Application of an enantiomerically pure bicyclic thiolactone in the synthesis of a farnesyl transferase inhibitor

Article abstract of DOI:10.1021/op700218j

An efficient manufacturing route to a novel farnesyl transferase inhibitor is described. The target molecule is a pro-drug, and its synthesis is complicated by the presence of labile functionality. The Medicinal Chemistry synthesis required trityl mercapt

Full text of DOI:10.1021/op700218j

Organic Process Research & Development 2008, 12, 202–212
Application of an Enantiomerically Pure Bicyclic Thiolactone in the Synthesis of a
Farnesyl Transferase Inhibitor
Sahar Abbas, Leigh Ferris, Alison K. Norton, Lyn Powell, Graham E. Robinson,* Paul Siedlecki, Rebecca J. Southworth,
Andrew Stark, and Emyr G. Williams
Process R&D, AstraZeneca, Silk Road Business Park, Charter Way, Macclesfield, Cheshire SK10 2NA, U.K.
Abstract:
Results and Discussion
Medicinal Chemistry Synthesis. The synthetic route to
AZD3409 used in Medicinal Chemistry was long, linear and
involved several unsatisfactory features, notably use of trityl
mercaptan to introduce the protected thiol and use of the
Weinreb amide 2 as a precursor for the reductive amination
(Scheme 1). Additionally, the presence of two labile groups in
the molecule posed problems on scale-up. This synthetic route
was used to make Campaigns 1 and 2 which delivered a total
of 20 kg of drug substance. There were significant quality issues.
Increased levels of impurity in the penultimate intermediate
affected the performance of the next stages, proving that the
crystallization processes were far from robust. An attempt to
improve the quality of the crude product by chromatography
on alumina was unsuccessful and merely resulted in loss of
one-third of the material on the stationary phase.
Synthetic Strategy. Trityl mercaptan is not readily available
in bulk and is a very atom-inefficient way of introducing a sulfur
atom. A more efficient way of generating the thiol at C-4 was
sought. N-Protected bicyclic thiolactones have found utility in
the synthesis of carbapenems such as meropenem.2,3 In this case,
the protected thiolactone is reacted with an amine to give a
proline amide containing the thiol at C-4. It occurred to us that
it should be possible to reduce thiolactone 11 to thiolactol 12.
The latter is a masked aldehyde which should be capable of
undergoing the required reductive amination whilst at the same
time liberating the thiol at C-4 (Scheme 2).
Manufacture of Amine 10. Stilbene 9 was prepared in 90%
yield by a Heck reaction between 4-fluorostyrene and methyl
2-chloro-5-nitrobenzoate. Catalytic reduction of the nitro and
alkene groups occurred in the same process, affording ester 6a
which was hydrolysed to acid 6b in 82% yield over the two
steps. The coupling between acid 6b and methionine isopropyl
ester required mild conditions to avoid racemisation or hy-
drolysis of the isopropyl ester. Several amide-coupling proce-
dures were screened, and the method using EDCI/HOBT/NMM
in NMP gave the best results. Reaction conditions had to be
defined carefully due to the risk of competition between the
two different amino groups. In particular, at low pH values
the rate of the desired reaction is reduced by protonation of the
methioninate amino group. Byproduct 13 from the reaction of
acid 6b first with itself and then with methionine isopropyl ester
(or from the reaction between amine 10 and acid 6b) was
observed, but its level in the product was controlled at about
An efficient manufacturing route to a novel farnesyl transferase
inhibitor is described. The target molecule is a pro-drug, and its
synthesis is complicated by the presence of labile functionality. The
Medicinal Chemistry synthesis required trityl mercaptan to
introduce a thiol group stereospecifically. An important objective
of a new route was avoidance of such an atom-inefficient protecting
group, and this was achieved by use of a bicyclic thiolactone.
Reduction of the thiolactone with DIBAL afforded a masked
aldehyde which participated cleanly in the key reductive amination
step without loss of stereochemical integrity. The reported pro-
cedure for making the thiolactone was found to give inconsistent
results. Development work resulted in a telescoped process that
was operated successfully and reproducibly on the large scale.
Removal of an N-Boc protecting group in the final step of the drug
synthesis required careful choice of conditions to avoid cleaving
other ester groups in the molecule. An impurity formed in the
deprotection step was identified as the S-tert-butyl analogue arising
from attack of the tert-butyl cation on the methionine residue; its
identity was confirmed by independent synthesis.
Introduction
AZD3409 is a novel, orally active antiproliferative agent with
potential application in the treatment of breast cancer and other
tumours.1 It acts by inhibiting farnesyl transferase and gera-
nylgeranyl transferase, two enzymes which prenylate Ras
proteins in cells. These proteins are involved in signaling
processes. Mutant forms of Ras protein are found in many types
of cancers where they cause uncontrolled cell division inde-
pendent of external signals. Prenylation leads to modification
of the protein structure by addition of a lipid chain to the
carboxy-terminus, thus allowing the protein to dock into the
cell membrane and initiate further biological activity. AZD3409
is unusual in being a double pro-drug with the active species
being the thiol acid. The nicotinoyl group is rapidly cleaved in
the gut to release the thiol which is detected in plasma and is
able to pass into cells. The isopropyl ester is removed by the
action of an intracellular esterase. This article describes the
development of an efficient synthetic route capable of being
operated on a commercial scale.
* Author for correspondence. Telephone: +44 (0)1625 514205. E-mail:
graham.robinson@astrazeneca.com.
(1) (a) Stephens, T. C.; Wardleworth, M. J.; Matusiak, Z. S.; Ashton, S. E.;
Hancox, U. J.; Bate, M.; Ferguson, R.; Boyle, T. Proc. Am. Assoc.
Cancer Res. 2003, 44, R4870. (b) Bell, I. M. J. Med. Chem. 2004, 47,
1869.
(2) Brands, K. M. J.; Marchesini, G.; Williams, J. M.; Dolling, U.-H.;
Reider, P. J. Tetrahedron Lett. 1996, 37, 2919.
(3) Matsumura, H.; Bando, T.; Sunagawa, M. Heterocycles 1995, 41, 147.
202
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development
10.1021/op700218j CCC: $40.75
2008 American Chemical Society
Published on Web 02/16/2008

Scheme 1. Medicinal Chemistry Route to AZD3409
Scheme 2
1% with the final process (Scheme 3). The kinetics of the
process were studied and have been reported elsewhere.4 The
first process to be scaled up suffered from a lengthy workup
involving multiple extractions and washes. In a subsequent
improvement, crystalline amine 10 was isolated in 89% yield
simply by adding water to the reaction mixture in a carefully
controlled manner to avoid the problem of oiling out. The batch
cycle time was dramatically reduced.
of intermediates were prepared from D-methionine isopropyl
ester. The L-and D-isomers of amine 10 could not be separated
using a range of chiral HPLC systems. Despite a considerable
amount of work, isomer separation in the downstream com-
pounds proved impossible using standard methodology. Eventu-
ally, capillary electrophoresis (CE) provided a way forward by
demonstrating that amine 10 manufactured on the plant scale
had >99.6% ee.
