The discovery and process development of a commercial route to the water soluble prodrug, fosfluconazole

Article abstract of DOI:10.1021/op010064+

A case history detailing the rationale behind the discovery of 2-(2,4-difluorophenyl)-1,3-bis(1H-1,2,4-triazole-1-yl)-2-propyl dihydrogen phosphate, fosfluconazole (2), a water-soluble prodrug of Diflucan, and the subsequent development of a commercial route is presented. Particular items to note are (i) that this compound was discovered in the Chemical Research and Development Department, hence Chemical Research and Development can play a key role in prodrug discovery, (ii) the strategy behind the selection of phosphate ester promoiety, by phosphorylation of a sterically hindered tertiary alcohol, (iii) the development of the initial route to remove thermally hazardous reagents and to improve processing to allow scale-up, and (iv) the identification and development of the proposed commercial process.

Full text of DOI:10.1021/op010064+

Organic Process Research & Development 2002, 6, 109−112
Articles
The Discovery and Process Development of a Commercial Route to the Water
Soluble Prodrug, Fosfluconazole
Arthur Bentley, Michael Butters, Stuart P. Green,* William J. Learmonth, Julie A. MacRae, Matthew C. Morland,
Garry O’Connor, and Joanne Skuse
Department of Chemical Research and DeVelopment, Pfizer Global Research and DeVelopment Laboratories,
Sandwich, Kent CT13 9NJ, United Kingdom
Abstract:
oral and intravenous formulations, the latter being a dilute
(2 mg/mL) infusion in saline.
A case history detailing the rationale behind the discovery of
2
-(2,4-difluorophenyl)-1,3-bis(1H-1,2,4-triazole-1-yl)-2-propyl di-
In the mid 1990s, the identification and development of
a highly water soluble prodrug of fluconazole became
attractive for Pfizer Global Research and Development since
this might offer both clinical and commercial advantages over
the Diflucan infusion formulation. A low-volume product
would allow for bolus administration, reducing fluid and
sodium load, and it could facilitate access to higher doses.
A smaller, concentrated product would therefore offer
advantages to the patient, as well as being more convenient
for storage and in clinical use.
hydrogen phosphate, fosfluconazole (2), a water-soluble prodrug
of Diflucan, and the subsequent development of a commercial
route is presented. Particular items to note are (i) that this
compound was discovered in the Chemical Research and
Development Department, hence Chemical Research and De-
velopment can play a key role in prodrug discovery, (ii) the
strategy behind the selection of phosphate ester promoiety, by
phosphorylation of a sterically hindered tertiary alcohol, (iii)
the development of the initial route to remove thermally
hazardous reagents and to improve processing to allow scale-
up, and (iv) the identification and development of the proposed
commercial process.
Results and Discussion
Fluconazole has two functional groups from which to
attach a promoiety, the tertiary alcohol group and the two
triazole rings. At the start of this project, workers at Abbott2
had recently shown that water-soluble prodrugs could be
readily accessed by alkylating a nitrogen-containing hetero-
cycle with bromomethyl acetate to give a quarternary
ammonium salt prodrug.
Introduction
1
Fluconazole (Diflucan) (1) Pfizer’s broad-spectrum an-
tifungal agent was launched in 1988, and its sales have grown
strongly, reaching $1 billion per annum in 2000. Its develop-
An approach of this type was considered for producing a
water-soluble prodrug of fluconazole, but it was deemed to
have a number of disadvantages:
(1) The use of an alkylating agent in the final step of a
drug synthesis was considered undesirable due to the need
for very tight control of the residual alkylating agent in the
drug substance.
(2) With two equivalent triazole rings, the alkylation of
fluconazole is likely to give mixed products. The plane of
symmetry in fluconazole is broken when producing a mono-
alkylated prodrug, leading to the generation of a chiral centre.
The two enantiomers may have different rates of cleavage
in vivo, and the development of a racemate is now generally
disfavoured in the pharmaceutical industry.
ment and subsequent approval during the 1980s came at a
time when the number of patients suffering from serious
fungal infections had risen sharply amongst cancer, trans-
plantation, and AIDS patients. Diflucan is marketed in both
(3) The achiral dialkylated prodrug was more straight-
forward to access, but as expected for a disalt it proved to
be hygroscopic and difficult to handle as a drug substance
*
Corresponding author: E-mail: stuart•green@sandwich.pfizer.com. Tele-
phone: +44 1304 641191.
1) (a) Richardson, K. Patent Appl. GB 2099818, 1982; Chem. Abstr. 1983,
9, 38467q. (b) Richardson, K.; J. Chemother. (Florence) 1990, 2, 51. (c)
(Scheme 1).
