The development of a manufacturable synthesis of LY213829

Article abstract of DOI:10.1021/op990197j

The development of a manufacturable synthesis of LY213829 (4-thiazolidinone-5-[3,5-bis(1,1-dimethylethyl)-4-hydroxyphenyl] methyl) is described. Rather than reduction to eliminate the thiocarbonyl from the rhodanine moiety, the new route utilizes a novel

Full text of DOI:10.1021/op990197j

Organic Process Research & Development 2000, 4, 295−297
The Development of a Manufacturable Synthesis of LY213829
Brian J. Slattery,† Douglas P. Kjell,*,† James D. Copp,‡ Francis O. Ginah,‡ Marvin M. Hansen,‡ and Stephen C. Stirm‡
Chemical Process DeVelopment, Lilly Research Laboratories, A DiVision of Eli Lilly & Company, P.O. 685,
Lafayette, Indiana, 47902, U.S.A., and Chemical Process Research and DeVelopment, Lilly Research Laboratories,
A DiVision of Eli Lilly & Company, Indianapolis, Indiana 46285, U.S.A.
Table 1: Results of Hydride Reduction of Rhodanine 6
Abstract:
The development of a manufacturable synthesis of LY213829
(4-thiazolidinone-5-[3,5-bis(1,1-dimethylethyl)-4-hydroxyphen-
yl] methyl) is described. Rather than reduction to eliminate the
thiocarbonyl from the rhodanine moiety, the new route utilizes
a novel concurrent ring opening with ammonia and re-
cyclization with formaldehyde. This change obviates a poten-
tially problematic zinc solid waste stream.
reagent
optimum result
NaBH4/MeOH
NaBH4/TiCl4
LiAlH4
LiAlH4/ZnBr2
Dibal-H
LiBHEt3
Zn(BH4)2
<55% 7
complex mixture
no reaction
∼50% 7
no reaction
no reaction
no reaction
20% 7, 65% 1
NaBH4/NiCl2 or CoCl2
Introduction
LY213829 (1) is a potent anti-oxidant and 5-lipoxygenase
inhibitor.1 Currently 1 is under clinical investigation for a
variety of inflammatory bowel diseases. The short and
efficient route used for manufacture of early clinical trial
materials is shown below (Scheme 1).2 Evaluation of this
route for commercial manufacture identified only one
potential liability. The condensation of 3,5-di-tert-butyl-4-
hydroxybenzaldehyde (2) and rhodanine (3) is straightfor-
ward and produced high quality adduct 4 in good yield.
While the reduction of olefin 4 via Hantzch’s ester 5 to form
6 is somewhat unusual, it presented no challenges to scale-
up. The challenge for full-scale manufacture occurred in the
last chemical transformation, in which the reduction of the
thiocarbonyl moiety to form 1 required 5 equiv of zinc. This
roughly translated to a kilo of solid Zn waste per kilo of 1.
At the time of this development a reclamation site could not
be identified, and thus the waste would need to be landfilled.
As the potential annual requirements of 1 are large, this
translates to a major environmental challenge. This study
examined an alternative method that avoids the use of Zn.3
catalysts, the spent catalyst can be sent back to the supplier
for reclaiming. However, the catalyst loading (340 mass %
of 5% Pd/C), and conditions (500 psi H2, 120 °C) were
unsuitable for manufacturing. We explored palladium, plati-
num, ruthenium, and rhodium catalysts on normal carbon
and sulfided carbon supports at 100 mass % loadings of 5%
metal catalysts. All catalysts were explored under acidic,
neutral, and basic conditions at up to 100 °C and at a
maximum pressure of 60 psig. In no case was greater than
1 area % (HPLC) of 1 observed. Interesting, if not particu-
larly useful for avoiding solid wastes, results were obtained
with a large excess of active nickel. While 6 was found to
be unreactive with Ni, as much as a 38% yield of 1 could
be obtained from the reduction of olefin 4 with most of the
remaining mass present as 6.
Hydride-reducing agents were also studied. The results
of these experiments are listed in Table 1. Either no reaction
or inefficient production of thiolamide 7 was observed.
Results
Initially, considerable effort was dedicated to developing
a catalytic hydrogenation version of the reduction of
intermediates 6 or 4 to 1. This approach was used in the
initial synthesis of LY213829 by the discovery chemists.4
These conditions are near optimal from the perspective of
solid waste. Due to the high value of the noble metal
Only the recently reported desulfurization conditions
utilizing NaBH4 in the presence of NiII or CoII provided a
reasonable yield of 1.5 In exploratory experiments a greater
than 50% in situ yield of 1 was obtained. Unfortunately, these
reactions require stoichiometric metal. Attempts to make the
reaction catalytic, or use a different catalyst, were unsuc-
cessful. This is in agreement with the literature. This reaction
was not optimized since stoichiometric Ni or Co, with the
environmental concerns those metals carry, was not superior
to the large excess of Zn.
† Chemical Process Development, Lilly Research Laboratories, Lafayette,
Indiana.
‡ Chemical Process Research and Development, Lilly Research Laboratories,
Indianapolis, Indiana.
(1) Panetta, J. A.; Shadle, J. D.; Phillips, M. L.; Benslay, D. N.; Ho, P. P. K.
Ann. N.Y. Acad. Sci. 1993, 696, 415.
(2) Hansen, M. M.; Harkness, A. R. Tetrahedron Lett., 1994, 35, 6971.
(3) Preliminary communication: Copp, J. D.; Ginah, F. O.; Hansen, M. M.;
Kjell, D. P., Slattery, B. J. Heterocycles 1998, 48, 1307.
(4) Philips, M. L.; Berry, D. M.; Panetta, J. A. J. Org. Chem. 1992, 57, 4047-
4049.
(5) Back, T. G.; Baron, D. L.; Yang, K. J. Org. Chem. 1993, 58, 2407.
10.1021/op990197j CCC: $19.00 © 2000 American Chemical Society and The Royal Society of Chemistry
Published on Web 05/06/2000
Vol. 4, No. 4, 2000 / Organic Process Research & Development
•
295