Reductive Amination. Several reagents were investigated
in the reductive amination step, and 8 M borane ·pyridine
complex was chosen for plant-scale operation. An acid catalyst
is necessary to achieve an acceptable rate of reduction, and
It was necessary to confirm the retention of stereochemical
integrity in the methionine residue throughout the synthesis.
This posed a significant analytical challenge. Reference samples
(4) Chan, L. C.; Cox, B. G. J. Org. Chem. 2007, 72, 8863.
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
203

Scheme 3
Scheme 4
chloroacetic acid was used to good effect in initial experiments.
Later this was discarded in favour of HCl in isopropanol due
to the possibility of carry-over into the next step and the risk
of transesterification between chloroacetic acid and the nicoti-
noyl ester (or formation of an activated amide with excess CDI).
Thiol 7 is an oil and likely to be susceptible to oxidation on
storage so there was no benefit in isolating the product from
the reductive amination. Instead, thiol 7 was reacted in situ with
the active amide prepared from nicotinic acid and CDI to give
thioester 8 which was isolated as a crystalline solid in 76%
yield from thiolactol 12. It had been assumed that thiolactol 12
would react via the open-chain aldehyde form. In view of the
acidity of protons alpha to a carbonyl or imino group, there
was concern that epimerisation would occur at C-2 of the
pyrrolidine ring during the reductive amination. In the event
and to our surprise, none of the trans-diastereomer was found
in the product (limit of detection <0.05%) (samples of both
trans-isomers were prepared by independent synthesis from the
appropriate cis-N-Boc-4-hydroxyproline using the route shown
in Scheme 1).
Mixing thiolactol 12 and amine 10 at room temperature leads
to rapid formation of cyclic thioaminal 14 (this conversion can
be monitored by HPLC canlysis) followed by slow and pH-
dependent reduction to thiol 7 (Scheme 4). Presumably,
existence of the intermediate in a cyclic form rather than the
open-chain imine protects it against epimerisation.
Regioselectivity of Acylation. Typical preparations of
thioester 8 afforded up to 1% of an impurity 16 containing an
additional nicotinoyl group on the aniline nitrogen. The impurity
can be produced by either of two pathways: acylation on N
followed by acylation on S or vice versa (Scheme 5). During a
programme of work aimed at synthesising key impurities, it
became apparent that the impurity cannot be prepared by
acylation of thioester 8; several attempts with different amide
coupling reagents (CDI, EDCI etc) failed to produce any of
the required compound. The importance of understanding the
origin of this impurity in the process was highlighted during a
manufacturing campaign when the second plant batch (73 kg
scale) generated 6% of the impurity. Although the impurity was
removed during downstream processing, its presence at such a
high level had a significant impact on the overall yield. The
unusually high level of impurity was attributed to an extended
addition time for the nicotinoyl imidazolide solution (2.5 h
versus the desired 1 h or less) which was caused by a blocked
orifice plate. This effect was simulated in the laboratory.
Conversely, an experiment in which the nicotinoyl imidazolide
solution was added quickly produced only 0.04% of the
impurity.
A possible explanation for this phenomenon involves general
base catalysis. The pKa values for the aniline and thiol groups
of 7 in water are estimated to be ∼4 and ∼10, respectively.
However, there appears to be relatively little information on
the ionisation of thiols in acetonitrile. A pKa value of 19 is
reported for 2-mercaptophenol in acetonitrile5a (this is due to
the weakly solvating effect of acetonitrile), and the figure for
the thiol group in 7 is likely to be even higher. The pKa value
for the aniline group in 7 is predicted to be ∼10.5b It seems
highly unlikely that the thiol function in 7 reacts via the ionised
form under the process conditions. Imidazole has the potential
to influence the selectivity of the reaction by acting as a general
base catalyst. It is reasonable to expect imidazole to catalyze
acylation of the thiol rather than the aniline due to their very
204
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

Scheme 5
Scheme 6
different pKa values. The abnormally long addition time that
occurred on the plant scale led to an extended period of time
where the reaction mixture was deficient in imidazole, resulting
in more of the impurity 16 being produced through nonselective
acylation; a low level of monoacyl compound 15 was also
detected during analysis of the product. Further experimental
evidence concerning the influence of imidazole was obtained.
When the acid charge in the reaction was increased, the amount
of impurity 16 also increased. On the other hand, when one
molar equivalent of imidazole was added to the reaction mixture
prior to the addition of nicotinoyl imidazolide, impurity 16 was
not produced. Addition of imidazole to the reaction mixture
would be a suitable precautionary measure in the event of
equipment failure.
Manufacture of Thiolactone 11. The published procedure
for making thiolactone 11 involves a multistage telescoped
process starting from N-Boc-4-hydroxyproline 1 (Scheme 6).2
Much experimental work was carried out within AstraZeneca
in an attempt to develop a process that could be scaled up
reliably. Initial investigations were hampered by the lack of a
robust quantitative analytical method (due to the relative
instability of the intermediates), and attempts to carry out the
preparation on any scale above 20 g resulted in greatly
diminished yields. Significant amounts (about 20%) of methyl
thioester 19b were observed when methyl chloroformate was
used to activate the carboxylic acid. Although the level of this
impurity was reduced to <5% in isolated thiolactone 11, the
yield loss was considered unacceptable. Ethyl chloroformate
gave poor-quality product (purity about 70% by HPLC) in
variable yield. Benzyl and phenyl chloroformate gave little or
none of the required product.
Finally, isobutyl chloroformate was chosen in place of the
methyl or ethyl analogues for activation of the acid group. This
had two advantages: the greater stability of the mixed anhydride
allowed the reaction to be monitored reliably by HPLC analysis,
and unwanted transfer of the alkyl group from the chloroformate
to give the isobutyl analogue of 19b was avoided.
Analytical monitoring of preparations which gave a
poor yield of thiolactone 11 indicated loss of product
during the workup. The stability of thiolactone 11 in
toluene solution was tested. Exposure to aqueous hydro-
chloric acid, sodium bicarbonate, sodium hydroxide, or
sodium sulfide at 45 °C for 3 h caused varying degrees
of decomposition and generated predictable byproduct
resulting from cleavage of the thiolactone group (including
disulfides). However, this did not explain the observation
that virtually complete disappearance of product some-
times occurred even when significant levels of the
expected impurities were not detected. In these cases,
disappearance of thiolactone 11 was accompanied by
precipitation of a very fine, white solid which proved
difficult to analyse as it was virtually insoluble in most
solvents. An NMR spectrum showed the solid to be closely
related to thiolactone 11, and MS analysis showed a spread
of high molecular weights indicative of a polymer.
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
205

Scheme 7
Anionic polymerization of thiolactones (e.g., induced by
methods for its removal have been described, but there does
not appear to be a single method of choice for scale-up. Early
manufacturing campaigns of AZD3409 used sulfuric acid (5
mol equivalents) in a mixture of toluene and isopropanol (35–40
°C for 6 h), but this generated 6–9% of an N-tert-butyl impurity,
resulting in a crude product that was difficult to purify. A
publication from Pfizer prompted us to develop a process using
85% phosphoric acid in 2-methyltetrahydrofuran (35 °C for 1 h)
that afforded crude AZD3409 in 88% yield.7 Whilst this process
worked well on the kilo scale, there were significant issues when
it was operated in the pilot plant. Use of a very large excess
(34 mol equivalents) of phosphoric acid and a correspondingly
large amount of aqueous base during the workup resulted in
long addition times because of the need to control exotherms.
