The chemical development and scale-up of sampatrilat

Article abstract of DOI:10.1021/op020091f

The discovery and scale-up of two routes to sampatrilat are described. The first Chemical R and D route used a side product from another development project to accelerate drug supply and expedite the early development programme. The second, more efficient Chemical R and D route had the potential for commercialisation and used an environmentally friendly variant of the Baylis-Hillman reaction, and an asymmetric Michael addition as key steps. Full preparative details for the aminomethacrylate 4, a potentially useful chiral synthon, are given for the first time, along with full experimental details of the asymmetric Michael addition to make the chiral glutarate 5. Finally, a striking polymorph case history is described.

Full text of DOI:10.1021/op020091f

Organic Process Research & Development 2003, 7, 244−253
The Chemical Development and Scale-Up of Sampatrilat1
Peter J. Dunn,* Michael L. Hughes, Patricia M. Searle, and Albert S. Wood
Department of Chemical Research and DeVelopment, Pfizer Global Research and DeVelopment,
Sandwich, Kent CT13 9NJ, United Kingdom
Abstract:
The discovery and scale-up of two routes to sampatrilat are
described. The first Chemical R and D route used a side product
from another development project to accelerate drug supply
and expedite the early development programme. The second,
more efficient Chemical R and D route had the potential for
commercialisation and used an environmentally friendly variant
of the Baylis-Hillman reaction, and an asymmetric Michael
Figure 1.
addition as key steps. Full preparative details for the ami-
Scheme 1. Medicinal Chemistry route to sampatrilata
nomethacrylate 4, a potentially useful chiral synthon, are given
for the first time, along with full experimental details of the
asymmetric Michael addition to make the chiral glutarate 5.
Finally, a striking polymorph case history is described.
Introduction
During the 1990s Pfizer conducted clinical trials on two
geminal-cycloalkylglutaramide derivatives candoxatrilat (1),
or its indanyl ester prodrug candoxatril, and sampatrilat (2).
These compounds are inhibitors of the zinc metalloprotease,
neutral endopeptidase 24.11, and potentiate the natriuretic
factor actions of the peptide hormone, atrial natriuretic factor
(ANF) in animals and man.2 In addition, sampatrilat also
inhibits the angiotensin converting enzyme (ACE) and was
thus a dual inhibitor (Figure 1).
Medicinal Chemistry Route. The medicinal chemistry
route, outlined in Scheme 1, provided access to a good range
of derivatives during the medicinal chemistry programme
by functionalisation of intermediates such as 5, 6, or 7.
Although this route was used to prepare the first 500 g in
Chemical R and D, it suffered from a number of problems:
(1) Several chromatographic purifications were required
because there were only two crystalline intermediates.
(2) The synthesis was 15 steps from commercially
available materials and proceeded in 2% overall yield from
diethyl malonate using either chiral or achiral variants.
(3) The formation of the bromomethacrylate 3 had
extremely poor atom utilisation.
a Conditions: (i) KHCO3, CH2CO. (ii) HBr (aq), heat. (iii) 2-Methyl-1-
propene, p-TSA (0.1 equiv), CH2Cl2, rt. (iv) N-Ethyldiisopropylamine (1.2 equiv),
toluene 40 °C, 5 h. (v) K2CO3 (1.2 equiv), CH3CN 60-70 °C, 18 h. (vi)
Cyclopentanecarboxylic acid, LDA (2.2 equiv), 0 °C, 1 h, then cool to -50 °C.
Add acrylate (1 equiv) -50 to -20 °C, 6 h. Quench into ice cold ether/1 M
HCl. (vii) 5 or 6 (1 equiv) S-tert-butyl-O4-butyltyrosinate (1.1 equiv), CH2Cl2 0
to 25 °C. (viii) H2, Pd(OH)2, ethanol, 20 °C, 60 psi.
Although the Michael addition reaction to give 5 pro-
ceeded with a high level of diastereoselectivity, it was
unreliable. The achiral variant proceeding through compound
6 was preferred by medicinal chemistry for compound
supply.
(4) The use of large volumes of constant boiling HBr in
a slow reaction (step 2) was a scale-up and corrosion
problem.
(5) The bromomethacrylate 3 was a severe lachrymator.
At an early stage it was found that the intermediate 7
was extremely crystalline and high melting (130 °C compared
with the diastereoisomeric aminomethyl compound which
melted at 5 °C). This intermediate provided significant
purification, and this was a strong driver for syntheses to
proceed though this intermediate. Indeed all three syntheses
that were scaled up used 7 as an intermediate.
(1) This chemistry was presented at the First International Conference on
Organic Process Research and Development, San Francisco, CA, U.S.A.,
November 5, 1997.
(2) James, K.; Alabaster, C. T.; Barclay, P. L.; Barnish, I. T.; Blackburn, K. J.;
Brown, D.; Cambell, S. F.; Cussans, N. J.; Danilewicz, J. C.; Palmer, M.
J.; Terrett, N. K.; Samuels, G. M. R.; Wythes, M. J. Perspect. Med. Chem.
1993, 45-60.
244
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development
10.1021/op020091f CCC: $25.00 © 2003 American Chemical Society
Published on Web 03/22/2003