Scheme 1
Scheme 2
Scheme 3
Reductions employing dissolving metal conditions were
Because an alternative reducing agent for zinc was not
found, the two-step route shown in Scheme 3 was investi-
gated.
also explored. Interesting results were obtained with lithium
and tert-butyl alcohol in liquid ammonia. An unknown
(tentatively assigned as disulfide 8 based on 1H NMR) was
formed under these conditions (Scheme 2). By careful
addition of stoichiometric lithium a 90% isolated yield of 8
could be obtained. Addition of additional lithium did not
lead to further reduction to 1; instead a new compound
assigned as the methyl amide 9 by HPLC/MS was formed.
The disulfide 8 could be further reduced to 1 by using Zn in
acetic acid. While interesting, this result is of no commercial
value. Attempts to increase the reducing power of the
dissolving metal conditions via a change to sodium or to
higher boiling amines lead to formation of the over-reduction
product 9.
Rhodanine 6 has been observed to react with ammonium
hydroxide at reflux to generate 7.3 All that would be required
was addition of formaldehyde to 7 to form 1. In fact, when
7 isolated from the ammonium hydroxide reaction was
exposed to formalin in a two-phase aqueous ammonia/toluene
system, 1 was formed. Isolation of 7 could be avoided by
direct addition of formalin and toluene to the reaction
mixture. Yields of 1 from these conditions varied from 30
to 50%.
The major source of yield loss and variability under these
aqueous conditions was via saponification to form the
thiolacid 10, which did not react appreciably with formal-
296
•
Vol. 4, No. 4, 2000 / Organic Process Research & Development