The problems were exacerbated by severe foaming that occurred
whilst the batch was being diluted with water. As a consequence
of the extended processing times, significant degradation of the
product occurred through hydrolysis of the labile ester groups;
impurity levels in the isolated product totalled 10% (HPLC
area). Further experiments showed that a significant reduction
in phosphoric acid usage could only be achieved at the expense
of very long reaction times. The environmental burden of the
process was also considered unacceptable for long-term use,
and it was necessary to re-examine the alternatives.
n-butyl lithium) has been reported.6 To test whether this could
explain the observed phenomena in the present case, a solution
of thiolactone 11 in THF was exposed to n-butyl lithium (2.5
mol %) in hexane at room temperature. The substrate disap-
peared, but no degradation products were observed by HPLC
analysis. Similar results were seen with some of the ingredients
used in the thiolactone process, for example diisopropylethy-
lamine hydrochloride (1 mol equivalent) and sodium sulfide (5
mol %) in water.
To reduce the risk of capricious and potentially catastrophic
polymerisation, key changes were made to the process: the
charge of sodium sulfide was reduced from 1.15 to 1.05 mol
equivalent, the reaction mixture was quenched with an aqueous
solution of citric acid, and additions of aqueous hydrochloric
acid and aqueous sodium bicarbonate were omitted from the
workup. This minimised the exposure of thiolactone 11 to any
residual sulfide or chloride anions, particularly during the
distillation stages. The improved process was successfully scaled
up (at a contract manufacturer) to produce over 1000 kg of
thiolactone 11 in an average yield of ∼76% (99.6% ee, total
organic impurities <1% w/w by GC) (an authentic sample of
the enantiomeric thiolactone was prepared for use as an
analytical reference standard).
DIBAL Reduction of Thiolactone 11. In early development
work, the reduction of thiolactone 11 with DIBAL was
performed at -70 °C to avoid the expected over-reduction of
the tautomeric aldehyde. Further investigation led to the
conclusion that thiolactol 12 is resistant to over-reduction as
similar yields (70–75%) and quality are obtained at -70 °C
and at around 0 °C with approximately 10% of product lost to
the crystallisation mother liquors. A satisfactory mass balance
was not obtained with the process; neither HPLC nor GC
analysis revealed significant amounts of any other components.
It is presumed that some loss of the Boc group occurred during
the workup. On scale-up, the reduction step was carried out at
0–5 °C, and it was beneficial to add the reaction mixture to a
warm aqueous solution of Rochelle salt rather than vice versa
as this allowed the two liquid phases to be separated more easily.
Crystallisation from toluene/n-heptane afforded dense, crystal-
line product of high purity. The average yield in plant
manufacture was 73%.
Hydrogen chloride (5 N) in isopropanol was selected for
further evaluation. Conditions were chosen to minimise cleavage
of other sensitive ester groups in the molecule. A simple and
efficient process was developed which scaled up successfully
in the pilot plant, affording crude AZD3409 in about 80% yield.
The product was purified through conversion to its malate salt.
Formation of an Unexpected Impurity during Depro-
tection. The identity of an impurity present at levels of 0.1–0.3%
in pure AZD3409 HCl salt prepared by the phosphoric acid
deprotection process was puzzling. LC-MS indicated a mass
of 694 which is 42 units higher than that of AZD3409. Initially,
the impurity was thought to be an N-acetyl derivative of
AZD3409, but its HRMS suggested that incorporation of an
isopropyl group was more likely. However, this suggestion was
disproved when an authentic sample of the isopropyl compound
20 was prepared by reductive amination with acetone and
sodium triacetoxyborohydride (Scheme 7).
Material recovered from AZD3409 mother liquors was
subjected to repeated crystallization followed by careful chro-
matography. This gave a sufficiently enriched sample of the
impurity to allow its tentative identification by NMR as the
S-tert-butyl derivative 22b. Evidently a proportion of the tert-
Boc Deprotection to Generate AZD3409. Use of the Boc
protecting group in drug synthesis is commonplace. Numerous
(5) (a) Izutsu, K. Acid-Base Dissociation Constants in Dipolar Aprotic
SolVents; IUPAC Chemical Data Series, No 35; Blackwell: Cambridge,
MA, 1990; p 23. (b) Izutsu, K. Acid-Base Dissociation Constants in
Dipolar Aprotic SolVents; IUPAC Chemical Data Series No 35;
Blackwell: Cambridge, MA, 1990; p 17.
(7) Li, B.; Bemish, R.; Buzon, R. A.; Chiu, C. K.-F.; Colgan, S. T.; Kissel,
W.; Le, T.; Leeman, K. R.; Newell, L.; Roth, J. Tetrahedron Lett. 2003,
44, 8113.
(6) Overberger, C. G.; Weise, J. K. J. Am. Chem. Soc. 1968, 90, 3533.
206
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

Scheme 8
Scheme 9
Since the use of hydrogen fluoride was unattractive, a
modified process was devised involving sulfuric acid and tert-
butanol followed by neutralization with calcium carbonate.
Application of this method to AZD3409 did not produce the
desired S-tert butyl impurity 22b due to the unstable nature of
the thioester group. Success was achieved starting with the
aniline precursor 10 (Scheme 10). The desired sulfonium ion
was formed (together with a small amount of the corresponding
acid resulting from hydrolysis of the isopropyl ester). Treatment
with sodium 2-hydroxyethylsulfide followed by chromatography
on silica afforded the S-tert-butyl derivative 25 in 30% yield.
Subjecting this to standard coupling conditions gave 22a which
was deprotected to give the required 22b.
Form and Quality of the Drug Substance. Initial small
batches of AZD3409 were prepared as an amorphous free base.
It quickly became apparent that the stability of the free base
was inadequate for development as a solid dosage form. In
particular, the nicotinoyl group has a propensity to migrate onto
the proline nitrogen when the latter is unprotonated. Various
salts were screened for crystallinity, but initially only the
monohydrochloride gave a positive result. This form showed
moderate stability and was selected for use in early clinical
studies where it was administered in capsules. Work aimed at
developing a tablet formulation showed the hydrochloride to
be unsuitable because it became amorphous on compression
and this led to unacceptable degradation. This threatened the
butyl carbocation generated in the deprotection is trapped by
the sulfur atom of the methionine moiety to give sulfonium
ion 21 (Scheme 8). In the presence of a suitable nucleophile
this could undergo cleavage of any of the bonds to sulfur: attack
at the methyl group would give the S-tert-butyl impurity, attack
at the tert-butyl group (disfavoured on steric grounds) would
regenerate AZD3409, and attack at the methylene group would
generate methyl tert-butyl sulfide (possibly with formation of
a five-membered lactone by intramolecular involvement of the
ester group). There was no evidence for an impurity derived
from the last pathway.