Scheme 2. Chemical R & D route 1a
Figure 2. Potential cheap starting materials for the synthesis
of 4.
diastereomer was easily removed from the desired (S,S)-
product 7 by crystallisation.
This chemistry was scaled up in the pilot plant to provide
multikilogram batches of sampatrilat; however, colleagues
in Chemical R and D working on the candoxatril project
developed an enantiospecific synthesis of the glutarate
intermediate 8 using an asymmetric hydrogenation reaction.3,4
Hence, the ready supply of the unwanted enantiomer dried
up. In addition the low bioavailability of sampatrilat and the
relatively low selling price in the hypertension market meant
that a more cost-effective, enantiospecific synthesis was
required.
a Conditions: (i) KOBut (2 equiv), ButOH, 25 °C. (ii) (S)-tert-Butyltyrosinate
(PPACA) (2.5 equiv), two-phase, 25 °C. (iii) BnNH2 (4 equiv), BF3‚OEt2 (1.1
equiv), ∆PrOH. (iv) H2 (55-60 psi), Pd/C, EtOH, 30-40 °C. (v) (S)-R-
Methylbenzylamine (4 equiv), ∆EtOH.
Chemical R and D Route 2. Most of the problems with
the Medicinal Chemistry route are associated with the
inefficient synthesis and use of the bromomethacrylate 3.
An efficient synthesis of the chiral aminomethacrylate 4 in
combination with some salt formations to reduce the
chromatographic purification would alleviate most of the
problems. In addition to the conventional disconnection
approach, we also tried to identify cheap raw materials that
had some structural similarity with the target. Three such
materials (see Figure 2) were tert-butyl methacrylate (which
although not a standard item in many western catalogues is
available at low cost on a large scale5), propargylic alcohol,
and tert-butyl acrylate. Our attempts to activate the methyl
group of the methacrylate by radical bromination led to
radical polymerisation. At Pfizer it has been shown that the
methyl group could be activated by an iodosulphonation-
dehydroiodination-isomerisation sequence,3 but this chem-
istry was too expensive to be used for sampatrilat. Car-
bonylation of inexpensive propargylic alcohol is not always
regioselective.6 Hence, the best approach seemed to be to
start from cheap tert-butyl acrylate and use the Baylis-
Hillman reaction7 to provide the additional carbon atom. The
successful realisation of this route is shown in Scheme 3.
Chemical R and D Route 1. The preparation of a 500 g
batch via the Medicinal Chemistry route was extremely
labour intensive, and in parallel with this activity an
alternative route for early drug supply was sought. At the
time, candoxatril was being prepared by a classical resolution
from the racemic glutarate 8 (Scheme 2). The desired (S)-
isomer was converted into candoxatril, but a stock of 200
kg of the unwanted (R)-isomer was available for potential
conversion to sampatrilat (Scheme 2). Base-induced elimina-
tion of methoxyethanol gave the acrylate 9. An attempted
asymmetric Michael addition of (S)-R-methylbenzylamine
to the acrylate 9 gave some diasteroselectivity (2.7:1) but
resulted in formation of the lactam 10. To avoid this
lactamisation the tyrosine moiety was introduced early to
give the acrylate 11. Michael additions to 11 were examined
using both chiral R-methylbenzylamine and also achiral
benzylamine. The chiral reaction gave encouraging selectivity
in the initial stages of the reaction (3:1), but this dropped to
1.4:1 as the reaction proceeded. It was harder to push the
chiral reaction to completion and more difficult to hydro-
genolyse the R-methylbenzyl protecting group. As the overall
yield from 11 to 7 was the same using either reagent, the
cheaper, benzylamine variant was selected for scale-up.
However, unacceptable levels of epimerisation of the tyrosine
chiral centre were encountered with either amine. A large
number of Lewis acids were screened for their ability to
accelerate the desired Michael addition reaction without
increasing the rate of tyrosine epimerisation, and boron
trifluoride was found to be the best catalyst. The addition of
boron trifluoride diethyl etherate complex eliminated the
epimerisation problem and gave a 4-fold increase in reaction
rate. Following hydrogenation, the low melting (R,S)-
(3) Challenger, S.; Derrick, A.; Mason, C. P.; Silk, T. V. Tetrahedron Lett.
1999, 40, 2187-2190.
(4) Burk, M. J.; Bienewald, F.; Challenger, S.; Derrick, A.; Ramsden, J. A. J.
Org. Chem. 1999, 64, 3290-3298.
(5) Available from Tokyo Kasei Organic Chemicals and Mitsubishi Chemicals.
(6) Rosenthal, R. W.; Schartzman, L. H.; Greco, N. P.; Proper, R. J. Org. Chem.
1963, 28, 2835-2838.
(7) Baylis, A. D.; Hillman, M. A. D. German Patent 2 155 133 (1972); Drewes,
S. E.; Roos G. H. P. Tetrahedron 1988, 44, 4653; Ciganek, E. Org React.
1997, 51, 201; Morita, K.; Suzuki, Y.; Hirose, H.; Bull. Chem. Soc. Jpn.
1968, 41, 2815; Mathias, L. J.; Kosefoglu, S. H.; Kress, A. O. Macromol-
ecules 1987, 20, 2326.
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
245

