Efficient syntheses of AZD4407 via thioether formation by nucleophilic attack of organometallic species on sulphur

Organic Process Research & Development 2005, 9, 555−569

Efficient Syntheses of AZD4407 via Thioether Formation by Nucleophilic Attack

of Organometallic Species on Sulphur

Marie-Lyne Alcaraz, Ste´phanie Atkinson, Philip Cornwall, Alison C. Foster, Duncan M. Gill, Lesley A. Humphries,^†

Philip S. Keegan, Richard Kemp, Eric Merifield,* Robert A. Nixon, Allison J. Noble, Darren O’Beirne,^‡

Zakariya M. Patel, Jacob Perkins, Paul Rowan, Paul Sadler, John T. Singleton, James Tornos, Andrew J. Watts, and

Ian A. Woodland

AstraZeneca R&D Charnwood, Process R&D, Bakewell Road, Loughborough, Leicestershire,

LE11 5RH, United Kingdom

Scheme 1. Previous route to AZD4407

Abstract:

The development of two efficient strategies for the synthesis of

AZD4407 is reported, both of which are considered suitable

for large-scale manufacture. In the first approach, 3-bro-

mothiophene is coupled with (2S)-2-methyltetrahydropyran-4-

one using Grignard chemistry. Following hydroxyl protection

and lithiation at thiophene C-2, reaction with a protected

5-mercapto-1-methyl-1,3-dihydro-indol-2-one derivative bearing

a leaving group on sulphur provides AZD4407 after acid-

catalysed deprotection and epimerisation. The second approach

starts from 2,4-dibromothiophene, which undergoes a selective

Grignard exchange reaction at C-2 followed by reaction with

similar protected mercapto-oxindole derivatives. Reprotection

of the oxindole ring, followed by a second Grignard exchange,

and reaction with (2S)-2-methyltetrahydropyran-4-one provides

AZD4407 after acid-catalysed deprotection and epimerisation.

Introduction

of hurdles needed to be overcome before further scale-up

could be considered:

AZD4407 4 is a 5-lipoxygenase inhibitor expected to have

activity in a variety of inflammatory conditions such as

asthma and chronic obstructive pulmonary disease (COPD).

To maintain supplies for development of this drug substance,

a route capable of delivering multikilogram quantities of

material in a short time frame was required, alongside

identification of the long-term manufacturing route. One

approach to the synthesis of AZD4407 has already been

disclosed,¹in which the key bond forming reaction involves

a palladium-catalysed coupling between the 2-mercap-

tothiophene 2 and 5-bromo-N-methyloxindole 3 affording

AZD4407 in moderate yield, Scheme 1. The crude product

mixture comprising a mixture of both the (2S,4R)- and

(2S,4S)-diastereomers was then epimerised with acid to

equilibrate predominantly to the thermodynamically more

stable (2S,4R)-isomer 4 (ratio 92:8), which was purified by

chromatography and crystallisation.

(1) The synthesis of thiol 2 from thienyl pyranol 1, for

which the final part of the process was performed at a

temperature in excess of 100 °C and utilised dimethyl

disulphide as reagent, thus generating a stoichiometric

amount of dimethyl sulphide as byproduct.

(2) The final bond forming reaction used a soluble

palladium catalyst at a level of 10 mol %, having cost of

goods implications and necessitating a removal regime to

ensure that the palladium burden in the drug substance was

reduced to an acceptable level.

(3) Lack of solid intermediates with the exception of

bromo-oxindole 3, providing little scope for purification

before isolation of crude AZD4407.

(4) Requirement for purification of the crude drug

substance by chromatography.

Consequently, a research program was instigated to find

alternative routes to AZD4407, and the results of those

studies are presented in this paper.

Whilst this route was successfully employed for the

synthesis of several kilograms of drug substance, a number

Thioether Formation via Pd-Catalysed Coupling. Hav-

ing sulphur on the oxindole moiety rather than on the

thiophene was considered preferable, since thiol 5 was

expected to be a much more stable solid intermediate than

2. Its synthesis was also expected to be relatively straight-

forward, avoiding the use of noxious reagents and formation

* To whom correspondence should be addressed. Fax: +44 (0)1509 645588.

E-mail: eric.merifield@astrazeneca.com.

^†Current address: GlaxoSmithKline, Gunnels Wood Road, Stevenage,

Hertfordshire, SG1 2NY, United Kingdom.

^‡Current address: Schering-Plough (Avondale) Co., Rathdrum, Co. Wicklow,

Ireland.

(1) Hutton, J.; Jones, A. D.; Lee, S. A.; Martin, D. M. G.; Meyrick, B. R.;

Patel, I.; Peardon, R. F.; Powell, L. Org. Process Res. DeV. 1997, 1, 61.

Published on Web 07/15/2005

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

555

Scheme 2. Reverse coupling strategy

Scheme 4. Synthesis of protected thienyl pyranols 15 and

16

Scheme 3. Lithio-thiophene approach to AZD4407

metallic species derived from readily available 3-bro-

mothiophene 9 and (2S)-2-methyltetrahydropyran-4-one 12.

Several syntheses of 12 and precursors have been published.³

In the work reported previously,¹this step was carried

out via lithium-halogen exchange followed by transmeta-

lation with chlorotitanium triisopropoxide, which generated

an intermediate titanate ester 10, Scheme 4. The titanate has

a reduced basicity, leading to cleaner carbonyl addition and

therefore has a lower propensity to generate a 2-metal-

lothiophene than the organolithium precursor. Reaction of

this species with pyranone 12 provided a mixture of thienyl

pyranols 13 and 14 with good regiocontrol (less than 1% of

the 2-isomer 17 produced, Figure 1). As this process involved

large jumps in temperature, it was further developed in order

to achieve a more constant reaction temperature profile and,

therefore, reduce the reaction time at larger scale from

unnecessary heating and cooling cycles.

A temperature of -20 °C was found to be convenient

and appropriate for the whole of this process as far as the

workup. Thus, sequential lithium-halogen exchange between

3-bromothiophene and n-butyllithium, transmetalation using

chlorotitanium triisopropoxide, and addition of pyranone 12

were all carried out at around -20 °C. After an aqueous

acidic workup and extraction, a mixture of products 13 and

14 was isolated as a toluene concentrate in 80-90% yield.

Whilst this procedure worked well on a laboratory scale

controlling the level of the undesired 2-thienyl isomer to less

than 1% GC area, scale-up to 450 L (12 kg of 12) resulted

in raised levels of the 2-isomer (at between 5 and 7% GC

of undesirable byproducts.²Initial investigations centered

around the use of a reverse coupling strategy, involving either

a palladium-catalysed or copper-mediated reaction between

oxindole thiol 5 and a bromothiophene 6, Scheme 2. Copper-

mediated couplings between 2-bromothiophene and a variety

of simple aryl thiols, e.g., p-methoxybenzene thiol, naph-

thalene-2-thiol, and 4-acetamidothiophenol as model systems

worked reasonably well. However, coupling between oxin-

dole thiol 5 and bromothiophene 6 gave an unacceptably low

yield of AZD4407 (4) under conditions in which other simple

thiols such as naphthalene-2-thiol could be coupled success-

fully with 6. We were unable to synthesise a suitable

carbonyl-protected version of 5 to determine if the oxindole

methylene unit was responsible for the lack of reactivity in

these coupling reactions.

Thioether Formation using Organolithium Chemistry.

By use of a suitable leaving group on sulphur, it was

envisaged that the coupling reaction could be accomplished

by, for example, organolithium chemistry, and the alternative

strategy shown in Scheme 3 was expected to overcome most

of the difficulties identified with the previous route (Scheme

1). This approach to AZD4407, using, for example, the

symmetrical oxindole disulphide, was proposed in the

previous paper;¹however very little work was undertaken

at that time. One of the potential problems identified with

this strategy was the acidity of the oxindole benzylic protons,

which would quench lithiothiophene. Hence appropriate

protection of the oxindole fragment was required.

(3) Atkinson, S.; Tornos, J. WO 2003051862 A1. Holt, R. A.; Rigby, S. R.;

Waterson, D. WO 9719185 A1. Crawley, G. C.; Briggs, M. T. J. Org. Chem.

1995, 60, 4264. Haslegrave, J. A.; Jones, J. B. J. Am. Chem. Soc. 1982,

104, 4666. Hetero Diels-Alder approaches to a dihydropyranone have been

reported: Unni, A. K.; Takenaka, N.; Yamamoto, H.; Rawal, V. H. J. Am.

Chem. Soc. 2005, 127, 1336. Mitsuda, M.; Hasegawa, J. GB 2304339 A1.

A related method for the synthesis of similar pyranone precursors has also

been recently reported: Reiter, M.; Ropp, S.; Gouverneur, V. Org. Lett.

2004, 6, 91.

Coupling of the Thiophene and Pyran Fragments. The

method used for formation of the bond between thiophene

C-3 and the pyran moiety involved reaction of an organo-

(2) Bird, T. G. C.; Ple, P.; Crawley, G. C.; Large, M. S. EP 623614 B1, 1994.

556

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

The Grignard exchange process reproducibly gave levels

of 2-isomers 17 (Figure 1) of <0.5% GC area on a laboratory

scale and only marginally higher on a 450 L (30 kg pyranone)

scale. It is believed that the small amount of this impurity

seen derives partly from the 2-bromothiophene present in

the starting material (limited to <0.5%). The slightly

increased level of the 2-isomer (1.09 area%) on scale-up was

not a concern, since a means of removal of impurities derived

from this contaminant further downstream in the process had

been identified (vide infra). The other key impurity carried

forward from this stage was derived from a small amount

of a dimethyl analogue of pyranone 12⁷leading to methyl

homologue 19 (Figure 1). The optical purity of pyranone 12

used was typically >99.5%; thus up to 0.5% of enantiomeric

compounds were also expected to carry through the synthesis.

An overcharge of 3-bromothiophene was deliberately used

in this process to ensure maximum consumption of the more

expensive pyranone 12, and the excess was partially removed

during distillation at the end of the stage. It was demonstrated

that any thiophene produced due to adventitious moisture in

the reaction medium would not undergo deprotonation by

isopropylmagnesium chloride, and hence the risk of forma-

tion of significant amounts of 2-metallothiophene was

eliminated. A further benefit of the Grignard exchange

process is that the need for cryogenic equipment is elimi-

nated, since the exchange step takes place at ambient

temperature or above and cooling only to between 0 and 15

°C is required for the exothermic pyranone addition. A range

of Grignard reagents have been screened in this process, and

so far, isopropylmagnesium chloride has proved the best.