Preparation of S-tert-Butyl Impurity 22b. To confirm the
identity of impurity 22b and provide an analytical reference
sample, an independent preparation was required. The conver-
sion of Boc-methionine into the tert-butyl sulfonium ion with
hydrogen fluoride followed by ion-exchange chromatography
and then treatment with sodium 2-hydroxyethylsulfide to give
S-tert-butylhomocysteine has been reported (Scheme 9).8
Scheme 10
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
207

long-term viability of the drug, so further screening of potential
salts was initiated, leading to the discovery of a crystalline salt
with L-malic acid. This salt (which could be obtained in high
purity) proved to have superior stability and formulation
characteristics to the hydrochloride and formed the basis of a
viable tablet formulation for use in phase II clinical studies.
The thioester and isopropyl ester groups in AZD3409 are
chemically labile and, not surprisingly, thiol 26 (together with
disulfide 28) and acid 27 were detected in all samples of drug
substance. These compounds are recognised metabolites of
AZD3409 and were qualified at relatively high levels in
toxicological studies. Accordingly, control levels in the drug
substance specification reflected this: 26 (1% maximum), 27
(1%), and 28 (2%). In addition, the sulfoxide (of the methionine
fragment) was controlled at 0.5%. In plant manufacture of the
malate salt, all these impurities were well within the required
limits (total organic impurities 2.4% w/w). The enantiomeric
purity of individual stereocentres was controlled by appropriate
analysis of key intermediates as described in earlier sections of
this paper.
Avatar FT-IR instrument. Melting points were obtained with a
Mettler Toledo DSC822e instrument. NMR spectra were
obtained using Bruker DPX 400, DRX 500, and Avance 600
1
instruments; H spectra were measured with reference to an
internal standard of tetramethylsilane at 0 ppm, and 13C spectra
were measured with reference to the DMSO signal at 39.5 ppm.
LC-MS and HRMS data were obtained using Waters Micro-
mass ZMD4000 and Waters Micromass LCT Classic instru-
ments, respectively. HPLC analyses were performed with an
Agilent 1100 instrument using the following conditions: syn-
thetic stages to acid 6b, HiChrom HI-5C18 column (4.6 mm
× 150 mm, 5 µm) at 40 °C and 220 or 230 nm with a flow
rate of 1.5 mL/min and a mobile phase gradient consisting of
5–90% CH3CN in water (with a constant level of 0.1% TFA
in the mixture); synthesis of thiolactone 11, Alltima HP C18EPS
column (3.0 mm × 150 mm, 3 µm) at 20 °C and 205 nm with
a flow rate of 0.43 mL/min and a mobile phase gradient
consisting of 5–85% CH3CN in water (with a constant level of
0.05% TFA in the mixture); conversion of thiolactone 11 to
thiolactol 12, Waters Spherisorb ODS2 column (4.6 mm × 100
mm, 3 µm) at 30 °C and 205 nm with a flow rate of 0.7 mL/
min and a mobile phase consisting of CH3CN and water (1:1
mixture containing 0.1% methanesulfonic acid); synthetic stages
from acid 6b to AZD3409, Jones Genesis C18 column (4.6
mm × 150 mm, 4 µm) at 50 °C and 265 nm with a flow rate
of 1.5 mL/min and a mobile phase gradient consisting of
40–90% CH3CN in water (with a constant level of 0.1% TFA
in the mixture) (approximate RRTs are 10, 0.40; 8, 1.00; 7,
1.04; 14, 1.08). The enantiomeric purity of thiolactone 11 was
determined by HPLC using the following conditions: Chiralpak
AD-H column (4.6 mm × 150 mm, 5 µm) at 15 °C and 210
nm with a flow rate of 1.0 mL/min and a mobile phase of
MeOH (approximate retention times R,R-isomer 2.61 min, S,S-
isomer 3.46 min). The chemical purity of thiolactone 11 was
measured with an HP6890 gas chromatograph equipped for
cool-on-column injection and fitted with a flame ionisation
detector using the following conditions: DB17MS capillary
column (15 m × 0.25 mm, 0.15 µm film), 40 °C for 1 min, 40
°C/min to 160 °C (no hold), 5 °C/min to 250 °C (no hold), 50
°C/min to 300 °C (5 min hold), carrier gas He at 1.2 mL/min.
Conclusion
A new, convergent route has been developed for large-scale
manufacture of the candidate drug AZD3409. Undesirable
features of the Medicinal Chemistry route (e.g., use of trityl
mercaptan and formation of a Weinreb amide) were conve-
niently avoided by employing a bicyclic thiolactol (itself derived
from a thiolactone) as a substrate in the key reductive amination
step. Problems of reproducibility in the preparation of the
thiolactone were solved, allowing >1000 kg of this intermediate
to be manufactured. From N-Boc-4-hydroxyproline was manu-
factured 170 kg of AZD3409 malate in 31% overall yield.
Selective removal of the N-Boc protecting group in the presence
of other labile esters was achieved with hydrogen chloride in
isopropanol. An impurity produced at low levels during the Boc
deprotection step was identified as the S-tert-butyl analogue
resulting from modification of the methionine fragment.
Experimental Section
General. Starting materials, reagents, and solvents were
obtained from commercial suppliers and used without further
purification. IR spectra were recorded using a Thermo Electron
(8) Chassaing, G.; Lavielle, S.; Marquet, A. J. Org. Chem. 1983, 48, 1757.
208
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

The enantiomeric purity of amine 10 was determined by CE
with a Beckman Coulter P/ACE MDQ instrument using the
following conditions: polyimide-coated fused silica capillary (50
µm i.d., 365 µm, 50 cm total length, 10 cm effective length),
capillary temperature 20 °C, 100 mM lithium phosphate buffer
(pH 2.5) containing 10 mM HS-ꢀ-CD, voltage +20 kV
(increased from 0 to 20 kV over 0.3 min), current draw 90 µA
(approx), injection of 15 s at 0.5 psi (hydrodynamic, capillary
inlet at cathode) ,and detection at 200 nm (approximate
migration times L-isomer 4.8 min, D-isomer 5.4 min).
in a precooled filter-drier, washed with a mixture of hexane
(49.5 L) and TBME (21.6 L), cooled to -8 °C, and dried in
Vacuo at 30 °C. The yield of thiolactone 11 was 39.5 kg (80%
based on trans-Boc-(2S,4R)-4-hydroxyproline). Chiral HPLC
assay: 99.6% ee.
1H NMR (400 MHz, 343 K, DMSO-d6) δ 4.44 (1H, br s),
4.34 (1H, br m), 3.75 (1H, dd, J ) 10.2, 2.7 Hz), 3.41 (1H, dd,
J ) 10.2, 1.6 Hz), 2.17 (1H, dt, J ) 11.5, 2.5 Hz), 2.11 (1H,
dq, J ) 11.5, 1.6, 1.2 Hz), 1.41 (9H, s).