(
9
Richardson, K.; Cooper, K.; Marriot, M. S.; Tarbit, M. H.; Troke, P. F.;
Wittle, P. J. ReV. Infect. Dis. 1990, 12, 5267. (d) Richardson, K. Contemp.
Org. Synth. 1996, 3, 125.
(2) Davidsen, S. K.; Summers, J. B.; Albert, D. H.; Holmes, J. H.; Heyman, H.
R.; Magoc, T. J.; Conway, R. G.; Rhein, D. A.; Carter, G. W. J. Med. Chem.
1994, 37, 4423.
1
0.1021/op010064+ CCC: $22.00 © 2002 American Chemical Society
Vol. 6, No. 2, 2002 / Organic Process Research & Development
•
109
Published on Web 12/18/2001

Scheme 1
Scheme 2. Initial synthesis
This initial route to fosfluconazole, although relatively
high-yielding, had a number of issues that made it inap-
propriate for further scale-up.
(
4) The drugs in this class of prodrugs liberate an
(1) The use of the weakly acidic 1H-tetrazole used to
promote the initial phosphorylation reaction had serious
safety issues, since this reagent is both shock- and heat-
sensitive, detonating when heated to its melting point (155
2
equivalent of formaldehyde in vivo when cleaved.
Consequently, we focused our efforts on identifying a
suitable polar or polarisable group with which to functionalise
the tertiary alcohol group. The potential benefits of such an
approach were that all of the issues delineated above could
be avoided.
6
°C).
(2) In addition to this, the use of the m-chloroperoxy-
benzoic acid oxidant also had associated thermal hazards,7
necessitating the use of an alternative oxidant.
3
Phosphate esters have frequently been employed as
prodrugs since they are readily ionisable and, in a salt form,
they significantly increase aqueous solubility. They are also
sufficiently stable in the solid state and in aqueous solution
to allow their development for intravenous administration.
Furthermore, they are readily cleaved in vivo by nonselective
alkaline phosphatases, thus releasing the parent drug. How-
ever, given the steric hindrance of the tertiary alcohol group
of fluconazole, we believed that making such a phosphate
ester would be extremely difficult, and indeed if such an
ester could be produced, we expected it to be unstable.
Nevertheless, we were delighted to be able to access
(3) Finally, the hydrogenation conditions proved to be
nonrobust, and the isolation of fosfluconazole was compli-
cated by its poor solubility in methanol.
We initially tackled the thermal hazards associated with
step 1, and after screening other weak acids, we found that
5-methyl-1H-tetrazole, pyridinium hydrobromide with cata-
lytic 4-(dimethylamino)pyridine, and imidazole hydrochloride
would each promote the coupling reaction. 1,2,4-1H-triazole,
which had been successfully used in the phosphorylation at
8
the anomeric position of a sugar, also gave excellent results.
We chose 1,2,4-1H-triazole for scale-up, and after screening
various oxidants we selected 30% aqueous hydrogen per-
oxide as it yielded a superior product quality. A more
comprehensive study of alternative additives has recently
been reported by Sanghvi and co-workers in this journal,9
although we have not evaluated their findings with flucona-
zole. We next addressed the processing of the final step.
Here, the insolubility of the product in methanol meant that
the catalyst became poisoned, giving variable product quality
together with losses of the product during the catalyst
removal. In addition to this, the product formed a tightly
bound methanol solvate. Fortunately, these processing issues,
together with the concerns of forming dialkyl phosphate
esters, could be avoided by changing the solvent medium to
aqueous sodium hydroxide in which the product, by design,
was soluble and stable. After the removal of the catalyst by
filtration and after separation of the toluene by-product, the
diacid could be easily crystallised by acidification with
sulfuric acid. Consultation with Johnson Matthey led to the
4
fosfluconazole (2) from fluconazole using the phosphory-
lation procedure developed by Fraser-Reid5 employing
dialkyl phosphoramidites. Hence, when fluconazole was
treated with dibenzyl diisopropylphosphoramidite in the
presence of 1H-tetrazole in dichloromethane, a smooth
conversion to the dibenzyl phosphite (3) was achieved. This
was oxidised in situ with m-chloroperoxybenzoic acid to the
corresponding phosphate (4) in an 87% overall yield. The
two benzyl-protecting groups were removed by hydrogenoly-
sis over a palladium on carbon catalyst in methanol to give
the free diacid (2) in a 65% yield (Scheme 2). Preliminary
studies showed that fosfluconazole did indeed have ap-
preciable aqueous solubility (>300 mg/mL), as the disodium
salt and that it was readily cleaved in vivo to fluconazole.