dehyde. We reasoned that nonaqueous conditions should be
used to avoid this side reaction. Therefore, gaseous ammonia
in methanol in a sealed system at 80 °C was explored. These
conditions were found to produce 7 contaminated with large
amounts of two very late eluting (by reverse phase HPLC)6
impurities. These impurities were assigned by HPLC-MS as
the diastereomers of the disulfide 11. Disulfide formation is
competitive well before consumption of the starting 6 is
complete. At 50% consumption of rhodanine 6 the yields
were 22% 7 and 17% 11. The mercaptan would be expected
to oxidize easily under these very basic and elevated
temperature conditions. Therefore, we added formalin to the
gaseous ammonia in a methanol solution of 6 prior to heating
the mixture in an attempt to trap 7 prior to oxidation. This
approach to the formation of the desired thiazolidinone 1
was successful. Under optimum conditions (10 equiv NH3,
1.1 equiv formalin, 80-95 °C) an in situ yield of 90%
thiazolidinone 1 was achieved. The remaining mass was
primarily the starting material 6 (3.3%) and the disulfides
11 (6.4%). At lower formaldehyde stoichiometries, consump-
tion of intermediate 7 is less complete. At higher stoichi-
ometries, reaction of a second equivalent of formaldehyde
to form 12 becomes competitive. These conditions are no
longer nonaqueous as some water is reintroduced with the
formalin, yet thiolacid 10 is not observed. We were unable
to achieve any measurable yield of thiazolidinone 1 with any
of the commercially available formaldehyde oligomers or
acetals.
to 50 psia develops under the chosen conditions at 80 °C,
which is well within the safe operating range of all of our
lab and production equipment. Care must be taken to follow
the manufacturer’s recommendations for safe operation of
any laboratory pressure equipment. We successfully, and
safely, ran the process in pressure tubes, shaker bottles, and
metal pressure reactors. Production could be executed in
standard alloy tanks with isolated overheads due to the
moderate pressure generated.
Isolation of pharmaceutical quality 1 from the reaction
mixture proved problematic. Addition of solvents in which
1 is insoluble to the reaction mixture typically resulted in
intractable, tarry solids. Addition of acetic acid and water
led to the crystallization of 1. Filtration and washing of the
solids with water led to 1 without measurable thiourea. This
isolation procedure marginally raised the quality from the
in situ results (to 91% 1), primarily by rejecting most of the
starting material 6 (from 3.4 to 1.1%). The disulfide was
not rejected. Further purification and a change in crystal form
were still required. These goals were achieved via recystal-
lization from ethyl acetate. Overall, the process as developed
showed yields of 67-72%, HPLC purities of greater than
99%, and total related substances (by HPLC) of <0.4%.
In conclusion, we have developed a modified synthesis
of 1 which avoids the generation of a large quantity of solid
Zn waste. The new route is competitive economically with
the Zn-based reduction and provides an option if Zn
reclamation is not viable.
Experimental Section
Compounds 4 and 6 were prepared as described.2,4
Compound 1 had spectral data identical to that reported.4
The purity of 1 was assessed by HPLC using a reference
standard.
4-Thiazolidinone-5-[3,5-bis(1,1-dimethylethyl)-4-hy-
droxyphenyl] methyl (1). Compound 6 (3.22 g, 9.2 mmol),
and methanol (13 mL) were combined in a 25 mL glass
pressure tube equipped with a magnetic stir bar. Ammonia
(1.6 g, 91 mmol), and formalin (0.76 mL, 0.8 g, 10 mmol)
were added sequentially at 0 °C. The pressure tube was
sealed and heated to 80 °C for 8 h with good stirring. The
resulting reaction mixture was transferred to a three-neck
50-mL flask equipped with mechanical stirring. Deionized
water (12 mL) was added in a dropwise manner followed
by acetic acid (6 mL). After 15 min, the reaction mixture
was cooled to 0 °C for 1 h. The resulting colorless crystals
were recystallized from ethyl acetate (3 mL), filtered, and
washed with water (2 × 10 mL) and 1/4 ethyl acetate/hexane
(2 × 10 mL). The product was then dried in an oven at 50
°C to give 2.04 g (69%) of white, crystalline 1. Purity by
HPLC was 99.9%.6
Thiourea is formed stoichiometrically in the reaction. This
is an area of concern since thiourea is a suspect carcinogen.
Therefore, a tradeoff exists between these conditions and
those of the zinc procedure. We have replaced a large solid
zinc waste stream, but we have also introduced a carcinogen
concern.
To ensure that the product was not contaminated by
thiourea a separate assay was developed with a detection
limit of 2 ppm. No thiourea contamination of the product
resulting from the isolation procedure described below was
detected. All the liquid waste streams could be incinerated
to obviate environmental impact.
The pressure generated by these conditions is another
added concern. At both the laboratory and manufacturing
scale pressure reactions have inherent risk. The conditions
chosen were designed to mitigate the risk. A pressure of 40
(6) Suitable HPLC conditions are: Zorbax RX-C8 column (25 cm × 4.6 mm,
5 µm), detect at 280 nm, 1 mL/min flow, isocratic elution with 70% CH3-
CN, 30% of aqueous 60 mM phosphoric acid, and 10 mM octanesulfonic
acid adjusted to pH 2 with NaOH.
Received for review September 28, 1999.
OP990197J
Vol. 4, No. 4, 2000 / Organic Process Research & Development
•
297