Preparation of (1S,4S)-3-Hydroxy-2-thia-5-azabicyclo-
[2.2.1]heptane-5-carboxylic Acid tert-Butyl Ester (12). A
mixture of thiolactone 11 (66.5 kg, 290 mol) and toluene (268
L) was stirred and cooled to -5 °C. A 1.0 M solution of
diisobutylaluminium hydride (DIBAL) in toluene (363 L, 363
mol) was added over about 1 h, keeping the temperature below
5 °C. When the reaction was complete, acetone (133 L) was
added. The reaction mixture was then added to a solution of
sodium potassium tartrate tetrahydrate (145 kg, 507 mol) in
water (145 L) at 60 °C at such a rate as to control the gas
evolution. The mixture was cooled to 20 °C, and the lower
aqueous phase was separated and discarded. The organic
solution was concentrated by distillation under reduced pressure
to 335 L. The temperature of the concentrated solution was
adjusted to 65 °C and diluted with n-heptane (335 L). The
solution was cooled to 20 °C over 1.5 h and held at that
temperature for 1 h. The crystalline product was collected by
filtration, washed with n-heptane (168 L), and dried in Vacuo
at 40 °C. The yield of thiolactol 12 (mean of three batches)
was 47.9 kg (73%).
Preparation of (1S,4S)-3-Oxo-2-thia-5-azabicyclo-
[2.2.1]heptane-5-carboxylic Acid tert-Butyl Ester (11). A
solution of trans-N-Boc-(2S,4R)-4-hydroxyproline 1 (49.45 kg,
213 mol) in tetrahydrofuran (THF) (725 L) was stirred and
cooled to -22 °C. A temperature of about -20 °C was
maintained throughout the following additions required to
produce mesylate ester 18. Diisopropylethylamine (62.0 kg, 479
mol) was added over 35 min followed by a line wash of THF
(11.9 L). Isobutyl chloroformate (32.0 kg, 234 mol) was added
over 30 min (slightly exothermic) followed by a line wash of
THF (11.9 L), and the mixture was stirred for 32 min.
Triethylamine (23.8 kg, 234 mol) was added over 22 min
followed by a line wash of THF (11.9 L). After 15 min,
methanesulfonyl chloride (27.0 kg, 236 mol) was added over
2.25 h (exothermic) followed by a line wash of THF (11.9 L),
and the mixture was stirred for a further 3 h. In a second reactor,
hydrated sodium sulfide (29.2 kg at 60% w/w content, 224 mol)
was dissolved in water (128 kg) and cooled to -5 °C (below
this temperature the sodium sulfide crystallizes out). The sodium
sulfide solution was added as quickly as possible to the reaction
mixture containing the mesylate ester 18; the addition was
completed in 100 s, and the temperature rose from -22 to -8
°C. A rinse wash of water (108 kg) was applied to the vessel
that contained the sodium sulfide solution and thence to the
main reactor. The reaction mixture was warmed to +20 °C over
about 40 min then agitated at this temperature for 6 h. A solution
of citric acid monohydrate (69.9 kg, 332 mol) in water was
added to the reaction mixture over about 30 min followed by
a rinse wash of water (21.4 kg). The mixture was agitated at
20 °C for about 30 min, and then the layers were separated.
The organic phase was concentrated by distillation under
reduced pressure (238 to 60 mbar) at 26–28 °C until 670 L of
distillate was collected. The aqueous phase was extracted with
tert-butyl methyl ether (TBME) (375 L), and the extract was
combined with the residue obtained by concentrating the organic
phase; the residue dissolved to give a solution at 25 °C. The
organic solution was washed with water (161 L), then with 10%
w/w aqueous brine (161 kg), and finally with more water (161
kg). The organic solution was then concentrated by distillation
under reduced pressure (293 to 94 mbar) at 25–30 °C until 390
L of distillate was collected. The residue was dissolved in more
TBME (215 L), and the solution was again concentrated by
distillation under reduced pressure (275 to 160 mbar) at about
25 °C until 205 L of distillate was collected. The residue was
dissolved in more TBME (72 L), and the solution was cooled
to -8 °C over 1 h 20 min and agitated at this temperature for
2 h. The suspension was diluted with hexane (99 L) and agitated
at -5 °C for another 2.5 h. The crystalline solid was collected
1H NMR (400 MHz, 343 K, DMSO-d6) δ 6.07 (1H, br d, J
) 6.6 Hz), 5.29 (1H, br d, J ) 5.3 Hz), 4.20 (1H, br s), 3.75
(1H, br t), 3.37 (1H, br d, J ) 9.7 Hz), 2.11 (1H, br d, J )
10.4 Hz), 1.97 (1H, br d, J ) 10.1 Hz), 1.41 (9H, s). IR νmax
(KBr disk) 3380br, 2972, 1666 cm-1.
Preparation of Methyl 2[(E)-2-(4-fluorophenyl)vinyl]-5-
nitrobenzoate (9). A mixture of methyl 2-chloro-5-nitroben-
zoate (31.0 kg, 144 mol), sodium carbonate (16.0 kg, 151 mol),
tetra-n-butylammonium bromide (4.67 kg, 14.5 mol), 4-fluo-
rostyrene (22.5 kg, 184 mol) and N,N-dimethylacetamide (110
kg) was stirred at ambient temperature under a nitrogen
atmosphere. Palladium chloride (1.06 kg, 5.98 mol) and triethyl
phosphite (1.01 kg, 6.11 mol) were added, and the mixture was
heated at 90 °C for 4 h. The hot mixture was diluted with
toluene (82.5 kg) and filtered to remove catalyst and other
insoluble material; the filter cake was washed with hot toluene
(34.5 kg). The total filtrate was concentrated by distillation under
reduced pressure until all the toluene had been removed. The
residue was diluted with methanol (159 kg) and cooled and
stirred at -5 °C for 30 min. The crystalline product was
collected by filtration and washed with cold methanol (49 kg).
The methanol-wet solid (60 kg) was used directly in the next
stage without further drying. The yield of stilbene 9 was about
90%.
1H NMR (400 MHz, 300 K, DMSO-d6) δ 8.56 (1H, d, J )
2.1 Hz), 8.39 (1H, dd, J ) 8.8, 2.6 Hz), 8.14 (1H, d, J ) 8.8
Hz), 7.84 (1H, d, J ) 16.4 Hz), 7.72–7.66 (2H, m), 7.47 (1H,
d, J ) 16.8 Hz), 7.31–7.24 (2H, m), 3.94 (3H, s); 13C NMR
(100 MHz, 300 K, DMSO-d6) δ 165.5, 162.4 (d, JCF ) 241.1
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
209

Hz), 145.7, 144.0, 134.0, 132.9 (d, JCF ) 3.0 Hz), 129.3 (d,
JCF ) 8.3 Hz), 129.2, 128.0, 126.4, 125.3, 124.4 (d, JCF ) 2.2
Hz), 115.9 (d, JCF ) 21.8 Hz), 52.8; IR νmax (KBr disk) 1728,
1585, 1505, 1331, 1227 cm-1.
After stirring the mixture at 40 °C for 1 h, the solid was
collected by filtration, washed with water (2 × 420 L), and
dried in Vacuo at 40 °C. The yield of amine 10 (mean of four
batches) was 208.7 kg (89%).