Additionally, accelerated stability studies showed that it had
sufficient stability to allow the preparation of a drug product
with a viable commercial shelf life. Consequently, it was
nominated for further development.
(
3) (a) Collis, A. J. Drug Access and Prodrugs. In Medicinal Chemistry: The
Role of Organic Chemistry in Drug Research, 2nd ed.; Ganellin, C. R.,
Roberts, S. M., Eds.; Academic Press: New York, 1993. (b) Bundgaard,
H. Design of Prodrugs; Elsevier Science: Amsterdam, 1985.
(6) Urben, P. G. Bretherick’s Handbook of ReactiVe Chemical Hazards, 5th
ed.; Butterworth Heinemann: London, 1995; Vol. 1, p 166.
(7) Urben, P. G. Bretherick’s Handbook of ReactiVe Chemical Hazards, 5th
ed.; Butterworth Heinemann: London, 1995; Vol. 1, p 864.
(8) Kondo, H.; Sim, M. M.; Wong. C. H. J. Am. Chem. Soc. 1993, 115, 2260.
(9) Sanghvi, Y. S.; Guo, Z.; Pfundheller, H. M.; Converso, A. Org. Process
Res. DeV. 2000, 4, 175.
(
4) Green, S. P.; Murtiashaw, C. W.; Stephenson, P. T.; Patent Appl. WO
9
728169, 1997; Chem. Abstr. 1997, 127, 220800s.
(
5) Fraser-Reid, B.; Yu, K.-L. Tetrahedron Lett. 1988, 29, 979.
110
•
Vol. 6, No. 2, 2002 / Organic Process Research & Development

selection of the optimum palladium-on-carbon catalyst, which
minimised both under-reaction, leading to the monobenzyl
ester (5), and over-reduction, leading to the defluorinated
product (6). This route was successfully scaled up in large-
scale laboratory equipment to prepare two 2-kg batches to
fund toxicology and early clinical studies.
hydrogen peroxide to a highly flammable solvent. We
subsequently changed the solvent to dichloromethane prior
to scale-up. We were keen to avoid the use of pyridine and
we have screened many different bases (e.g., trialkylamines,
N-methylimidazole, and DBU) to achieve this conversion
without success. We have now scaled up this process to
produce approximately 1 tonne of the dibenzyl phosphate
ester (4) in our manufacturing facilities in Sandwich.
We have experienced challenges scaling up the hydro-
genation step in our pilot plant and manufacturing facilities
during the fosfluconazole development program. For ex-
ample, we found that as we scaled up the process, drying
times increased, and under these conditions, hydrolysis of
the product back to fluconazole was observed. This was
minimised by giving the filter cake a thorough wash with
acetone to displace the water required to hydrolyse the
phosphate ester, which in turn, reduced the drying times
required. Other challenges have pertained to the control of
various impurities that may be present in the drug substance
at parts per million levels. The use of a palladium-on-carbon
catalyst in the final step of a drug substance synthesis has
led to the introduction of an additional operation to remove
trace levels of this metal. This has been successfully achieved
by incorporating a carbon treatment of an aqueous solution
of the disodium salt of the prodrug after the catalyst filtration,
and prior to the crystallisation of the diacid.
Although the initial route had been quickly developed into
one that was more appropriate for scale-up, we did not
consider this route to be commercially viable. This was
because there was only a limited bulk availability of the
dibenzyl diisopropylphosphoramidite reagent. This in turn
was due to difficulties in preparing and purifying this liquid
reagent to a sufficiently high quality on a bulk scale, and as
a result, both the high costs and poor availability led us to
seek an alternative set of phosphorylation conditions. We
were initially attracted to forming the phosphorus oxygen
bond with the phosphorus already at the correct oxidation
state, but we were unable to identify any conditions that
would allow us to achieve this. Instead we were forced to
establish conditions that coupled the phosphorus at the more
reactive phosphite oxidation state, with subsequent oxidation
in situ. Fortunately, we managed to successfully establish a
procedure employing cheap and readily available materials
that was suitable for scale-up into both the pilot plant and
our production facilities. In this procedure, fluconazole was
combined with pyridine in a suitable solvent (initially ethyl
acetate, later dichloromethane) with sequential additions of
phosphorus trichloride, benzyl alcohol, and hydrogen per-
oxide to provide the desired protected phosphate ester (4)
in 65-70% (Scheme 3). We found that the process produced
marginally purer products when the reaction was run in ethyl
acetate, and the processing was more straightforward in this
solvent during work-up. Nevertheless, we had concerns about
the potential formation of ethyl chloride during the initial
phases of the reaction and the potential engineering measures
that would be required to manage the safety issue of adding
Conclusions
A highly water-soluble prodrug of Diflucan (1) has been
rationally designed by functionalising the tertiary alcohol as
the phosphate ester to give fosfluconazole (2). The develop-
ment of the initial phosphoramidite route to remove the
thermal hazards and to improve the processing in the final
step that are described in this contribution, allowed the
support of exploratory development. The subsequent iden-
tification and development of a viable manufacturing route
has allowed several hundred kilograms of fosfluconazole to
be prepared in both pilot-plant and production facilities.