Cyclohexylmagnesium chloride is an alternative that merits

further investigation as it avoids the issues surrounding

generation of propane on quenching excess isopropylmag-

nesium chloride.

Protection of the Thienyl Pyranol Moiety. The protect-

ing group chosen for pyranols 13 and 14 was trimethylsilyl

(TMS), which provided a sufficiently stable intermediate and

a robust enough protecting group for use with the subsequent

organo-lithium chemistry¹. This protecting group also influ-

ences the site of lithiation and favours formation of the

2-lithio-4-pyranyl thiophene. The procedure reported previ-

ously¹was redeveloped, and imidazole was identified as a

cheaper alternative to DMAP. Thus, reaction between

trimethylsilyl chloride and imidazole followed by addition

of the toluene solution of pyranols 13 and 14 provided a

mixture of 15 and 16, again isolated as a toluene concentrate,

in 97% yield on a 50 kg scale. The level of 3-bromothiophene

was further reduced to 4.4% GC area, due to codistillation

with toluene. Although the residual 3-bromothiophene was

expected to undergo lithiation in the next stage and react

with the oxindole derivative 8, we had shown that the levels

of the resultant impurities were tolerable in the anticipated

purification regime.⁸This proved to be a useful hold point

in the synthesis and preferable to the previous stage, as this

Figure 1. Some impurities.

area). The reasons for this were not proved conclusively;

however it is thought that formation of 2-lithiothiophene is

assisted by the presence of small amounts of protic material,

which quench 3-lithiothiophene to provide a source of

thiophene which can then undergo deprotonation at C-2.

Control of the level of this impurity was thought to be crucial

as it was expected to give a regioisomer of AZD4407 on

progression through the synthesis. Given the difficulties of

maintaining scrupulously anhydrous conditions and control-

ling the quality of the titanium reagent, an alternative

organometallic species and more robust synthesis of inter-

mediates 13 and 14 were sought.

Since 3-thienyl Grignard reagents are known to be

difficult to prepare using standard procedures^1,4and being

keen to avoid the use of an organolithium species in a

transmetalation process,⁵a Grignard exchange process⁶was

evaluated and found to work well in this situation. Reaction

between 3-bromothiophene and isopropylmagnesium chloride

in THF required up to 18 h at ambient temperature or around

3 h at 40 °C to achieve a maximum conversion of around

80%. The reasons behind this incomplete exchange are not

fully understood; however, it seems that towards the end of

the exchange, there is a competing reaction which consumes

isopropylmagnesium chloride, perhaps by reaction with the

byproduct, isopropyl bromide. A hazard assessment study

demonstrated that there were no safety concerns about the

all-in nature of this reaction, the adiabatic temperature rise

on mixing the reagents being only 30 °C. Reaction with

pyranone 12 was accomplished after cooling slightly, and

the product mixture comprising both 13 and 14 was isolated

as a toluene concentrate for continuation into the next stage

as these compounds exist as oils at ambient temperature. The

combined yield on a 30 kg scale was 90%, and GC purity

was 82 area% as the sum of 13 and 14, with thiophene and

3-bromothiophene at 6% and 6.6% GC areas, respectively,

as the major impurities. The diastereoselectivity of this

addition was not a concern at this point in the synthesis,

since previous work had shown that the mixture of epimers

could be enriched in the desired (2S,4R)-isomer downstream

by acid-catalysed epimerisation of the 3° alcohol¹. In practice,

the ratio was found to be up to 2:1 in favour of the undesired

epimer 14, although this ratio was dependent chiefly on the

length of time in contact with the aqueous hydrochloric acid

during workup.

(4) Rieke, R. D.; Seung-Hoi, K.; Xiaoming, W. J. Org. Chem. 1997, 62, 6921.

(5) Jayasuriya, N.; Kagan, J. Heterocycles 1986, 24, 2261.

(6) Boymond, L.; Rottlander, M.; Cahiez, G.; Knochel, P. WO 9951609 A1.

Abarbri, M.; Dehmel, F.; Knochel, P. Tetrahedron Lett. 1999, 40, 7449.

Tre´court, F.; Breton, G.; Bonnet, V.; Mongin, F.; Marsais, F.; Que´guiner,

G.; Tetrahedron Lett. 1999, 40, 4339. Tre´court, F.; Breton, G.; Bonnet, V.;

Mongin, F.; Marsais, F.; Que´guiner, G. Tetrahedron 2000, 56, 1349. Martin,

G. J.; Mechin, B.; Leroux, Y.; Paulmier, C.; Meunier, J. C. J. Organomet.

Chem. 1974, 67, 327.

(7) The synthesis of pyranone 12 by an intramolecular Prins cyclisation between

3-buten-1-ol and acetaldehyde is reversible and can go back either to the

same starting materials or to formaldehyde and 4-penten-2-ol. The latter

will undergo a Prins reaction with acetaldehyde to generate 2,6-dimeth-

ylpyran-4-ol.

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

557

Scheme 5. Synthesis and protection of oxindole disulphide

quantitative formation of the bis-O-silyl enol ether 24 and

liberation of triethylammonium triflate, which separated out

as a lower liquid phase. After removal of this layer, the

amount of triethylammonium triflate remaining in the product

solution was found to be variable and often too high, since

it quenched lithiothiophene in the next stage. Therefore,

further processing steps were introduced to control the level

to an acceptable limit. It was found that addition of

ethyldiisopropylamine achieved further removal of residual

triflate by crystallisation of the salt of this base, affording

residual levels reproducibly below 2 mol % after two

treatments. Consequently, an aqueous workup was avoided

which was important due to the instability of 24 at elevated

temperatures in the presence of moisture. The product was

isolated as a toluene concentrate in a yield of 99% by assay

on a 35 kg scale. This intermediate could be stored for several

days at 10-20 °C without any detrimental effect, although

our preference was for immediate utilisation in the subse-

quent stage. Due to the extremely high acid sensitivity of

this material, HPLC analysis could not be performed. Instead,

product assays were carried out on samples treated with acid

to reconvert to disulphide 23, in conjunction with ¹H NMR

spectroscopy to demonstrate that complete TBDMS protec-

tion had been accomplished in the reaction. Attempts to

accomplish protection of 23 using ethyldiisopropylamine

alone as base gave much slower conversion which was

attributed to the thick heterogeneous nature of the reaction

mixture.

Due to the relatively high cost of tert-butyldimethylsilyl

protecting agents, particularly the triflate, many alternative

protecting groups were considered and investigated. These

included O-alkyl enol ethers;¹⁰however these were dis-

counted due to the requirement of reagents such as trialkyl-

oxonium tetrafluoroborates for their synthesis. Trimethylsilyl

protection and formation of the bis-pivaloyl enol ether¹¹were

possible; however, both were found to be incompatible with

the subsequent lithiothiophene chemistry.

Coupling the Thienyl Pyranol and Oxindole Frag-

ments (Scheme 6). TMS-protected thienyl pyranol 1 is

known to react with butyllithium selectively to give the

2-lithio-4-pyranylthiophene, presumably due to the bulk of

the trimethylsilyl group inhibiting deprotonation at the

adjacent carbon¹. To avoid having to cope with butane

evolution, the base for this stage was changed to hexyllithium

in hexane, which made no difference to the chemistry.¹²The

starting material was transferred from the previous stage as

a toluene concentrate and required dilution with THF before

deprotonation would take place. The charge of hexyllithium

was regarded as critical, since an undercharge would result

in an underconsumption of 1, whereas an overcharge would

result in a competing reaction of hexyllithium with 24 or a

second deprotonation of the product 25 at C-5, followed by

intramolecular silyl abstraction to generate a C-silyl analogue

solution was stable for at least 3 months at ambient

temperature.

Synthesis and Protection of Oxindole Disulphide. The

synthesis of disulphide 23 via chlorosulphonylation of

N-methyloxindole 21 followed by reduction with hydriodic

acid (Scheme 5)²was chosen for development and scale-

up. This provided an efficient preparation of 23 and was

successfully used for the manufacture of over 200 kg of this

intermediate.

Initial studies into the reaction of disulphide 23 with

lithiothiophene 7, Scheme 3, demonstrated that protection

of the oxindole carbonyl functionality was critical for two

reasons. First, the lithiothiophene would be quenched by

deprotonation of the relatively acidic oxindole benzylic

protons in preference to the desired reaction at the disulfide

bridge. Second, the solubility of oxindole disulphide in

solvents commonly used for organometallic chemistry was

very low. It was found that both limitations could be

overcome by formation of a bis(tert-butyldimethylsilyl)enol

ether.⁹A screen encompassing a wide range of organic and

inorganic bases failed to find conditions under which tert-

butyldimethylsilyl (TBDMS) chloride could be utilised.

However, we found that a clean reaction occurred when

TBDMS triflate was used in conjunction with an organic

base, preferably triethylamine. This resulted in a rapid,

(8) Wiped-film evaporation (WFE) was briefly investigated as a technique for

the purification of the mixture of TMS ethers 15 and 16. Results on a small

scale demonstrated the feasibility of this technique since the purified mixture

of 15 and 16 with an assay of around 100% by HPLC was produced Via a

2-stage process. The first stage of this process distilled off residual

3-bromothiophene with the bulk of the toluene at 55 °C/1-3 mbara. In the

second stage, the product itself was distilled at 110 °C/1 mbara. This process

resulted in the 3-bromothiophene level being reduced from as much as 7%

w/w to 0.5% w/w. However, this additional level of purification was not

found to be necessary.

(10) Esses-Reiter, K.; Reiter, J. J. Heterocycl. Chem. 2000, 37, 927. Deberly,

A.; Bourdais, J. J. Heterocycl. Chem. 1977, 14, 781.

(11) Yamada, S.; Yaguchi, S.; Matsuda, K. Tetrahedron Lett. 2002, 43, 647.

(12) The cost of hexyllithium at the time this work was done was approximately

the same as n-butyllithium on a w/w basis; therefore in terms of molar

amounts, hexyllithium was only slightly more expensive.

(9) Sawada, T.; Fuerst, D. E.; Wood, J. L. Tetrahedron Lett. 2003, 44, 4919.

Moody, C. J.; Miah, S.; Slawin, A. M. Z.; Mansfield, D. J.; Richards, I. C.

Tetrahedron 1998, 54, 9689.