Preparation of Methyl 5-Amino-2-[2-(4-fluorophenyl)-
ethyl Benzoate] (6a). Methanol (216 kg) was added to a
mixture of a stilbene 9 (60 kg of methanol-wet solid containing
an estimated 39 kg at 100% strength, 129 mol) and 10%
palladium on carbon wet paste (water content 57.0% w/w, Pd
content on dry weight 9.84% w/w) (0.78 kg, 0.31 mol). The
mixture was stirred and heated at 35 °C under a hydrogen
atmosphere (3.0 bar gauge) for 2 h. When the reaction was
complete, the mixture was heated to 50 °C and passed through
a filter aid to remove the catalyst; the filter cake was washed
with methanol (26 kg). The total filtrate was diluted with water
(56.7 kg), cooled to -5 °C with stirring and maintained at -5
°C for 30 min. The crystalline product was collected by
filtration, washed with a mixture of methanol (23 kg) and water
(28.7 kg) and dried at ambient temperature and pressure. The
yield of aminoester 6a (mean of four batches) was 31.8 kg
(88%).
1H NMR (400 MHz, 300 K, DMSO-d6) δ 7.24–7.16 (2H,
m), 7.11–7.04 (3H, m), 6.95 (1H, d, J ) 8.4 Hz), 6.69 (1H, dd,
J ) 8.3, 2.6 Hz), 5.16 (2H, br s), 3.80 (3H, s), 2.99–2.92 (2H,
m), 2.76–2.68 (2H, m); MS (ES) m/z 274.1 (M+H)+.
Preparation of 5-Amino-2-[2-(4-fluorophenyl)ethyl Ben-
zoic Acid (6b). A mixture of aminoester 6a (73.3 kg, 268 mol),
methanol (220 L), and 47% w/w sodium hydroxide solution
(34.5 kg, 402 mol) was stirred and heated at 60 °C for 6.5 h.
The solution was cooled to room temperature and adjusted to
pH 5 with 1 M aqueous hydrochloric acid (402 L, about 402
mol). The precipitated solid was collected by filtration, washed
with water (2 × 147 L), and dried in Vacuo at 40 °C. The yield
of acid 6b (mean of three batches) was 65.0 kg (93%).
1H NMR (400 MHz, 300 K, DMSO-d6) δ 7.25–7.17 (2H,
m), 7.11–7.02 (3H, m), 6.90 (1H, d, J ) 8.2 Hz), 6.64 (1H, dd,
J ) 8.1, 2.6 Hz), 3.33 (1H, br s), 3.01–2.93 (2H, m), 2.77–2.69
(2H, m); 13C NMR (100 MHz, 300 K, DMSO-d6) δ 169.2,
160.5 (d, JCF ) 241.1 Hz), 146.6, 138.2 (d, JCF ) 3.0 Hz),
131.4, 130.4, 129.9 (d, JCF ) 7.9 Hz), 129.3, 117.2, 115.5, 114.8
(d, JCF ) 21.1 Hz), 37.0, 35.5; MS (ES) m/z 260.1 (M + H)+,
301 (MH + CH3CN)+.
Preparation of Isopropyl N-{5-Amino[2-(4-fluorophenyl)-
ethyl]benzoyl}-L-methioninate (10). A mixture of L-methion-
ine isopropyl ester hydrochloride (135.3 kg, 594 mol), 1-(3-
dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (113.9
kg, 594 mol) and N-methylpyrrolidin-2-one (NMP) (420 L) was
stirred at about 20 °C. N-Methylmorpholine (27.3 kg, 270 mol)
was added to the mixture followed by a solution of 1-hydroxy-
benzotriazole (29.2 kg, 217 mol) in N-methylmorpholine (87.6
kg, 866 mol) and water (29.1 kg) and a wash of NMP (35 L).
A solution of the amino acid 6b (140.0 kg, 540 mol) in NMP
(350 L) was added over 2 h, keeping the temperature at about
20 °C, and this was followed by a wash of NMP (35 L). The
mixture was maintained at 20 °C for a further 8 h to complete
the reaction. Water (420 L) was added, and after stirring for
1 h at 20 °C the product began to crystallize. The mixture was
warmed to 40 °C, and more water (980 L) was added over 2 h.
1H NMR (500 MHz, 300 K, DMSO-d6) δ 8.56 (1H, d, J )
7.7 Hz), 7.23–7.16 (2H, m), 7.08–7.01 (2H, m), 6.86 (1H, d, J
) 7.9 Hz), 6.60 (1H, d, J ) 2.1 Hz), 6.52 (1H, dd, J ) 8.1, 2.4
Hz), 5.06 (2H, br s), 4.97 (1H, septet, J ) 6.3 Hz), 4.48 (1H,
q, J ) 7.1 Hz), 2.84–2.66 (4H, m), 2.64–2.51 (2H, m), 2.04
(3H, s), 1.99 (2H, q, J ) 6.4 Hz), 1.20 (3H, d, J ) 6.3 Hz),
1.17 (3H, d, J ) 6.2 Hz); 13C NMR (125 MHz, 300 K, DMSO-
d6) δ 171.4, 170.2, 160.5 (d, JCF ) 240.7 Hz), 146.4, 138.2 (d,
JCF ) 2.8 Hz), 136.7, 130.2, 130.0 (d, JCF ) 8.9 Hz), 125.9,
114.8, 114.7 (d, JCF ) 20.7 Hz), 112.9, 67.9, 51.4, 36.8, 34.4,
30.0, 29.8, 21.5, 21.4, 14.5; MS (ES) m/z 433.2 (M + H)+.
Preparation of Isopropyl N-{5-({[(2S,4S)-1-(tert-butoxy-
carbonyl)-4-(pyridin-3-ylcarbonyl)thiopyrrolidin-2-
yl]methyl}amino)-2-[2-(4-fluorophenyl)ethyl]benzoyl}-L-
methioninate (8). A mixture of amine 10 (73.0 kg, 169 mol),
thiolactol 12 (39.4 kg, 170 mol), 5 M hydrogen chloride in
isopropanol (14.6 L, 73 mol), and toluene (507 L) was stirred
at about 20 °C for about 30 min. Then 8 M borane ·pyridine
complex (16.8 L, 134 mol) was added over about 30 min,
followed by a wash of toluene (24.8 L). A further portion of 5
M hydrogen chloride in isopropanol (22.6 L, 113 mol) was
added over about 20 min followed by a wash of toluene (12.4
L). The mixture was stirred for 4.5 h at 20 °C to complete the
formation of thiol 7. 1,1′-Carbonyldiimidazole (CDI) (34.3, 212
mol) was dissolved in acetonitrile (219 L) at 40 °C, and the
solution was added to a stirred suspension of nicotinic acid (24.8
kg, 201 mol) in acetonitrile (146 L), keeping the temperature
below 20 °C, followed by a wash of acetonitrile (53 L). The
mixture was stirred at about 15 °C for 50 min and then added
to the solution of thiol 7 over about 25 min, followed by a wash
of acetonitrile (53 L). The mixture was stirred at about 20 °C
for 4 h and then diluted with water (358 L). The mixture was
heated to 40 °C, and the lower aqueous phase was separated
and discarded. The organic solution was washed with an
additional portion of water (358 L) at 40 °C and then
concentrated by distillation under reduced pressure to about 330
L. The temperature of the residual solution was adjusted to 43
°C, and n-heptane (110 L) was added. The solution was cooled
to about 20 °C, seeded, and held at 20 °C for 5 h until
crystallization was well established. A second portion of
n-heptane (110 L) was added, and the mixture was stirred at
20 °C for 3 h. The crystalline solid was collected by filtration,
washed with a mixture of toluene and n-heptane (1:1 v/v, 146
L) and dried in Vacuo at 40 °C. The yield of thioester 8 (mean
of three batches) was 96.1 kg (76%).