Experimental Section
Proton NMR data were recorded on a Varian Unity 300
spectrometer operating at 300 MHz. Microanalytical data
were obtained from Exeter Analytical UK Ltd. Mass spectra
were obtained on a Micromass Autospec-Q. Melting points
were determined on a Buchi melting point apparatus.
Dibenzyl 2-(2,4-difluorophenyl)-1,3-bis(1H-1,2,4-tria-
zole-1-yl)-2-propyl phosphate (4). Method 1. A solution
of 2-(2,4-difluorophenyl)-1,3-bis(1H-1,2,4-triazole-1-yl)-2-
propan-2-ol (10 g, 32.6 mmol), 1H-tetrazole (6.85 g, 97.8
mmol), dibenzyl diisopropyl phosphoramidite (22.55 g, 65.2
mmol) in dichloromethane (100 mL) was stirred at ambient
temperature under nitrogen for 2 h. The mixture was then
cooled to 0 °C, and a solution of m-chloroperoxybenzoic
acid (13.5 g, 50-55% w/w, 39 mmol) in dichloromethane
Scheme 3. Proposed commercial synthesis
(50 mL) was added, maintaining the temperature at 0 °C.
The resulting mixture was allowed to warm to ambient
temperature for 1 h before washing with aqueous sodium
metabisulfite and sodium bicarbonate. After drying (MgSO )
4
Vol. 6, No. 2, 2002 / Organic Process Research & Development
•
111

the solvent was removed and replaced with methyl isobutyl
ketone (37 mL) and tert-butyl methyl ether (74 mL). After
granulating at -10 °C for 1 h the product was filtered and
washed with ice cold methyl isobutyl ketone and tert-butyl
methyl ether (1:3, 15 mL) and dried at 50 °C under vacuum
and washed with tert-butyl methyl ether (2 × 70 L) and dried
at 40 °C under vacuum for 8 h to afford the title compound
(97.7 kg, 66%) as a white solid.
2-(2,4-Difluorophenyl)-1,3-bis(1H-1,2,4-triazole-1-yl)-
2-propyl dihydrogen phosphate (2). A slurry of dibenzyl
2-(2,4-difluorophenyl)-1,3-bis(1H-1,2,4-triazole-1-yl)-2-pro-
pyl phosphate (30.1 kg, 53.13 mol), 5% palladium-on-carbon
catalyst (50% wet, type 5R39, 1.5 kg), and sodium hydroxide
(4.36 kg, 108.9 mol) in low-endotoxin water (75.7 L) was
hydrogenated at ambient temperature and 414 kPa (60 psi)
for 12 h. The slurry was filtered, and the catalyst was washed
with low-endotoxin water (9.8 L). After separating the
toluene by-product, the aqueous phase was slurried with
carbon (3.1 kg) for 30 min. After the carbon was removed
by filtration, the aqueous phase was acidified to pH 1.45 by
that addition of sulfuric acid (6.69 kg) in low-endotoxin water
(25 L) over 2 h. The resulting slurry was granulated at
ambient temperature for 1 h and then filtered. The product
was sequentially washed with filtered low-endotoxin water
(103 L) and filtered acetone (103 L). The product was dried
under vacuum at 50 °C for 12 h to give the title compound
for 18 h to afford the title compound (16.05 g, 87%) as a
1
white solid: mp 92-93 °C. H NMR (CDCl
3
) δ 4.90 (2H,
d), 4.95 (2H, d), 5.05 (2H, d), 5.19 (2H, d), 6.58-6.73 (2H,
m), 6.88-6.95 (1H, m), 7.20-7.30 (4H, m), 7.32-7.38 (6H,
+
m), 7.80 (2H, s), 8.36 (2H, s). m/z 567 [M + H] . Found:
C, 57.12; H, 4.46; N, 14.85. C27
7.24; H, 4.46; N, 14.84.