558

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

Scheme 6. Synthesis of AZD4407 from 1 and 24

Scheme 7. Impurity formation from 3-bromothiophene

Epimerisation and Isolation. Deprotection and epimeri-

sation of AZD4407 was carried out in a two-phase system

comprising dilute aqueous sulphuric acid with an acetonitrile/

toluene solvent mixture. Acetonitrile was initially chosen to

ensure that sufficient water was present in the reaction

medium to avoid dehydration and toluene was present due

to its use as the solvent for isolation of the substrate 25.

This solvent system was favoured over ethyl acetate reported

previously¹due to hydrolysis of the latter during the course

of the reaction. At ambient temperature, the acid treatment

rapidly removed both protecting groups and equilibrated the

epimers of AZD4407 at the 3° alcohol to a 92:8 mixture of

4 and 26. When the epimerisation was carried out at 30 °C

it was found that the ratio of diastereoisomers remained

approximately the same. However, the 2,5-regioisomer of

AZD4407 derived from progression of 2-thienyl pyranol 17

(Figure 1) through the synthesis selectively underwent

irreversible dehydration, enabling easier removal during

crystallisation. Hence, the temperature of 30 °C that was

chosen as the rate of dehydration of AZD4407 was negli-

gible, whereas the 2,5-regioisomer dehydrated readily over

the course of the reaction period. At elevated temperatures

in the presence of acid, AZD4407 dehydrates irreversibly;

therefore, it was important to ensure complete removal of

acid prior to any distillation.

Whilst early development batches of AZD4407 drug

substance were purified by chromatography, isolation by

crystallisation was vital for large-scale manufacture. Toluene

proved a suitable solvent for crystallisation of AZD4407,

since it not only gave a good recovery of product with

acceptable quality and good physical form but also was used

as solvent in preceding stages and facilitated removal of

acetonitrile by distillation. It was found, however, that

seeding was essential, otherwise the product failed to

crystallise, and this was done at an appropriate point in the

cooling ramp. On scale-up of this process to a 20 kg scale,

the solid filtered in less than 30 min on a 0.3 m²14 µm

filter-dryer and the isolated yield from disulphide 24 was

58%.

of AZD4407. The hexyllithium charge required was highly

sensitive to the quality of input TMS ether. Consequently,

the requirement was determined from the GC assay data for

the particular batch of 1 used, taking into account the amount

3-bromothiophene present and residual unreacted thienyl

pyranols 13 and 14. The temperature at which deprotonation

was carried out was found to be critical; below -20 °C, the

regioselectivity of lithiation is at least 98:2, but above this

limit, selectivity is lost. At -40 °C or below deprotonation

still takes place at titration rate, but no advantage is gained

from this additional cooling. It proved impossible to consume

all the starting material 1 in this reaction, since after about

0.85 molar equiv of TBDMS-protected disulphide had been

charged, no further conversion took place and excess

disulphide was carried through after acid treatment. The

reasons for this incomplete conversion remain unknown.

Studies into the isolation of AZD4407 had indicated that the

presence of excess disulphide during crystallisation led to

an inefficient recovery of AZD4407. Therefore, the process

was fixed with a stoichiometry of 0.85 equiv of TBDMS-

protected disulphide 24 until means of improving the

conversion in this stage could be found. During workup,

oxindole thiolate partitioned cleanly into the aqueous layer,¹³

with all the protected AZD4407 25 remaining in the organic

phase.

Toluene was also selected as the solvent for recrystalli-

sation since longer-term this would have facilitated progres-

sion of the damp crude product.¹⁴It also provided a solid

with the desired polymorphic form, suitable filtration char-

acteristics, and very good quality. Seeding was used during

recrystallisation to ensure that the correct polymorphic form

was produced, and meta-stable zone width experiments

defined the optimal temperature window at which to seed.

3-Bromothiophene brought though the synthesis to this

point was shown to undergo lithium-halogen exchange and

react with TBDMS-protected disulphide 24. Rearrangement

of 3-lithiothiophene to the more stable 2-lithio isomer was

expected to result in a mixture of the regioisomeric oxindole

thiophene sulphides 28, Scheme 7. By tracking the fate of

these impurities during crystallisation of AZD4407, it was

determined that a level of residual 3-bromothiophene of up

to 5% carried into this stage was tolerable.

(14) One issue concerning the use of toluene in the final stage of the process is

the level of benzene in the final product, and until we had more information

on the amount in the dried product, a grade of low-benzene toluene was

chosen for this stage. A change in the risk phrases for toluene published

recently may have necessitated a change of solvent for this recrystallisation.

This change was made as a result of the 29th ATP to EU Directive 67/

548/EEC.

(13) If this route had been selected for longer-term supply of AZD4407, recovery

and recycling of the thiolate from the aqueous phase by oxidation to

disulphide 23 would have been considered.

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

559

Scheme 8. Synthesis of AZD4407 from oxindoles having

sacrificial leaving groups on sulphur

Figure 2. Some impurities in AZD4407.

This afforded AZD4407 pure in 88% recovery on a 22 kg

scale with a residual toluene level of below 0.1% w/w after

drying.

Impurities. A consequence of telescoping the whole route

from 3-bromothiophene and pyranone 12 through to crude

AZD4407 meant that particular attention had to be paid to

the carry through of impurities. The key impurities found to

cocrystallize with AZD4407 and therefore not reduce in level

were the dimethyl analogue 29, the enantiomer [(2R,4S)-

isomer], and 2,5-regioisomer 30, for which a method of

removal had been discovered (see above)(Figure 2). No

enhancement in optical purity was achieved during crystal-

lisation of AZD4407 so the optical purity mirrored the optical

purity of the input pyranone 12. Likewise, the dimethyl

impurity 29 was present at a level of 0.5%, unchanged from

the level in the input pyranone. The (2S,4S)-diastereoisomer

26 was reduced to below 1 area%, and disulphide 23, to

below 0.1 area% when the level in the input crude product

solution was below 1 area% relative to AZD4407.

Scheme 9. Synthesis of a mixed oxindole trityl disulphide

Scheme 10. Synthesis of oxindole thiocyanate 32

Further Improvement Using Sacrificial Leaving Groups.

Whilst the chemistry described above utilising oxindole

disulphide was highly successful and was expected to be

suitable for further scale-up, the drive for a more efficient

process led us to consider alternative, sacrificial leaving

groups on sulphur. A number of substrates were identified

as potential replacements including a mixed disulphide from

mercaptobenzothiazole, a mixed oxindole trityl disulphide,

p-toluenethiosulphonate, thiocyanate, and a thiosulphate salt.

Several of these were selected for further evaluation, Scheme

8, and the results are presented herein.

Mixed Trityl Oxindole Disulphide. A mixed disulphide

of the form 31 was anticipated to undergo nucleophilic attack

at the least sterically crowded sulphur atom, thus forming

protected AZD4407 and liberating trityl thiolate on reaction

with lithiothiophene 7. This intermediate was synthesised

in high yield on a small laboratory scale by the reaction

between triphenylmethylsulphenyl chloride and oxindole

thiol 5, Scheme 9. Formation of the silyl enol ether and

further conversion through to AZD4407 proceeded in an

unoptimised solution yield of 34% after acid treatment.

Oxindole Thiocyanate. Use of thiocyanates in the forma-

tion of thioethers by reaction with either Grignard reagents

or organo-lithium species is known,¹⁵and this strategy was

investigated for the synthesis of AZD4407. Direct electro-

philic thiocyanation of N-methyloxindole using well-

established conditions such as N-bromosuccinimide with

sodium thiocyanate in acetic acid¹⁶proved very difficult;

therefore the synthesis given in Scheme 10 was used for the

preparation of thiocyanate 32.¹⁷Transformation of thiols to

thiosulphate (Bunte) salts using pyridine-sulphur trioxide

complex is precedented,¹⁸as is conversion of an organic

(16) Toste, F. D.; De Stefano, V.; Still, I. W. J. Synth. Commum. 1995, 25, 1277.

(17) A recently reported procedure describes direct electrophilic thiocyanation

of a range of aromatic and heteroaromatic compounds including oxindoles

using ammonium thiocyanate and iodine: Yadav, J. S.; Reddy, B. V. S.;

Shubashree, S.; Sadashiv, K. Tetrahedron Lett. 2004, 45, 2951.

(18) Kice, J. L.; Anderson, J. M.; Pawlowski, N. E. J. Am. Chem. Soc. 1966,

88, 5245.

(15) Pakulski, Z.; Pieroz˘yn´ski, D.; Zamojski, A. Tetrahedron 1994, 50, 2975.

560

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

thiosulphate to a thiocyanate, although this is uncommon.¹⁹

Once its viability had been demonstrated, this synthesis was

developed using dichloromethane as solvent, since the

synthesis of oxindole sulphonyl chloride 22 in this solvent

was already established and its use allowed a telescope of

some of the later stages.

Scheme 11. Synthesis of oxindole thiosulphonate 33

It has been reported that sulphonyl chlorides can be

reduced to thiols directly using triphenylphosphine in aque-

ous dioxane.²⁰This procedure was successfully applied to

oxindole sulphonyl chloride 22 in dichloromethane in the

presence of water, providing thiol 5 in 80-90% yield.

Conversion to Bunte salt 37 was achieved using pyridine-

sulphur trioxide complex in up to 80% yield from N-

methyloxindole 21. Reaction of pyridinium thiosulphate 37

with excess potassium cyanide in a biphasic aqueous toluene

mixture gave thiocyanate 32 in 73% yield.

TBDMS-protection was carried out using TBDMS triflate

in conjunction with Hu¨nig’s base as described below for

oxindole thiosulphonate 36. Reaction with excess lithio-

thiophene 7 at around -50 °C proved best giving AZD4407

with a yield in solution of 52%, Scheme 8. When the reaction

was carried out at slightly higher temperatures (-30 to -45

°C) the yield decreased to 44% and dropped further still to

34% at -20 °C.²¹Although proven as a viable approach for

the synthesis of AZD4407, the use of oxindole thiocyanate

32 was discontinued for the following reasons: the require-

ment for a lower reaction temperature, suboptimal stoichi-

ometry, potential competing attack at carbon of the -SCN

unit, and the formation of tarlike byproducts during the

course of the reaction.