1H NMR (500 MHz, 343 K, DMSO-d6) δ 9.03 (1H, d, J )
2 Hz), 8.83 (1H, dd, J ) 5, 2 Hz), 8.27 (1H, d, J ) 8 Hz), 8.22
(1H, ddd, J ) 8, 2, 2 Hz), 7.58 (1H, dd, J ) 8, 5 Hz), 7.21–7.18
(2H, m), 7.03–6.99 (2H, m), 6.92 (1H, d, J ) 8 Hz), 6.66 (1H,
s), 6.65 (1H dd, J ) 9, 2 Hz), 5.61 (1H, br t, J ) 6 Hz), 4.93
(1H, septet, J ) 6 Hz), 4.51 (1H, q, J ) 7 Hz), 4.11 (1H, qn,
J ) 7 Hz), 4.10–4.04 (1H, m), 4.06 (1H, dd, J ) 11, 7 Hz),
3.49 (1H, ddd, J ) 13, 5, 5 Hz), 3.25 (1H, dd, J ) 11, 6 Hz),
3.15 (1H, ddd, J ) 14, 8, 6 Hz), 2.85–2.79 (2H, m), 2.79–2.74
210
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development

(2H, m), 2.65–2.53 (3H, m), 2.04 (3H, s), 2.05–1.98 (3H, m),
1.46 (9H, s), 1.20 (3H, d, J ) 6 Hz), 1.18 (3H, d, J 6 Hz); 13C
NMR (125 MHz, 343 K, DMSO-d6) δ 189.4, 170.8, 169.6,
160.2 (d, JCF ) 242 Hz), 153.8, 153.4, 147.1, 146.3, 137.8 (d,
JCF ) 3 Hz), 136.7, 134.1, 131.7, 129.9, 129.5 (d, JCF ) 8 Hz),
125.9, 123.7, 114.2 (d, JCF ) 21 Hz), 112.2, 111.3, 79.0, 67.6,
55.5, 52.1, 51.3, 46.2, 38.7, 36.3, 34.4, 33.7, 30.2, 29.7, 27.8,
21.1, 14.3. MS (ES) m/z 753.3 (M+H)+.
170.1, 160.5 (d, JCF ) 240.6 Hz), 154.4, 147.5, 146.2, 138.2
(d, JCF ) 3.3 Hz), 136.8, 134.6, 131.7, 130.4, 130.0 (d, JCF
)
7.5 Hz), 126.6, 124.3, 114.7 (d, JCF ) 20.7 Hz), 113.0, 111.7,
68.0, 66.5, 57.6, 51.5, 51.1, 45.8, 40.7, 39.5, 36.7, 34.8, 34.3,
30.1, 29.9, 21.5, 21.4, 14.5; HRMS (accurate mass) calcd for
C34H41N4O4S2F (free base) 653.2632, found 653.2661. Anal.
calcd for C34H41N4O4S2F·C4H6O5: C, 58.0; H, 6.0; F, 2.4; N,
7.1; S, 8.1. Found: C, 57.7; H, 5.9; F, 2.6; N, 6.8; S, 8.4.
Preparation of Isopropyl (2S)-2-({2-(4-Fluorophenyl)-
ethyl-5-[({(2S,4S)-4-[(pyridin-3-ylcarbonyl)thio]pyrrolidin-
2-yl}methyl)amino]benzoyl}amino)-4-(tert-butylthio)bu-
tanoate (22b). (a). Introduction of S-tert-Butyl Group. Sulfuric
acid (10 mL, 187 mmol) was added to a stirred solution of
amine 10 (25.0 g, 57.8 mmol) and tert-butanol (50 mL, 526
mmol) in toluene (200 mL) at ambient temperature; the
temperature of the mixture rose to 36 °C. The solution was
stirred at 30 °C for 2 h and then at 45 °C for 2 h, whereupon
HPLC analysis showed the sulfonium species to be the
predominant component. The mixture was cooled to 9 °C and
diluted with water (100 mL) and adjusted to pH 7.3 by addition
of precipitated calcium carbonate (26.0 g, 260 mmol). The
mixture was filtered and the filter cake washed with water (50
mL). The layers were separated, and the organic phase was
discarded. 2-Mercaptoethanol (8.90 g, 114 mmol) and 47% w/w
sodium hydroxide solution (5.5 mL, 104 mmol) were added to
the aqueous phase containing the sulfonium species. The
mixture was stirred overnight at ambient temperature and for
30 min at 40 °C, then it was cooled to ambient temperature
and extracted with ethyl acetate (150 mL). The organic phase
was washed with water (100 mL), dried over magnesium
sulfate, and evaporated under reduced pressure to give a yellow
oil (14 g) containing mainly the required product 25 (59% by
area on HPLC analysis). The crude oil was purified in two
portions by chromatography on silica gel 60 (500 g) using ethyl
acetate/isohexane (1:1 v/v) as eluent to give 25 (7.2 g, 26%)
(95% by area on HPLC analysis).
(b). Coupling between tert-Butyl Analogue 25 and Thiolactol
12. S-tert-Butyl sulfide 25 (6.56 g, 13.8 mmol) was dissolved
in toluene (50 mL). The solution was concentrated to about 35
mL under reduced pressure to remove any water or residual
ethyl acetate and then diluted with toluene (20 mL). Whilst the
temperature was maintained at about 20 °C, thiolactol 12 (3.28
g, 14.0 mmol) was added, followed by 5 M hydrogen chloride
in isopropanol (1.22 mL, 6.1 mmol), 8 M borane ·pyridine
complex (1.38 mL, 11.1 mmol), and a further portion of 5 M
hydrogen chloride in isopropanol (1.82 mL, 9.1 mmol). The
solution was stirred at 20–23 °C for 24 h to complete the
reductive amination step. Meanwhile, 1,1′-carbonyldiimidazole
(2.80 g, 17.3 mmol) was added portionwise to a solution of
nicotinic acid (2.04 g, 16.6 mmol) in acetonitrile (35 mL) at
10 °C. The solution was warmed to 20 °C to form the activated
nicotinic acid and then was added to the pyrrolidinethiol
solution. After 3.5 h, HPLC anlysis indicated substantial
conversion to the required Boc-protected thioester 22a. The
mixture was diluted with water (32 mL) and warmed to 43 °C
whereupon the suspended solid dissolved and enabled the layers
to be separated. The organic phase was washed with water (2
× 32 mL) at 40 °C, dried over magnesium sulfate, and
Preparation of Isopropyl N-{5-({[(2S,4S)-4-(Pyridin-3-
ylcarbonyl)thiopyrrolidin-2-yl]methyl}amino)-2-[2-(4-fluo-
rophenyl)ethyl]benzoyl}-L-methioninate L-Malate (AZD3409
Malate). (a). RemoVal of Boc-Protecting Group. A mixture
of thioester 8 (Boc-protected AZD3409) (85.0 kg, 113 mol)
and toluene (255 L) was stirred and heated to 50 °C. 5 N
Hydrogen chloride in isopropanol (90.1 L, 451 mol) was added
over 20 min followed by a wash of toluene (8.5 L). The mixture
was stirred at 50 °C for 4 h then cooled to 20 °C. A solution of
potassium hydrogen carbonate (45.2 kg, 451 mol) in water (340
L) was added followed by a wash of water (17 L), giving a pH
of 7–8. The lower aqueous phase was separated and discarded.