Method 2. A solution of 2-(2,4-difluorophenyl)-1,3-bis-
1H-1,2,4-triazole-1-yl)-2-propan-2-ol (1.5 kg, 4.9 mol), 1H-
,2,4-triazole (507 g, 7.35 mol), dibenzyl diisopropyl phos-
25 2 6 4
H F N O P requires: C,
5
(
1
phoramidite (2.53 kg, 7.35 mol) in dichloromethane (7.5 L)
was refluxed under nitrogen for 20 h. The mixture was then
cooled to 10 °C, and 30% hydrogen peroxide (1 L, 8.8 mol)
was added, maintaining the temperature between 10 and 25
°
C. The resulting mixture was allowed to warm to ambient
temperature for 1 h before separating the phases. The organic
phase was washed with aqueous sodium metabisulfite,
hydrochloric acid, and water. The solvent was removed and
replaced with methyl isobutyl ketone (5 L) and tert-butyl
methyl ether (10 L). After granulating overnight at ambient
temperature and at 0 °C for 1 h, the product was filtered
and washed with methyl isobutyl ketone and tert-butyl
methyl ether (1:3, 2 L). The product was finally dried at 60
1
(18.1 kg, 88%) as a white powder: mp 223-224 °C. H
NMR (DMSO) δ 5.07 (2H, d), 5.24 (2H, d), 6.77-6.83 (1H,
m), 7.00-7.18 (2H, m), 7.75 (2H, s), 8.53 (2H, s). Found:
+
C, 40.28; H, 3.39; N, 21.63; [MH] 387.0786. C13
H
13
F
2
N
6
O
4
P
+
requires: C, 40.43; H, 3.39; N, 21.78; [MH] 387.0782.
Benzyl 2-(2,4-difluorophenyl)-1,3-bis(1H-1,2,4-triazole-
1-yl)-2-propyl hydrogen phosphate (5) characterisation
1
°C under vacuum for 18 h to afford the title compound (2.26
data: H NMR (DMSO) δ 4.99 (2H, d), 5.11 (2H, d), 5.26
kg, 81%) as a white solid.
(2H, d), 6.77-6.83 (1H, m), 7.00-7.18 (2H, m), 7.30-7.42
+
Method 3. A mixture of 2-(2,4-difluorophenyl)-1,3-bis-
(5H, m), 7.81 (2H, s), 8.59 (2H, s). Found: [MH] 477.1255.
+
(1H-1,2,4-triazole-1-yl)-2-propan-2-ol (79.8 kg, 261 mol) and
C
H F
20 19 2
N O
6 4
P requires: [MH] 477.1246.
pyridine (68.0 kg, 861 mol) in dichloromethane (640 L) was
2-(2-Fluorophenyl)-1,3-bis(1H-1,2,4-triazole-1-yl)-2-
stirred at -13 °C under nitrogen. Phosphorus trichloride
propyl dihydrogen phosphate (6) characterisation data:
H NMR (DMSO) δ 5.13 (2H, d), 5.28 (2H, d), 6.88-6.94
1
(38.0 kg, 276 mol) was added to the mixture over 5 min
followed by a dichloromethane (30 L) wash. The mixture
was then reacted at 13 °C for 2 h before benzyl alcohol (65.0
kg, 602 mol) was added over 2 h, maintaining the temper-
ature between 14 and 16 °C. A dichloromethane (30 L) line
wash was added, and the mixture was stirred between 10
and 15 °C for 2 h. The mixture was then cooled to 0 °C,
and 30% hydrogen peroxide (106 kg, 935 mol) was added
over 3 h, maintaining the temperature below 20 °C. After
reacting for 1 h, the mixture was allowed to settle. After
separating, the organic phase was washed with sodium
metabisulfite (12 kg in 160 L of water), dilute hydrochloric
acid (6 L of concentrated acid in 160 L of water), and water
(1H, m), 7.03-7.11 (2H, m), 7.21-7.26 (1H, m), 7.72 (2H,
+
s), 8.52 (2H, s). Found: [MH] 369.0868. C13
requires: [MH] 369.0871.
H14FN O
6 4
P
+
Acknowledgment
We acknowledge the advice and assistance of many Pfizer
colleagues throughout this development work. In particular
we thank Peter Stephenson and the late Bill Murtiashaw for
their input during the early stages of the project, Steve Ray
and Ciaran Byrne for their helpful suggestions during transfer
to production, and Steve Robinson, Gina Coghlan, Jon
Beaman, and David Andrews for their analytical support.
(320 L). The product solution was then concentrated under
vacuum and the dichloromethane replaced with methyl
isobutyl ketone (229 L) and tert-butyl methyl ether (296 L).
After granulating at 2 °C for 1 h the product was filtered
Received for review August 10, 2001.
OP010064+
112
•
Vol. 6, No. 2, 2002 / Organic Process Research & Development