Oxindole Thiosulphonate. The synthesis of thiosulpho-

nates by reaction of thiols with a sulphonyl halide has

precedence²²and was investigated for the synthesis of

oxindole thiosulphonate 33. The results from a small number

of experiments with p-toluenesulphonyl chloride suggested

that this reagent was unsuitable, since oxindole disulphide

23 was formed as the sole product, presumably via nucleo-

philic attack of unreacted thiol on the thiosulphonate as soon

as it forms. Use of tosyl bromide²²did afford the desired

thiosulphonate; however, at present this reagent is not

commercially available and had to be synthesised. The best

yields were obtained by a slow addition of thiol 5 to a

solution of p-toluenesulphonyl bromide and pyridine in

dichloromethane at -10 °C, which minimised disulphide

formation.

approach was selected for development and scale-up, Scheme

11. For a multikilogram scale synthesis, a modified process

was developed and implemented in which a solution of iodine

in dichloromethane was added slowly to a hot solution of

disulphide and p-toluene sulphinic acid in dichloromethane,

whilst allowing solvent to distill out so as to maintain a

roughly constant reaction volume. Thiosulphonate 33 was

prepared in a yield of 71% using this method with a purity

by HPLC of 97 area% on a 2 kg scale.

Silylation of oxindole thiosulphonate 33 was achieved

using a modification of the procedure developed for the

symmetrical disulphide 23, including a reduced TBDMS

triflate charge, Scheme 8.²⁴In contrast to disulphide 23,

thiosulphonate 33 was found to undergo successful protection

using ethyldiisopropylamine alone, which provided the

triflate byproduct as a solid and would have facilitated

recovery and recycling of the expensive triflic acid. It also

had the benefits of a simpler and more effective removal of

the byproduct and provided a more stable product solution.

Concentration of the solution by distillation was not neces-

sary and in fact was not possible with this material due to

thermal instability.

Reaction between lithiothiophene 7 and protected thio-

sulphonate 36 was carried out at -30 °C, Scheme 8. A rapid

reaction followed, and on completion, the byproduct, a salt

of p-toluenesulphinic acid, was removed in the aqueous layer

during workup, leaving a solution of protected AZD4407

25 in toluene/THF/hexanes which was progressed as de-

scribed above for the route using oxindole disulphide 23.

After acid treatment, AZD4407 crude was isolated by

crystallisation from toluene in 32% overall yield from TMS-

protected thienyl pyranol 1 on a 1.6 kg scale. This proved

the viability of this approach for the large-scale manufacture

of AZD4407, and this route was preferred over that using

the symmetrical disulphide 23 due to lower cost from more

efficient utilisation of oxindole disulphide. The discovery

of an alternative strategy for the synthesis of AZD4407

offering the additional benefits of a solid isolable intermediate

and later incorporation of the enantiopure pyranone 12 (see

below) meant that no further development and optimisation

of this approach was carried out.

A convenient method for the conversion of disulphides

to thiosulphonates has recently been published.²³Moreover,

a well-established synthesis of oxindole disulphide 23 was

already in place. Therefore, after proving its viability, this

(19) Tyrrell, A. W. R. Tetrahedron Lett. 1985, 26, 1753.

(20) Oae, S.; Togo, H. Bull. Chem. Soc. Jpn. 1983, 56, 3802.

(21) The reasons for this reduction in yield with increasing temperature were

not fully elucidated; however competing attack of the nucleophile at the

carbon in the thiocyanate unit was thought possible leading to a 2-cyan-

othiophene and liberation of oxindole thiolate. This in turn could react as

a nucleophile with oxindole thiocyanate forming the symmetrical disulphide

23, significant quantities of which were seen in many of the coupling

reactions tried.

(24) Protection of 33 was attempted using TBDMS chloride in the presence of

an alkoxide base but without success. Previous attempts at the coupling of

33 without protection using an excess of lithiothiophene 7 had demonstrated

that the enolate of 33 was unstable, and this was thought to be the reason

protection of the enolate under anionic conditions could not be accomplished.

(22) Ranasinghe, M. G.; Fuchs, P. L. Synth Commun. 1988, 18, 227. Prasad, J.

V. N. V. Org. Lett. 2000, 2, 1069.

(23) Fujiki, K.; Tanifuji, N.; Sasaki, Y.; Yokoyama, T. Synthesis 2002, 343.

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

561

Scheme 12. Alternative approach to AZD4407 starting

from 2,4-dibromothiophene

Scheme 14. Synthesis of AZD4407 from 38 with oxindole

protection

was more successful, giving an 80% solution yield of 38.

The problem here was that workup under basic conditions

gave an inseparable emulsion. This was resolved by adopting

an acidic workup, but the evolution of stoichiometric HCN

under these conditions precluded scale-up of this chemistry.

In contrast, TBDMS-protected thiosulphonate 36 proved

to be an ideal substrate, and reaction occurred on mixing

with the thienyl Grignard reagent at -20 °C followed by

warming to ambient temperature.²⁷After an acidic quench,

the sulphinate byproduct was washed out during the workup

using aqueous sodium acetate and thioether 38 isolated by

crystallisation from toluene²⁸65% yield.

Initial efforts into progression of intermediate 38 focused

upon a deprotonation-Grignard exchange tactic, using excess

isopropylmagnesium chloride, obviating the need for a

second oxindole protection step. However, this led to

formation of addition products between the oxindole enolate

and pyranone carbonyl group. A variety of other reagents

were therefore tried for deprotonation prior to Grignard

exchange. Using hexyllithium as base, lithium-bromine

exchange was found to occur faster than deprotonation. Other

bases (NaH, LiHMDS, KHMDS, KOMe, KOMe/ TMEDA)

all gave product mixtures containing the oxindole-pyranone

adducts.

Scheme 13. Thioether formation from

2,4-dibromothiophene

Thioether Formation on 2,4-Dibromothiophene. While

the strategy described above worked well and was success-

fully scaled up in our pilot plant to deliver a total of 25 kg

of AZD4407, an alternative approach was devised and

developed, Scheme 12, in which the pyranone was incor-

porated late in the synthesis. This strategy was appealing on

the grounds that intermediate 38 was predicted to be

crystalline, and thus represent a potential control point. It

was envisaged that 38 would derive from a selective

metalation of 2,4-dibromothiophene 39 at C-2 followed by

reaction with a suitable electrophilic sulphur derivative 40.

Although the selective lithiation of 39 with butyllithium

has been described,²⁵in our hands this reaction was not

selective, even at -78 °C; therefore we turned again to the

Grignard exchange reaction pioneered by Knochel.⁶This

procedure has been described for this substrate, although

temperatures below 0 °C were stated to be a requirement.

In contrast, we found that, at +5 °C, magnesium-bromine

exchange occurred exclusively at C-2,²⁶Scheme 13. Three

possibilities were considered for the coupling partner: the

symmetrical disulphide 23, thiocyanate 32, and p-tolyl

thiosulphonate 33. In each case the oxindole carbonyl group

was protected as the tert-butyldimethylsilyl enol ether, as

described previously. In view of the lack of reactivity of the

TBDMS-protected disulphide with Grignard reagents, this

candidate was dismissed. TBDMS-protected thiocyanate 35

Protection of the oxindole was therefore required, Scheme

14. Formation of the magnesium enolate with isopropyl-

magnesium chloride followed by quenching with chlorotri-

methylsilane (TMSCl) failed to give the TMS protected

oxindole 41. Changing the base to sodium hydride however

resulted in successful protection with TMSCl. After the

derived Grignard reagent 43 was reacted with pyranone 12,

(27) The excess ⁱPrMgCl was much less reactive towards 36 than the aryl

i

Grignard reagent. A reaction between stoichiometric PrMgCl and 36 was

(25) Crawley, G. C.; Briggs, M. T. J. Med. Chem. 1995, 38, 3951.

(26) This reaction was also applied to 2,3- and 2,5-dibromothiophene, to

determine the fate of impurities in 39. The former was totally unreactive

towards Grignard exchange, and the latter gave only the product arising

from a single Grignard exchange. An iterated process did give the 2,5-

dithioether, but this was found to be extremely sensitive towards acid-

catalysed decomposition.

only 80% complete after 24 h. The isopropyl sulphide derived from this

side reaction was not detectable in crude solutions.

(28) Several other solvents were examined for the isolation: IMS (industrial

methylated spirit), IPA, THF, acetonitrile, ethyl acetate. IMS gave an

excellent recovery, but in one experiment a different, less soluble polymorph

was produced. In view of the success and practicality of the toluene

procedure, this was not pursued further.

562

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

Scheme 15. One-pot synthesis of an AZD4407 precursor

from 2,4-dibromothiophene

diastereoisomers 4 and 26 were obtained at a combined

solution yield of 48%, still accompanied by significant

amounts of the oxindole-pyranone adducts. Substitution of

TMSCl by tert-butyldimethylsilyl chloride (TBDMS-Cl)

reduced formation of these byproducts to undetectable levels;

therefore this strategy was adopted for further development.

The most practical way of carrying out the deprotonation

was to charge a solution of 38 in THF to the sodium hydride.

The addition was performed in this manner to allow control

of the hydrogen evolution and dispersion in the large-scale

laboratory setup used. A solution of TBDMS chloride was

then charged, and the extent of formation of silyl enol ether

1

42 was monitored by H NMR spectroscopy. This process

worked well on a small scale; however, on the sole occasion

it was scaled up to 374 g, multiple sodium hydride and

TBDMS chloride charges were needed to achieve complete

conversion, and the reasons behind this difference were not

determined.

Telescoped Synthesis. Although there was a keen desire

to incorporate the crystalline intermediate 38 in the synthesis

as a point of control and hold point, it was considered feasible

that the second protection stage could be omitted by carrying

out the second Grignard exchange reaction on intermediate

42 without isolation and deprotection, Scheme 15. This

would therefore represent a one-pot synthesis of AZD4407

from 2,4-dibromothiophene, a TBDMS-protected oxindole

thiol derivative and pyranone 12. In practice, this chemistry

only worked successfully with oxindole thiocyanate 35. Thus,

an excess of 2,4-dibromothiophene 39 was treated with

isopropylmagnesium chloride at below 5 °C to form the

Grignard reagent at C-2 which was then reacted with

TBDMS-protected thiocyanate 35 to give intermediate 42

in solution. The second Grignard exchange was carried out

at 40 °C over 18 h and was followed by reaction with

pyranone 12. An aqueous work up gave 45 with a solution

yield calculated as AZD4407 of 21% based on the amount

of input pyranone 12. Although this telescoped sequence was

only done on a small laboratory scale and remains unopti-

mised, it proved the concept of coupling the three fragments

together in a one-pot process.

