
was separated and discarded before a line-filtration of the
disodium salt solution into the crystallisation vessel. The
product 1 was formed by acidifying to pH 5 with acetic acid,
collecting the fine solid by filtration and washing with water
followed by acetone. This allowed isolation of very high
quality material, in a virtually quantitative yield. The HPLC
purity was typically >99.9%, and the largest impurity in all
batches was acetone (typically 0.3% w/w determined by
NMR) which could not be reduced even by extended drying
at 50° under vacuum. Levels of palladium were typically 2
ppm, and copper, <1 ppm.
Process Improvements
It was soon established that the process used for the in
situ formation of the catalyst by mixing PdCl2 and Ph3P was
very inefficient (as judged by the almost complete insolubility
of PdCl2 in EtOAc) compared to using preformed (Ph3P)2-
PdCl2, and hence the preformed catalyst was used in all
further experiments. The Pd catalyst loading was reduced
from 1.5 mol % to 0.5 mol % which maintained an
acceptable reaction rate, but further reduction to 0.1 mol %
gave a very slow reaction. For the sake of robustness and
reaction rate a level of 0.5 mol % was adopted. Similarly,
the loading of CuI was reduced from 10 mol % to 0.5 mol
% but further reduction or omission gave an unacceptable
conversion. These modifications had a number of clear
benefits: reduced catalyst purchase cost, reduced heavy metal
contamination of drug substance, intermediate, and waste
streams. These experiments were carried out at a reaction
temperature of 25 °C which gave ca. 95% conversion in 6
h. More importantly, the colour of the reaction mixture was
maintained as pale yellow throughout, with no precipitation
of dark solids, which would indicate catalyst decomposition
and lead to high levels of heavy metal contamination in the
isolated intermediate 3. It was found that effective de-
oxygenation of the reaction mixture was necessary to
minimise heavy metal contamination; hence, the reaction
mixture was degassed by application of three vacuum-
nitrogen purge cycles prior to introduction of the TMSA.
A screen of various solvents for the Sonogashira reaction
(ethyl acetate, DMF, THF, MeOH, EtOH, and n-PrOH) was
carried out. This confirmed ethyl acetate as the solvent of
choice despite the very low solubility of both 2 and 3. The
low solubility of 3 allowed direct isolation from the reaction
mixture (10 vol of ethyl acetate were required to achieve a
mobile slurry) by filtration with minimal losses in the filtrate.
The previous process had required a slurry with aqueous
EDTA to reduce the levels of copper, but with the massively
reduced input of CuI this became unnecessary. Instead, a
water slurry was used to remove the triethylammonium
iodide formed in the reaction. This was best carried out in a
mechanical filter-dryer which allowed effective mixing of
water with the initial filter cake followed by in situ vacuum-
Issues during Development
It was found in some laboratory-scale reactions that the
copper iodide was not properly suspended but formed a solid
mass at the bottom of the flask and resulted in a failed
reaction. Copper iodide is a very dense solid (d 5.62 g/mL),
and even finely divided material requires vigorous agitation
to remain suspended. We were concerned about this issue
in large conical-bottomed vessels in the pilot plant, and thus
calculations were carried out to determine a minimum stirring
rate for the particular plant we were planning to use. These
calculations indicated that a stirring speed of 120 rpm would
effectively suspend copper iodide with a particle size of less
than 150 µm. By adding the copper iodide powder to a
reaction mixture stirred at 120 rpm we avoided any failed
reactions on plant.
During a routine user test of a new batch of 5-IU it was
noticed that the reaction mixture turned a dark brown colour
(normally yellow) and achieved only 11.7% conversion with
the normal charge of catalyst. By increasing the catalyst
loading to 1.5 mol % a 93.1% conversion could be achieved,
but this would not be acceptable for processing because of
the much higher heavy metal contamination in the IG 3. We
discussed the possible contaminants with the supplier of the
(11) More precisely (Ph3P)2PdCl2 is a precursor to the true catalyst: (Ph3P)2Pd.