The organic phase was diluted with toluene (425 L) and
concentrated to about 595 L at 40–50 °C under reduced
pressure. The temperature of the residual solution was adjusted
to 45 °C, and seed crystals were added. n-Heptane (85 L) was
added over 20 min, and the mixture was cooled to 20 °C over
3 h. After 1 h at 20 °C, the solid was collected by filtration,
washed with a mixture of toluene and n-heptane (1:1 v/v, 170
L), and dried in Vacuo at 40 °C. The yield of AZD3409 free
base (mean of three batches) was 58.5 kg (79%) (total impurities
2.5–7.0 area % by HPLC).
(b). Salt Formation. AZD3409 free base (40.0 kg, 61.3 mol)
was dissolved in a mixture of ethyl acetate (360 L) and water
(4 L) at 35 °C. The warm solution was screened to remove
any extraneous material, and the filter was washed with ethyl
acetate (40 L). The filtrate was warmed to 40 °C. A solution of
L-malic acid (8.54 kg, 63.7 mol) in water (12 L) was prepared,
and approximately half of this solution was added to the
AZD3409 solution, keeping the temperature at 40 °C. Seed
crystals of AZD3409 malate (400 g, 0.5 mol) were added, and
the mixture was stirred at 40 °C for 1 h until crystallization
was established. The remainder of the malic acid solution was
added over 2 h followed by a wash of water (4 L). The mixture
was kept at 40 °C for another 2 h, cooled to 20 °C over 4 h,
and stirred at that temperature for 1 h. The white crystalline
solid was collected by filtration, washed with ethyl acetate (2
× 40 L), and dried in Vacuo at 45 °C. The yield of AZD3409
malate (mean of four batches) was 42.6 kg (89%) (total organic
impurities 2.4% w/w by HPLC). Mp 148 °C (DSC analysis).
1H NMR (500 MHz, 300 K, DMSO-d6) δ 9.05 (1H, dd, J
) 2.4, 0.8 Hz), 8.86 (1H, dd, J ) 4.8, 1.6 Hz), 8.60 (1H, d, J
) 7.5 Hz), 8.26 (1H, ddd, J ) 8.0, 2.4, 1.8 Hz), 7.61 (1H, ddd,
J ) 8.0, 4.8, 0.9 Hz), 7.21 (2H, dd, J ) 8.7, 5.6 Hz), 7.06 (2H,
t, J ) 8.9 Hz), 6.97 (1H, d, J ) 8.1 Hz), 6.62 (2H, m), 5.79
(1H, br s), 4.92 (1H, m), 4.50 (1H, m), 4.15 (1H, m), 4.05 (1H,
dd, J ) 7.7, 5.6 Hz), 3.64 (3H, m), 3.25 (2H, m), 3.09 (1H, dd,
J ) 11.8, 6.7 Hz), 2.78 (4H, m), 2.59 (4H, m), 2.37 (1H, dd,
J ) 15.6, 5.8 Hz), 2.04 (3H, s), 2.00 (2H, m), 1.19 (6H, m);
13C NMR (125 MHz, DMSO-d6) δ 190.1, 175.8, 172.0, 171.4,
Vol. 12, No. 2, 2008 / Organic Process Research & Development
•
211

concentrated under reduced pressure to about 22 mL. The
residual solution was diluted with isohexane (28 mL). Attempts
to crystallize the product failed, and thus the mixture was
evaporated under reduced pressure to give 22a (9 g, 82%) as
an amorphous solid.
4.49–4.56 (1H, m), 4.10–4.19 (1H, m), 3.59–3.73 (2H, m),
3.27–3.35 (2H, m), 3.07–3.15 (1H, m), 2.69–2.86 (4H, m),
2.57–2.65 (3H, m), 1.93–2.01 (2H, m), 1.65–1.76 (1H, m), 1.23
(9H, s), 1.20 (3H, d, J ) 6.30 Hz), 1.17 (3H, d, J ) 6.14 Hz).
(c). RemoVal of Boc-Protecting Group. Phosphoric acid
(36.0 g, 310 mmol) was added over 30 min to a solution of
Boc-protected thioester 22a (9 g, 10.2 mmol) in 2-methyltet-
rahydrofuran (18 mL) whilst keeping the temperature below
36 °C. The mixture was stirred at 36 °C for 1 h 20 min (HPLC
analysis showed complete conversion) and then cooled to 10
°C and diluted with water (30 mL). The mixture was neutralized
by addition of potassium bicarbonate (31.5 g, 315 mmol)
dissolved in water (250 mL), and the gum that deposited was
extracted into ethyl acetate (300 mL). The organic solution was
dried over magnesium sulfate and evaporated under reduced
pressure to give 22b (6.6 g, 81%) as a solid that was dried in
Vacuo at 40 °C.
Acknowledgment
The idea for the new route was conceived in discussions
between Euan Arnott, Anne Butlin, and Andy Godfrey. Initial
evaluation of the route was carried out by Martin Bowden and
co-workers at Syngenta (Huddersfield). Anne O’Kearney-
McMullan carried out the first successful preparation of the
crystalline malate salt of AZD3409. We thank teams led by
Stuart Hadley and Anna Powell for providing analytical support
and Ian Jones for providing interpretation of NMR spectra. We
also thank Ian Ashworth and Brian Cox for helpful discussions
on physicochemical aspects of the processes. We are grateful
to teams of chemists and engineers in the pilot plant for ensuring
successful scale-up of the processes.
1H NMR (600 MHz, 300 K, DMSO-d6) δ 9.03 (1H, s), 8.84
(1H, d, J ) 4.20 Hz), 8.65 (1H, d, J ) 7.27 Hz), 8.23 (1H, d,
J ) 7.76 Hz), 7.59 (1H, dd, J ) 7.60, 5.01 Hz), 7.17–7.25
(2H, m), 7.05 (2H, t, J ) 8.65 Hz), 6.96 (1H, d, J ) 8.08 Hz),
6.61–6.70 (2H, m), 6.00–8.00 (2H, br s), 4.89–4.96 (1H, m),
Received for review September 21, 2007.
OP700218J
212
•
Vol. 12, No. 2, 2008 / Organic Process Research & Development