The second Grignard exchange using isopropylmagnesium

chloride was carried out 40 °C, and the extent of exchange

was followed by GC-MS analysis of a sample quenched

with CD₃OD. From this, it was shown that maximum

conversion to 44 occurred after heating for between 2 and

20 h and required 2.0 molar equiv of isopropylmagnesium

chloride, giving a ratio of 44:42 of around 92:8 on a

laboratory scale. Increased reaction times and reagent stoi-

chiometry did not lead to higher conversion. Conducting the

reaction at reflux while distilling off solvent in an attempt

to remove the byproduct, isopropyl bromide, did not improve

conversion, while the higher temperature led to the formation

of new byproducts. The use of a substoichiometric quantity

of isopropylmagnesium chloride as an initiator with sto-

ichiometric magnesium metal again offered no improvement

in conversion, but this did provide evidence that the reaction

could be run with catalytic isopropylmagnesium chloride.

Again, on the occasion this was run at a 374 g scale, this

step required additional reagent charges to achieve the target

conversion.

Conclusions

Two new strategies for the synthesis of AZD4407 have

been developed, which involve reaction of a lithio-thiophene

or thienyl Grignard reagent with electrophilic sulphur deriva-

tives to form the key thioether linkage. During the course

of this work, the utility of the Grignard exchange process⁶

has been demonstrated for the formation of stable 3-metal-

lothiophenes, allowing clean addition to a ketone under

noncryogenic conditions. In addition, a simple method for

synthesis and isolation of moisture and acid sensitive TBS-

protected oxindoles in solution has been developed enabling

their use on a large scale.

The first strategy, starting from 3-bromothiophene, was

successfully scaled up to pilot plant affording AZD4407 with

good chemical purity and excellent control over the formation

of regioisomeric impurities. This approach avoided the

majority of the challenges presented by the previously

published route,¹specifically no requirement for chroma-

tography at any stage of the process and no volatile thiol-

or thioether-containing waste streams to treat, and avoided

the use of palladium and other heavy metals. The number

of stages was similar, and the overall yield was significantly

The reaction mixture containing Grignard reagent 44 was

reacted with pyranone 12 at 0 °C and then quenched with

aqueous sulphuric acid to remove the protecting group, which

provided a 60:40 mixture of diastereomers 4 and 26.

Removal of THF from the solvent mixture was found to be

important, as epimerisation of the diastereomeric mixture to

give predominantly AZD4407 4 was found to be impracti-

cably slow in the presence of this solvent. Epimerisation was

carried out under conditions described previously affording

crude AZD4407 in a 58% solution yield from 38. Crystal-

lisation from a relatively dilute toluene solution gave only a

23% recovery of AZD4407 4 in 94.6% purity²⁹which was

upgraded upon recrystallisation to 97% purity, comprising

an almost identical impurity profile to material prepared via

the alternative strategy described above. Preliminary experi-

ments suggested that more concentrated solutions afforded

up to 75% recovery of AZD4407 4 without seriously

compromising product quality.

(29) By HPLC peak area. The largest impurity was (2S,4S)-AZD4407 at 3.7%,

with 38 and the desbromo analogue reduced to 0.3% and 0.5%, respectively.

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

563

improved at 45%, providing AZD4407 in good physical form

and with high purity after a single recrystallisation.

In addition, more efficient utilisation of oxindole disul-

phide was implemented by conversion to a thiosulphonate.

Although requiring an additional chemical step, the amounts

of materials required for protection of the oxindole were

consequently halved and the need for longer-term recycling

of oxindole thiol was avoided.

A second approach, starting from 2,4-dibromothiophene,

has been identified, which has the important additional

advantage of proceeding via a solid isolable intermediate.

This provided a convenient hold point in the synthesis

together with a point of control for removal of impurities

and incorporated the enantiopure pyranone 12 later in the

synthesis. Although not developed or scaled up to the same

extent, it was intended that this route would have been

subjected to further in-depth study to develop it to a point

where it would have been used for further supply of

AZD4407 had the project continued.

reaction vessel was rinsed with THF (53.1 kg), and this was

added to the quenched reaction mixture. The phases were

allowed to separate, and the upper organic layer was washed

with 6-8% sodium bicarbonate solution (303.2 kg) and the

basic lower layer (pH 9) was discarded. The organic layer

was distilled at atmospheric pressure to remove TBME

(213.4 kg, maximum temperature 61.6 °C), and then toluene

(259.0 kg) was added. A further distillation under reduced

pressure was carried out to provide a mixture of 13 and 14

as a solution in toluene. Assay by HPLC was 32% w/w,

giving a contained weight of 13 and 14 of 47.1 kg (90.4%

yield). The isomer ratio was 2:3 by GC analysis. Spectral

data are quoted for isolated, purified samples of each

diastereoisomer.

(2S,4R)-isomer 13. GC retention time ) 9.1 min. ¹H NMR

(400 MHz, CDCl₃) δ 7.31 (dd, J ) 5.0, 2.9 Hz, 1H), 7.20

(dd, J ) 3.1, 1.3 Hz, 1H), 7.13 (dd, J ) 5.0, 1.4 Hz, 1H),

4.00-3.87 (m, 3H), 2.07 (ddd, J ) 13.8, 12.3, 5.6 Hz, 1H),

1.84 (dt, J ) 13.8, 2.4 Hz, 1H), 1.77-1.69 (m, 2H), 1.21

(d, J ) 6.4 Hz, 3H). ¹³C NMR (75 MHz, CDCl₃) δ 150.64

(s), 126.25 (d), 125.11 (d), 119.26 (d), 70.17 (s), 68.92 (d),

63.48 (t), 46.23 (t), 38.34 (t), 21.69 (q). GC-MS: m/z 198

(M⁺).

(2S,4S)-isomer 14. GC retention time ) 9.0 min. ¹H NMR

(400 MHz, CDCl₃) δ 7.35 (dd, J ) 5.1, 2.8 Hz, 1H), 7.23

(dd, J ) 2.8, 1.5 Hz, 1H), 7.17 (dd, J ) 5.0, 1.4 Hz, 1H),

3.95 (ddd, J ) 11.8, 4.9, 2.1 Hz, 1H), 3.47-3.36 (m, 2H),

2.31 (dt, J ) 13.2, 2.3 Hz, 1H), 2.25 (dq, J ) 13.3, 2.3 Hz,

1H), 1.98 (td, J ) 12.9, 4.9 Hz, 1H), 1.68 (dd, J ) 13.1,

11.3 Hz, 1H), 1.21 (d, J ) 6.2 Hz, 3H). ¹³C NMR (75 MHz,

CDCl₃) δ 145.90 (s), 126.52 (d), 126.24 (d), 121.97 (d), 71.12

(d), 69.89 (s), 65.08 (t), 46.74 (t), 39.07 (t), 21.76 (q). GC-

MS m/z 198 (M⁺)

(2S,4R)- and (2S,4S)-2-Methyl-4-(trimethylsilyloxy)-4-

thiophen-3-yltetrahydropyrans (15 and 16). Imidazole

(21.2 kg, 1.3 mol equiv) was dissolved in THF (150.0 kg),

and the solution heated to 23 °C. A solution of trimethylsilyl

chloride (31.5 kg, 1.2 mol equiv) in toluene (123.3 kg) was

added to the imidazole solution below 25 °C and washed in

with toluene (4.5 kg). The solution of 13 and 14 in toluene

prepared above (147.3 kg, 32% w/w, 1 mol equiv) was added

at such a rate to maintain the temperature between 20 and

30 °C and was washed in with toluene (5.3 kg). The reaction

mixture was stirred at 35 °C for 4 h when less than 1%

starting material remained by GC analysis. After cooling to

23 °C, the reaction mixture was washed with water (2 × 90

kg), and the organic layer distilled under reduced pressure

to provide a mixture of 15 and 16 as a solution in toluene.

Assay by HPLC was 52.9% w/w, giving a contained weight

of 15 and 16 of 62.3 kg (97% yield). Purity by GC (area%):

sum of isomers 15 and 16 89.0, 3-bromothiophene 4.37,

2-isomers 0.72.

Experimental Details

General. All reactions were performed under a nitrogen

atmosphere unless otherwise stated. NMR spectra were

recorded using a Varian Unity-Inova 300 spectrometer

(operating at 300 MHz for proton and 75 MHz for 13-carbon

detection) or a Varian Unity-Inova 400 spectrometer (operat-

ing at 400 MHz for proton and 100 MHz for 13-carbon

detection) at a probe temperature of 25 °C. Chemical shifts

are reported in ppm downfield relative to TMS as an internal

standard in CDCl₃or DMSO-d₆. GC-MS data were obtained

on an HP6890 Series GC and HP5973 Mass Selective

Detector (MSD) using electron impact ionisation at 70 eV.

Assays by HPLC were performed against purified and fully

characterised reference materials. Samples of TBDMS-

protected oxindoles for analysis were collected in plastic

containers, since decomposition was catalysed by certain

types of glass vials. Full analytical methods are provided in

the Supporting Information.

(2S,4R)- and (2S,4S)-2-Methyl-4-thiophen-3-yltetrahy-

dropyran-4-ols (13 and 14). THF (106.0 kg) was added to

a solution of isopropylmagnesium chloride in THF (20%

w/w, 187.0 kg, 1.4 mol equiv) and the temperature was

adjusted to 20 °C. 3-Bromothiophene (51.5 kg, 1.2 mol

equiv) was added to the Grignard reagent, maintaining the

temperature below 25 °C during the addition. THF (5.5 kg)

was used to wash in any remaining 3-bromothiophene. The

reaction mixture was stirred at 40 °C for 3 h, and then a

quenched sample was analysed by GC to check the extent

of Grignard exchange, showing a ratio of thiophene/3-

bromothiophene of 4.6:1. The reaction mixture was cooled

to 3 °C, and 12³(30.0 kg, 1 mol equiv) was added,

maintaining the temperature between 0 and 15 °C (note that

the vessel jacket was cooled to -15 °C during this addition

to control the exotherm). This was followed by THF (4.9

kg) as a line rinse, and the reaction mixture was held at 10

°C for 40 min. The reaction was quenched by addition to a

mixture of tert-butylmethyl ether (222.3 kg) and 2 M

hydrochloric acid (307.5 kg). The temperature was main-

tained below 20 °C by controlling the rate of addition. The

Bis[2-tert-butyldimethylsilyloxy)-1-methyl-1H-indol-5-

yl]disulphide (24). Triethylamine (23.3 kg, 2.3 mol equiv)

was added to a slurry of 23²(34.8 kg) in toluene (241.5 kg)

and washed in with toluene (16.0 kg). The suspension was

cooled to 0 °C, and TBDMS-triflate (57.8 kg, 2.2 mol equiv)

was added at such a rate to maintain the temperature below

564

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

10 °C. This was followed by a toluene rinse (15.5 kg), and

then the reaction mixture stirred at 0-10 °C for 15 min.