384
•
Vol. 5, No. 4, 2001 / Organic Process Research & Development

batch of 5-IU and began to suspect that elemental sulfur may
be present from sodium thiosulfate used to quench excess
iodine in the iodination of uracil. A re-work procedure was
devised, consisting of dissolving the rather insoluble 5-IU
in 2 M sodium hydroxide, clarifying the hazy solution
through a 1 µm line filter and precipitating the 5-IU by
acidifying with concentrated hydrochloric acid. This proce-
dure gave a 94.2% recovery on a 200 g scale, and the
resultant material gave a satisfactory user test. The re-work
was rapidly scaled-up to process >200 kg of 5-IU and to
allow us to continue with our campaign with minimal loss
of time. It is interesting to note that in all respects, other
than the user test, this batch of 5-IU passed our specification,
and it was shown that the sulfur impurity was present at only
0.2% w/w. The specification was subsequently revised to
incorporate a clarity test of a caustic solution of 5-IU.
It seemed prudent to try to integrate stages 1 and 2 by
processing the initial filter cake (including the triethylam-
monium iodide) directly in stage 2. This would have the
benefits of fewer washes and less waste, as well as one less
drying step, and would therefore save time and energy. To
this end the initial filter cake from stage 1 was washed with
ethyl acetate, dissolved in boiling THF/MeOH, and treated
with charcoal (0.4 wt). 3 was isolated in the normal way,
and it was shown that the triethylammonium iodide had been
successfully removed but the material was bright yellow
rather than white. The yellow-coloured impurity was isolated,
and inspection of its NMR spectrum showed that it was not
drug-related but indicated that it was derived from butylated
hydroxytoluene 4 (BHT) the standard stabiliser found in
THF. An authentic sample of a known12 BHT dimer 5 was
prepared by treatment of BHT with manganese dioxide
followed by chromatography to give a pure sample of 5 as
a deep orange solid, which was identical to that isolated from
the yellow 3. This indicated that the stabiliser in the THF
was being oxidised by the palladium and copper contami-
nants in the IG 3 but only in the presence of triethylammo-
nium iodide. Although it was possible to remove the BHT
dimer 5 at stage 3, we elected not to implement this
integrated stage for the pilot-plant campaigns.
the pilot plant at Stevenage and subsequently validated at a
manufacturing site. At the manufacturing site stages 1 and
2 were combined into a single process with the IG 3 being
dissolved in the filter-dryer. This led to further improvements
in the environmental impact of the process and the oc-
cupational health benefit of reduced solids handling.
Experimental Section
Reactions were carried out in 1500 L glass-lined carbon
steel reactors and the intermediate isolated in a Hastelloy
(C22) Schenk mechanical filter-dryer with a poroplate filter
of 1.2 m2. The drug substance was isolated in a stainless
steel enclosed pan filter (internal diameter 890 mm) and dried
in a vacuum oven. All temperatures are in degrees Celsius.
Levels of palladium and copper were determined using
inductively coupled plasma-optical emission spectroscopy
(ICP-OES) and are quoted in parts per million. HPLC was
carried out using the following method: eluent A: 970:30:1
water:methanol:acetic acid, eluent B: 100:900:1 water:
methanol:acetic acid. An isocratic elution of 100% A for 10
min and then a gradient to 100% B over 20 min; column
150 mm × 4.6 mm YMC-AQS-3 120A, 3 µm; flow rate 1.0
mL/min; temperature 40 °C; UV detection at 281 nm.
Stage 1: Intermediate grade 5-(2-trimethylsilylethy-
nyl)uracil 3. (Ph3P)2PdCl2 (0.95 kg, 1.35 mol, 0.005 equiv)
and finely ground CuI (0.26 kg, 1.37 mol, 0.005 equiv) were
added to a well-stirred suspension of 2 (64.20 kg, 269.8 mol)
in ethyl acetate (514 L) at 15-25°. The mixture was then
deoxygenated by evacuating and flushing with nitrogen three
times. TMSA (31.3 kg, 318.7 mol, 1.15 equiv) followed by
triethylamine (31.7 kg, 313.3 mol, 1.15 equiv) was added
and the addition flask rinsed with ethyl acetate (128 L) into
the reactor. The suspension was stirred under nitrogen at 25°
for 19 h. The solids were isolated by filtration in a
mechanical filter-dryer under nitrogen and washed sequen-
tially with ethyl acetate (2 × 128 L), water (3 × 128 L) and
finally ethyl acetate (2 × 128 L). The product was dried
under vacuum in the filter-dryer at 50° overnight to give IG
3 as an off-white powder 52.42 kg, 251.7 mol, 93.3% theory
yield, 81.7% w/w yield. HPLC 98.5% purity. 1H NMR (400
MHz; d6-DMSO) δ: 11.35 (br s, 2H), 7.79 (s, 1H), 0.18 (s,
9H). ICP-OES: Pd 550 ppm, Cu 120 ppm.