The reaction mixture was warmed to 19 °C and analysed by

¹H NMR to ensure complete reaction. The biphasic mixture

was allowed to separate, and the lower triethylammonium

triflate/toluene layer separated off. Diisopropylethylamine

(12.9 kg) and toluene (15.3 kg) were added to the upper layer,

and the reaction mixture concentrated under reduced pressure

(maximum temperature 32.3 °C) and then cooled to 20 °C.

The mixture was filtered to remove diisopropylethylammo-

nium triflate. Then a second diisopropylethylamine charge

(6.3 kg) in toluene (142.2 kg) was added, and the distillation/

filtration process was repeated to provide 24 as a solution

in toluene, 140.3 kg. Assay by HPLC³⁰was 40.2% w/w,

giving a contained weight of 24 of 56.4 kg (99% yield). ¹H

NMR (400 MHz, CDCl₃): δ 7.27 (d, J ) 1.7 Hz, 2H), 6.74

(d, J ) 8.3 Hz, 2H), 5.19 (s, 2H), 3.20 (s, 6H), 0.73 (s, 18H),

0.00 (s, 12H). The remaining resonances were coincident

with the toluene signals at 7.1-7.3 ppm.

confirmed steady growth in the 93-250 µm range during

this time alongside attrition of the larger particles. The mean

particle size (chord length) was consistently around 35 µm.

After holding at this temperature overnight, the product was

filtered, washed with toluene (30.9 kg), and dried under

vacuum (jacket temp 50 °C) to provide AZD4407 crude,

22.15 kg. Yield based on HPLC assay of 95.5% w/w was

58% from 24. Purity by HPLC (area%): 95.8, largest

impurity (2S,4S)-isomer 26 at 2.54.

AZD4407 Pure (4). Toluene (159.3 kg) was added to

AZD4407 crude (22.05 kg), and the mixture was heated to

85 °C to dissolve the solid. The solution was filtered, cooled

to 59 °C, and then seeded with AZD4407 of the desired

polymorphic form (33.1 g). The crystallisation was further

cooled to 10 °C and then stirred at this temperature for 15

h. The solid was filtered off, washed with toluene (38.0 kg),

and dried under vacuum (maximum jacket temperature 50

°C) to afford AZD4407 pure, 19.5 kg (88% yield, including

the heel). HPLC purity (area%): 98.4, largest impurity

(2S,4S)-isomer 26 at 0.83.

AZD4407 Crude. THF (106.3 kg) and toluene (31.2 kg)

were added to a mixture of 15 and 16 in toluene (57.9 kg

solution weight, 30.7 kg contained weight, 1.0 mol equiv).

The resulting solution was cooled to -23.5 °C, and n-

hexyllithium in hexanes (33% w/w, 40.1 kg, 1.28 mol

equiv)³¹was added, maintaining the temperature below -20

°C, followed by a THF rinse (5.4 kg). A solution of 24 in

toluene (40.2% w/w, 56.4 kg contained weight, 0.85 mol

equiv) was added, keeping the temperature below -20 °C.

The reaction mixture was warmed to -5 °C and quenched

with aqueous sodium chloride (11.1 kg in 211 kg of water).

After warming to ambient temperature, the aqueous layer

was removed and the organic phase distilled under reduced

pressure to remove THF and hexane to leave intermediate

25 as a toluene solution. Acetonitrile (156.0 kg) and aqueous

sulphuric acid (2 M, 67.8 kg) were added to the toluene

concentrate, and the reaction mixture stirred at 30 °C for 21

h and then at 20 °C for 44 h. The lower acidic layer was

removed, and the organic layer was washed with sodium

acetate solution (2.5 M, 68.6 kg) followed by isohexane (89.8

kg). Water (141.0 kg) and toluene (160.8 kg) were added,

and the aqueous layer was removed. The organic layer was

concentrated by distillation under reduced pressure until

197.4 kg of distillate had been removed. Further toluene

(152.7 kg) was added, and distillation continued until 156.5

kg of distillate had been removed. The solution was cooled

to 50 °C, seeded with AZD4407 (27 g) and cooled further

to 10 °C. During laboratory development work, particle size

analysis using focused beam reflectance measurements

1-Methyl-5-trityldisulfanyl-1,3-dihydro-indol-2-one (31).

A solution of 5 (3.0 g, 16.8 mmol) in dichloromethane (10

mL) was cooled to 0 °C and then was slowly added to a

suspension of triphenylmethanesulfenyl chloride (5.2 g, 16.7

mmol) and pyridine (1.32 g, 16.7 mmol) in dichloromethane

at 0 °C. The mixture was allowed to warm to ambient

temperature, stirred for ca. 20 h, and then washed with water

(2 × 30 mL). The organic phase was dried over magnesium

sulphate, and the dichloromethane was removed to leave 31

1

as a solvent-damp, pale yellow solid, 8.40 g. By H NMR

analysis this contained 14% w/w dichloromethane; therefore

contained product weight was 7.22 g (95% yield). ¹H NMR

(400 MHz, CDCl₃): δ 7.31-7.14 (m, 15H), 6.98 (dd, J )

8.1, 1.2 Hz, 1H), 6.78 (d, J ) 1.2 Hz, 1H), 6.53 (d, J ) 8.2

Hz, 1H), 3.36 (s, 2H), 3.17 (s, 3H).

2-(tert-Butyldimethylsilyloxy)-1-methyl-5-trityldisul-

fanyl-1H-indole (34). TBDMS-protection of 31 was carried

out on a 2 g scale using a similar process to that given above

for the symmetrical disulphide 24. Product 34 was isolated

as a toluene solution and was progressed without further

treatment. ¹H NMR (400 MHz, CDCl₃): δ 7.02 (d, J ) 1.3

Hz, 1H), 6.89 (d, J ) 8.5 Hz, 1H), 6.83 (dd, J ) 8.2, 1.8

Hz, 1H), 5.39 (s, 1H), 3.54 (s, 3H), 1.04 (s, 9H), 0.32 (s,

6H). The remaining protons were coincident with toluene

resonances at 7.1-7.3 ppm.

Synthesis of AZD4407 (4) Using 34. The solution of 34

prepared above (assumed 90% yield, 3.96 mmol) was reacted

with the lithiothiophene 7 (1.07 g, 3.96 mmol) at -40 °C as

described above for the symmetrical disulphide 24. After

workup and acid-catalysed epimerisation/deprotection, the

yield of AZD4407 in solution by HPLC assay was 0.50 g

(34% from 34).

Pyridinium S-(1-Methyl-2-oxo-2,3-dihydro-1H-indol-

5-yl) Thiosulphate (37). Chlorosulfonic acid (272 mL, 4.09

mol) was heated to 70 °C. A solution of 21 (80 g, 0.54 mol)

in dichloromethane (240 mL, 3 vol) was added to the

chlorosulfonic acid over 45 min, whilst allowing the dichlo-

romethane to distil out. Dichloromethane (40 mL, 0.5 vol)

(30) Assays were performed on samples of material treated with acid to hydrolyze

back to disulphide 23, since TBDMS-protected disulphide 24 was unstable

to HPLC. ¹H NMR was carried out before acid treatment to demonstrate

that the TBDMS-protection had gone to completion.

(31) A multiplication factor for the hexyllithium charge to account for reagent

consumed by reaction with impurities was determined as follows. Area

percentages of impurities undergoing lithiation (thiophene, 3-bromothiophene,

pyranone 12, and residual thienyl pyranols 13 and 14) were converted to

mole percentages by multiplying each by molecular weight ratios relative

to TMS thienyl pyranols 1 (3.07, 1.61, 2.35, 1.36, respectively) and their

relative response factors (1.0, 1.7, 1.0, 1.0, respectively). The multiplication

factor for the hexyllithium charge was then calculated as 1 + (Σ mol %

impurities/total area% 15 and 16).

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

565

was used as a line wash. The reaction was allowed to stir at

70 °C for a further 25 min. The mixture was cooled to 20

°C and then added dropwise to a rapidly stirred mixture of

dichloromethane (1400 mL, 17.5 vol) and ice/water (1400

mL, 17.5 vol) over a period of 20 min, maintaining the

temperature below 10 °C (exothermic). The mixture was

stirred for 10 min, and then the layers were allowed to

separate and the aqueous phase was discarded. The dichlo-

romethane phase was washed twice with water (320 mL, 4

vol each) and then concentrated by distillation removing 400

mL (5 vol) of solvent to yield 22 as a solution in dichlo-

8.09 (m, 2H), 7.39-7.45 (m, 2H), 6.93 (d, J ) 7.7 Hz, 1H),

3.55 (s, 2H), 3.11 (s, 3H). HPLC purity: 98.5 area%.

1-Methyl-5-thiocyanato-1,3-dihydro-indol-2-one (32).

Bunte salt 37 (50.3 g, 59.6% w/w, 30 g contained weight,

0.089 mol) was slurried in water (150 mL). Aqueous sodium

carbonate was added slowly until pH 10 was achieved (20%

w/w, 150 mL used), causing effervescence and producing a

yellow solution. Toluene (450 mL) was then added to the

vessel,³³and the mixture vigorously stirred. A solution of

potassium cyanide (46.1 g, 0.71 mol) in water (100 mL) was

then added to the reaction vessel. After stirring for 60 min,

the biphasic mixture was filtered. The two layers were

separated, the organic phase was washed with water (120

mL) and then two portions of 80% aqueous ethanol (120

mL each). The volume was reduced to 90 mL by distillation

at 35 °C/40-70 mbar (360 mL of toluene was removed).

The resulting slurry was heated to 95 °C to give a solution

and then allowed to cool to 80 °C. Acetonitrile (4.5 mL,

5% v/v with respect to toluene) was then added, and the

solution was allowed to cool to ambient temperature

overnight. The resulting slurry was cooled to 0 °C, held for

3 h, and then filtered. The filter cake was washed with

isohexane (90 mL) and then dried in a vacuum oven at 40

1

romethane. Typical H NMR data for an isolated sample

(CDCl₃, 300 MHz): δ 8.03 (m, 1H), 7.90 (s, 1H), 6.97 (d,

J ) 8.3 Hz, 1H), 3.65 (s, 2H), 3.29 (s, 3H).

A solution of triphenylphosphine (427 g, 1.63 mol) in

dichloromethane (200 mL, 2.5 vol), prepared by warming

the mixture to 30 °C, was added to the solution of 22

prepared above over a period of 25 min (exothermic

addition), followed by a dichloromethane line wash (40 mL,

0.5 vol). Water (80 mL, 1 vol) was then added to the reaction

mixture, and after heating under reflux for 30 min no starting

material remained, so the mixture was allowed to cool to

ambient temperature and held overnight. Water (600 mL,

7.5 vol) was added followed by dropwise addition of aqueous

sodium hydroxide until pH 12 was achieved (10 M, 90 mL

used, 1.13 vol). The layers (both orange in colour) were

allowed to separate, the dichloromethane phase was dis-

carded, and the aqueous phase was washed with one portion

of dichloromethane (400 mL, 5 vol). Dichloromethane (1200

mL, 15 vol) was then added to the aqueous phase followed

by HCl until pH 2 was achieved (11.65 M, 50 mL used,

0.63 vol). The mixture was stirred for 5 min before allowing

the layers to separate. The organic phase was concentrated

by distillation removing 400 mL (5 vol) of solvent to yield

1

°C to provide 37 as a white solid 13.25 g (73%). H NMR

(CDCl₃, 400 MHz): δ 7.46-7.55 (m, 2H), 6.86 (d, J ) 8.2

Hz, 1H), 3.57 (s, 2H), 3.23 (s, 3H). HPLC purity (area%):

98.8, disulphide 23, 1.2.

2-(tert-Butyldimethylsilanyloxy)-1-methyl-5-thiocyanato-

1H-indole (35). This was prepared using the method

described below for thiosulphonate 33. 35 was isolated as a

solution in toluene in an assumed 100% yield. ¹H NMR (400

MHz, CDCl₃): δ 7.58-7.43 (d, J ) 1.8 Hz, 1H), 5.54 (s,

1H), 3.51 (s, 3H), 1.02 (s, 9H), 0.31 (s, 6H). The remaining

resonances were coincident with toluene at 7.1-7.3 ppm.

Synthesis of AZD4407 Using 35. This was carried out

on a 12 g scale of 15 and 16 using a similar process to that

described above for disulphide 24, with a relative stoichi-

ometry of 1:0.83 15 and 16/35, except that the reaction was

performed at between -50 and -54 °C. After epimerisation,

the yield of AZD4407 in solution by HPLC assay was 52%

from 35. Crystallisation from toluene afforded 4.0 g (64%

recovery) of crude AZD4407. Recrystallisation from toluene

gave 3.0 g (75%) of purified AZD4407 (22% overall yield

from 35). Purity by HPLC (area%): 97.3, largest impurity

was the (2S,4S)-diastereoisomer 26 at 1.22.

1

5 as a solution in dichloromethane. Typical H NMR data

for an isolated sample (CDCl₃, 300 MHz): δ 7.20-7.29 (m,

2H), 6.70 (d, J ) 8.1 Hz, 1H), 3.49 (s, 2H), 3.42 (s, 1H),

3.19 (s, 3H).

Pyridine-sulfur trioxide complex (86.5 g, 0.54 mol) was

slurried in dichloromethane (400 mL, 5 vol), and the solution

of 5 in dichloromethane prepared above was added. The

mixture was heated under reflux for 2 h and then cooled to

5 °C and stirred at this temperature for 100 min. The solid

product was filtered off, and the filter cake was washed with

dichloromethane (2 × 200 mL). The resulting solid was dried

in a vacuum oven at 40 °C to yield 37 as a white solid, 155

g, assay by ¹H NMR 95% w/w, giving an overall yield from

Toluene-4-thiosulfonic Acid, S-(1-Methyl-2-oxo-2,3-

dihydro-1H-indol-5-yl) Ester (33). Disulphide 23 (2.2 kg,

6.17 mol), anhydrous sodium p-toluenesulfinate (2.9 kg, 16.3

mol), and dichloromethane (33 L) were charged to a reaction

vessel. Separately, iodine (1.74 kg, 6.86 mol) was dissolved

in dichloromethane (55 L). The suspension in the reaction

vessel was heated to reflux, and the solution of iodine in

1

N-methyloxindole (21) of 80%. H NMR (d₆-DMSO, 400

MHz): δ 8.92 (dd, J ) 6.5, 1.4 Hz, 2H), 8.58 (m, 1H), 8.02-

(32) It was found that the product was unstable in the presence of excess aqueous

sodium thiosulphate, particularly during a prolonged workup. To generate

a more robust process by minimising risk of degradation in the workup,

the amount of sodium thiosulphate solution charged was reduced by

calculating a slight excess over the amount of excess iodine input over the

1 molar equiv required for reaction. Although this process was not proven

on any scale larger than 10 g, the stability of the reaction mixture in the

presence of the wash solution over a period of 2 h indicated that no problems

with workup were expected on scale-up. This wash regime was preferred

over the use of mildly basic aqueous systems which led to emulsion

formation.

(33) The rationale behind use of this two-phase aqueous organic solvent system

was to minimise contact between the product 32 and cyanide, by extraction

into the organic phase. Degradation occurred when water alone was used

as solvent leading to formation of large amounts of oxindole disulphide by

nucleophilic displacement of thiolate from thiocyanate by cyanide, which

would then react with thiocyanate to form disulphide and liberate cyanide.

It is important to add the toluene before the potassium cyanide, since the

reaction is rapid.

566

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

dichloromethane was added to this in portions (6 L every

30 min). As the iodine solution was added, a similar volume

of dichloromethane was removed from the batch by distil-

lation so as to keep the reaction volume constant. At the

end of the addition, the mixture was stirred for 1 h at reflux.

The amount of starting material remaining was 4%, so further

iodine (157 g, 0.62 mol) was added in one portion and the

mixture was heated overnight at 38 °C. This led to comple-

tion of reaction, so the mixture was cooled to room

temperature and washed with sodium thiosulfate solution (1

M, 22 L) by stirring for 23 min.³²The phases were separated,

and the dichloromethane solution was washed again with

sodium thiosulfate solution (1 M, 24 L) and then with water

(22 L). The resulting dichloromethane solution was distilled

to leave a volume of approximately 12 L, and then toluene

(17 L) was added. The remaining dichloromethane was

removed by distillation under vacuum, and the resulting

toluene solution cooled from 50 °C to 0 °C over 6 h and

then stirred at 0 °C overnight. The product was filtered,

washed with toluene (4.4 L), and dried to constant weight

at 40 °C under vacuum to provide 33 as an off-white solid

30 min and then warmed to -5 °C over 15 min. A line wash

of toluene (3 L) was added, followed by 5% aqueous sodium

chloride solution (9.2 L), and then the mixture was warmed

to 20-25 °C over 30 min and stirred at this temperature for

30 min. The phases were separated, and the upper layer was

washed again with 5% aqueous sodium chloride solution by

stirring for 30 min at 20-25 °C. The upper layer was then

concentrated to 8.5 L by distillation under vacuum at below

50 °C, and then the contents of the reaction vessel were

cooled to 20-25 °C. Acetonitrile (14 L) and 2 M sulphuric

acid (4.5 L) were added, and the mixture stirred at 30 °C

for 24 h. Stirring was stopped, the lower aqueous layer was

removed, the upper layer was cooled to 20-25 °C and then

washed with 20% aqueous sodium acetate solution (4.7 L),

followed by isohexane (9.3 L). The lower product-containing

layer was washed with water (4.7 L). The toluene/acetonitrile

solution was concentrated to 8 L by distillation under vacuum

at below 50 °C, additional toluene (20.5 L) was added, and

the distillation was repeated to reduce the volume to 8 L.

The mixture was heated to 80 °C and then cooled to 56 °C

over 40 min and seeded with AZD4407 (5 g). The mixture

was held at 56 °C for 10 min, cooled to 10 °C over 5.5 h,

and then stirred at 10 °C overnight. The product was filtered

off, washed with toluene (3.1 L), and dried to constant weight

at 40 °C under vacuum to give AZD4407 (4) as an off-white

solid (0.76 kg, 32%). HPLC purity (area%): 94.3, (2S,4S)-

isomer 26 was 2.13.

1

(2.92 kg, 71%). H NMR (400 MHz, CDCl₃): δ 7.49 (d, J

) 8 Hz, 2H), 7.23-7.28 (m, 4H), 6.77 (d, J ) 8 Hz, 1H),

3.49 (s, 2H), 3.22 (s, 3H), 2.43 (s, 3H). HPLC purity: 96.8

area%.

Toluene-4-thiosulfonic Acid S-(2-tert-Butyldimethyl-

silyloxy)-1-methyl-1H-indol-5-yl) Ester (36). Diisopropyl-

ethylamine (0.95 kg, 7.35 mol) was added over 13 min to a

stirred suspension of 33 (2.16 kg, 6.48 mol) in toluene (13.7

L) at between -2 °C and +2 °C followed by a toluene line

wash (0.5 L). TBDMS triflate (1.85 kg, 7.00 mol) was then

added to the reaction mixture over 1 h at -5 °C to -3 °C

followed by a toluene line wash (0.5 L). The mixture was

stirred at between -3 °C and -1 °C for 30 min, warmed to

A portion of this material was recrystallised from toluene

on a 20 g scale as described above to afford AZD4407 pure,

16.6 g (83%). HPLC purity (area%): 97.6, (2S,4S)-isomer

26, 0.94, dimethyl analogue 29, 0.54.

5-(4-Bromothiophen-2-ylsulfanyl)-1-methyl-1,3-dihydro-

indol-2-one (38). 2,4-Dibromothiophene (30.89 g, 94% w/w,

29 g contained weight, 120 mmol) was dissolved in THF

(154 mL) and cooled to 0 °C. A solution of isopropylmag-

nesium chloride in THF (2 M, 66 mL, 132 mmol)³⁴was

added whilst maintaining the reaction temperature between

0 and 5 °C. The reaction was left stirring for 1 h on

completion of the addition, at which point a sample was

removed and analysed by HPLC to check conversion and

confirmed that the Grignard reagent had formed. The reaction

was cooled to -20 °C, and a solution of 36 in toluene (320

g, 16.8% w/w, 120 mmol) was added whilst maintaining the

reaction temperature between -10 and -20 °C. Once the

addition was complete the reaction was warmed to 20 °C

over 2 h. The reaction was then quenched by the addition of

aqueous sulphuric acid (2 M, 230 mL, 460 mmol). The layers

were separated, the organic layer was washed with aqueous

sodium acetate solution (20% w/v, 2 × 220 mL) followed

by water (220 mL), and then toluene (290 mL) was added.

The organic phase was concentrated by distillation under

vacuum until the concentration of 38 was around 20% w/w.