Stage 2: 5-(2-Trimethylsilylethynyl)uracil 3. A mixture
of IG 3 (50.44 kg, 242.2 mol) and decolourising charcoal
(Norit SX+, 21.50 kg) in a mixture of THF (504 L) and
MeOH (504 L) was heated at reflux (61-62°) for 1 h. The
suspension was filtered hot (ca. 60°) by pumping through a
filter train consisting of a pressure filter, a GAF filter (5 µm
bag), and a 1 µm line filter in series. The charcoal filter cake
was washed with hot (ca. 60°) THF/MeOH (1:1, 252 L).
The wash was added to the distillation vessel and the mixture
concentrated to 252 L under reduced pressure (batch tem-
perature ca. 20-30°) to give a slurry. The temperature of
the slurry was adjusted to 25-30°. The slurry was diluted
with water (252 L) over 30 min at 25-30°, cooled to 20°,
and aged at 15-20° for 30 min. The product was isolated
by filtration in a mechanical filter-dryer, washed with a
mixture of water and MeOH (2:1, 2 × 100 L), and dried in
the filter-dryer at 50° overnight to give 3 as a white
Conclusions
A Sonogashira coupling of 5-IU with TMSA has been
optimized and scaled up to >60 kg batch size in 1500 L
plant. Particular attention was paid to minimising input of
palladium and copper catalysts, which allowed the prepara-
tion of good quality material with a minimum amount of
charcoal to remove heavy metal contaminants. The process
has been successfully operated on numerous occasions in
(12) Liu, L. K.; Lee, Y. H.; Leu, L. S. J. Chin. Chem. Soc. (Taipei) 1989, 36
(3), 219.
Vol. 5, No. 4, 2001 / Organic Process Research & Development
•
385

powder: 47.92 kg, 230.1 mol, 95.0% yield. HPLC 99.9%
purity. 1H NMR (400 MHz; d6-DMSO) δ: 11.35 (br s, 2H),
7.79 (s, 1H), 0.18 (s, 9H). ICP-OES: Pd 1.6 ppm, Cu <0.7
ppm.
overnight to give 1 as an off-white powder: 30.06 kg, 220.9
mol, 99.0% theory yield, 64.7% w/w. HPLC 99.95% purity.
1H NMR (400 MHz; d6-DMSO) δ: 11.36 (br s, 2H), 7.82
(s, 1H), 4.03 (s, 1H). ICP-OES: Pd 2 ppm, Cu <1 ppm.
Stage 3: 5-Ethynyluracil 1 (Eniluracil). A 1 M aqueous
solution of NaOH (1162 L) was prepared by adding 50%
w/w aqueous NaOH (93.0 kg, 1162 mol, 5.2 equiv) to
purified water (744 L) and diluting with further purified water
(372 L). 3 (46.48 kg, 223.2 mol) was added at 15-20°, and
the mixture stirred vigorously at 20° for 2 h. The agitator
was stopped and the HMDSO allowed to separate. The lower
aqueous layer was clarified by filtration through a GAF filter
(5 µm bag) and a 1 µm line filter, in series. The HMDSO
was extracted with purified water (93 L), the phases were
allowed to separate, and the lower aqueous phase was
transferred into main batch via the filter train. The upper
HMDSO layer was discarded. The pH of the combined
filtrate and wash was adjusted to 5.0 by adding filtered acetic
acid (98.6 kg, 1642 mol, 7.36 equiv) at 20-25°. The resultant
suspension was cooled to 20° and aged at 15-20° for at
least 30 min. The product was isolated by filtration, washed
with filtered, purified water (2 × 116 L), washed with filtered
acetone (2 × 116 L), and dried in a vacuum oven at 50°
Acknowledgment
We are grateful to Sanford Strunk, Marc Caddell, John
Eaddy, and Roy Flanagan for the early process research and
development of this route and initial scale-up to the pilot
plant. We are grateful to Antony Janes for carrying out the
calculations concerning effective suspension of CuI. We are
grateful to Keith Freebairn, Guy Wells, and Jacky Hicks for
analytical support.
Supporting Information Available
Additional detail of the original process at the point of
transfer to Stevenage.. This material is available free of
charge via the Internet at http://pubs.acs.org.
Received for review January 30, 2001.
OP0100100
386
•
Vol. 5, No. 4, 2001 / Organic Process Research & Development