The mixture was heated to 80 °C and then cooled to ambient

temperature. The solid product was collected by filtration,

washed with toluene (29 mL), and then dried in a vacuum

1

20-25 °C over 30 min, and then sampled. Analysis by H

NMR showed complete reaction, so the suspension was

filtered. The collected solid was washed with toluene (2.1

L), and the filtrates were combined to provide 36 as a pale

brown solution in toluene, which was stored at 0 °C until

used in the subsequent step. The yield was assumed to be

100%. ¹H NMR (300 MHz, CDCl₃): δ 7.39-7.43 (m, 3H),

6.97-7.05 (m, 2H), 5.49 (s, 1H), 3.53 (s, 3H), 2.38 (s, 3H),

1.02 (s, 9H), 0.31 (s, 6H). The remaining resonances were

coincident with toluene at 7.1-7.3 ppm.

Synthesis of AZD4407 (4) from 36. n-Hexyllithium in

hexanes (2.02 kg, 32.9% w/w, 7.22 mol)³¹was added over

50 min to a stirred solution of 15 and 16 (3.2 kg, 52.3%

w/w in toluene, 6.19 mol) diluted with THF (4.25 L) and

toluene (4.25 L) at -28 °C to -25 °C. The solution was

stirred at -25 °C to -28 °C for 5 min. TMS-chloride quench

1

of a sample and analysis by H NMR showed incomplete

lithiation, so additional n-hexyllithium in hexanes (0.1 kg,

32.9% w/w, 0.36 mol) was added. The solution was stirred

for 5 min and then resampled. Analysis as above demon-

strated complete lithiation, so the reaction solution was

cooled to -33 °C. The solution of 36 (assumed 6.48 mol)

in toluene prepared above was then added over 70 min at

-33 °C to -25 °C. The mixture was stirred at -25 °C for

(34) The excess ⁱPrMgCl was much less reactive towards 36 than the aryl

i

Grignard reagent. A reaction between stoichiometric PrMgCl and 36 was

only 80% complete after 24 h. The isopropyl sulphide derived from this

side reaction was not detectable in crude solutions.

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

567

oven at 40°C to afford a beige crystalline solid, 26.6 g (65%

yield). ¹H NMR (300 MHz, CDCl₃): δ 7.34-7.37 (m, 1H),

7.26-7.28 (m, 2H), 7.10 (d, J ) 1.5 Hz, 1H), 6.76 (d, J )

8 Hz, 1H), 3.52 (s, 2H), 3.19 (s, 3H). Purity by HPLC: 98.52

area%.

0.784 kg). The mixture was heated to 30 °C with stirring,

and stirring continued at this temperature for 18 h. The layers

were allowed to separate, and the lower aqueous layer was

discarded. The organic (toluene/acetonitrile) phase was

washed with sodium acetate solution (20% w/v, 0.75 L),

isohexane (1.088 kg), and then water (1.7 kg). The product-

containing solution was diluted with toluene (2.196 kg) and

distilled under reduced pressure (200-80 mbarg) to give

1.871 kg of a solution of AZD4407 in toluene (12.7% w/w,

contained weight of AZD4407 therefore 237.4 g, 58% yield).

This was heated to 60 °C, and a cooling ramp of 10 °C/h

was set. When the temperature had fallen to 45 °C, the

mixture was seeded with AZD4407 (2.37 g, 1 mol %).

Stirring was continued at 10 °C for 72 h. The product was

collected by filtration, and the filter cake was washed with

toluene (0.375 L) and dried under vacuum at 40 °C for 4 h,

to give AZD4407 crude (54.9 g, 23%) in 95% purity by

HPLC peak area. This was slurried in toluene (0.46 L) and

heated to 85 °C and filtered at this temperature, and a cooling

ramp of 10 °C/h was set. When the temperature had fallen

to 45 °C, the mixture was seeded with AZD4407 (2.37 g, 1

mol %). Stirring was continued at 10 °C for 18 h. The

product was collected by suction filtration, and the filter cake

was washed with toluene (0.11 L) and dried under vacuum

at 50 °C for 4 h, to give AZD4407 (43.9 g, 80%) in 97.2%

purity by HPLC peak area.

Synthesis of AZD4407 (4) from 2,4-Dibromothiphene

via a One-Pot Process. A solution of isopropylmagnesium

chloride in THF (16.7 mL, 20.5% w/w, 33.5 mmol) was

added over 10 min to a solution of 2,4-dibromothiophene

(7.37 g, 30.5 mmol) in THF (26.5 mL) at 3 °C causing an

exotherm to +12 °C. The resulting reaction mixture was

stirred at this temperature for 30 min during which time a

thick light-coloured precipitate formed. A solution of 35 in

toluene (21.9 g, 31% w/w, 21.3 mmol) was added over a

period of 5 min. The reaction mixture was allowed to warm

to ambient temperature and stirred overnight. A solution of

isopropylmagnesium chloride in THF (12.1 mL, 20.5% w/w,

24.4 mmol) was added, and the reaction mixture was heated

to 40 °C. After 18 h the exchange reaction had proceeded

to >80% conversion, so the reaction mixture was cooled to

12 °C and pyranone 12 (2.78 g, 24.4 mmol) was added over

5 min. The reaction was stirred for 1 h at 10 °C and then

warmed to 20 °C. Aqueous sodium chloride (5% w/v, 100

mL) was added, and stirring continued for 20 min. Toluene

(100 mL) was added, and the mixture was allowed to stand

for 1 h, after which the layers had separated. The lower

aqueous layer was re-extracted with toluene (100 mL), and

the combined organic layers were washed with water (100

mL) to leave a dark brown solution of 45, weight 33.4 g.

Assay by HPLC of an acidified sample gave a result of 5.7%

w/w for the combined isomers 4 and 26, corresponding to a

21% yield of AZD4407 and the (2S,4S)-epimer (26) from

12.

Synthesis of AZD4407 (4) from 38. Compound 38 (0.374

kg, 1.10 mol) was dissolved in THF (2.91 kg) by heating to

50 °C with stirring. The solution was cooled to 25 °C and

then added to a suspension of sodium hydride (0.048 kg of

a 60% dispersion in mineral oil, 1.20 mol) in THF (0.355

kg) followed by a THF line wash (0.172 kg). A solution of

TBDMS chloride (0.185 kg, 1.23 mol) in THF (0.353 kg)

was added followed by a THF line wash (0.170 kg). The

mixture was stirred at 20 °C for 2 h and then analysed by

¹H NMR which showed only 50% conversion had taken

place. The mixture was then added to sodium hydride (0.026

kg of a 60% dispersion in mineral oil, 0.65 mol). The mixture

1

was stirred at 20 °C for 2 h and then analysed by H NMR

which showed only 60% conversion had taken place. A

solution of TBDMS chloride (0.101 kg, 0.67 mol) in THF

(0.179 kg) was added. The mixture was stirred for 16.5 h at

1

20 °C and then analysed by H NMR which showed only

75% conversion had taken place. The mixture was then added

to sodium hydride (0.023 kg of a 60% dispersion in mineral

oil, 0.58 mol). A solution of TBDMS chloride (0.085 kg,

0.56 mol) in THF (0.168 kg) was added. The mixture was

stirred for 1 h at 20 °C and then analysed by ¹H NMR which

showed that complete conversion to 42 had taken place.

A solution of isopropylmagnesium chloride in THF (2

M, 1.142 kg, 2.34 mol) was added to the reaction mixture,

followed by a THF line wash (0.189 kg), resulting in a 5.4

°C exotherm. The mixture was heated to 40 °C and stirred

at this temperature. After 5 h, analysis by ¹H NMR indicated

that 50% conversion to the arylmagnesium halide 44 had

taken place. The mixture was cooled to 35 °C, and a further

charge of isopropylmagnesium chloride in THF (2 M, 0.564

kg, 1.16 mol) was added. The mixture was heated to 40 °C

and stirred at this temperature for 1.5 h. Analysis by ¹H NMR

indicated that 92% conversion to the arylmagnesium halide

44 had taken place.

The mixture was cooled to 0 °C, and a solution of 12

(0.127 kg, 1.11 mol)) in THF (0.334 kg) was added, followed

by a THF line rinse (0.165 kg). The mixture was warmed to

20 °C and stirred at this temperature for 14 h and then added

to aqueous sulphuric acid (0.24 M, 7.00 kg) held at 5 °C

(exotherm of 9 °C observed) followed by a toluene line wash

(1.414 kg). The mixture was warmed to 20 °C over 30 min

with stirring. The layers were allowed to separate, and the

lower layer was discarded. The organic layer was washed

with sodium acetate solution (20% w/v, 0.78 L), followed

by water (1.523 kg), then distilled under vacuum (150 mbarg)

to remove THF. A further charge of toluene (1.426 kg) was

added once the volume had been reduced by approximately

50%, and distillation continued until 1.936 kg remained.

1

Analysis of this mixture by H NMR showed the solvent

Acknowledgment

composition to be toluene/THF 97:3 w/w.

The mixture was cooled to 20 °C, and acetonitrile (1.943

kg) was added, followed by aqueous sulphuric acid (2 M,

We thank the following who were also involved in this

project: Members of the Process Analysis Section at

AstraZeneca Charnwood, in particular Annabelle Germon,

568

•

Vol. 9, No. 5, 2005 / Organic Process Research & Development

Lee Mallard, and Kevin Sutcliffe for their input to analytical

aspects. The staff of the Avlon and Macclesfield Large Scale

Laboratories for scale-up studies of some of the processes

published in this paper. Members of the AstraZeneca Process

R&D Laboratories in Bangalore for additional investigations

into the synthesis of 23. Neil Barnwell, Steve Eyley, Melvyn

Giles, Paul Hurved, and Anil Mistry of AstraZeneca Charn-

wood for their contribution to the project. We also acknowl-

edge the contribution to this project made by Laporte plc,

subsequently part of Degussa AG, for development and

scale-up of the synthesis of 23 and Orgamol S.A. for scale-

up of the synthesis of 32.

Supporting Information Available

HPLC and GC analytical methods for AZD4407 and

various intermediates. This material is available free of

charge via the Internet at http://pubs.acs.org.

Received for review March 24, 2005.

OP0500483

Vol. 9, No. 5, 2005 / Organic Process Research & Development

•

569

Article Doi

Efficient syntheses of AZD4407 via thioether formation by nucleophilic attack of organometallic species on sulphur

DOI: 10.1021/op0500483

Source and publish data:

Authors:

Article abstract of DOI:10.1021/op0500483

Full text of DOI:10.1021/op0500483

Products guided by the article

R&D Labs maybe for 406463-50-1

Relevant to this article

Hot Product