Scheme 3. Chemical R & D route 2 to sampatrilata
thermic decomposition (by DSC) until 200 °C. Finally the
“penny dropped” when it was noted that the chromato-
graphed Baylis-Hillman products polymerised and solidified
faster than the crude materials. This led to the realisation
that the acidic phenolic stabilisers (such as dihydroquinone
and monomethylhydroquinone) that are present in tert-butyl
acrylate were being removed on aqueous work up by the
basic 3-quinuclidinol. The addition of 50-100 ppm of fresh
stabiliser before the toluene solution was stripped restored
sufficient margin of safety (DSC in the presence of stabiliser
showed an exotherm at 177 °C which in combination with
other standard tests gave us confidence to transfer to the pilot
plant with no safety concerns). The Baylis-Hillman reaction
is an inherently environmentally friendly, atom-efficient
reaction in which all the atoms of the two starting materials
get incorporated into the product. The process developed by
Pfizer has added environmental benefit since it allows simple
reuse of the 3-quinuclidinol as shown in Figure 4 and gen-
erates no waste stream. A small amount of acetonitrile is
added back into the aqueous layer before it is reused in the
next reaction. This is to replace the acetonitrile which is
partitioned into the toluene layer. It is critical to have enough
acetonitrile to maintain a homogeneous reaction, otherwise
side reactions are much greater.
a Conditions: (i) CH2O (1.6. equiv), 3-quinuclidinol (0.25 equiv), ∆H2O/
CH3CN. (ii) SOCl2 (0.88 equiv), Et3N (1.02 equiv), Py (0.1 equiv). (iii) (S,S)-
Bis(R-methylbenzyl)amine (15) (0.66 equiv). (iv) Cyclopentanecarboxylic acid
(1.1 equiv), LDA (2.2 equiv), THF -30 to 20°C, add 4/hexane, quench citric
acid. (v) CDI (1.05 equiv), HOBT (0.1 equiv), tert-butyl-O4-tert-butyltyrosine
(1.05 equiv), toluene 90-95 °C. (vi) H2 50-60 psi, Pd/C, IMS, 25-35 °C.
The hydroxymethacrylate 13 was chlorinated with thionyl
chloride in acetonitrile and reacted with the chiral amine 15
to give the desired aminomethacrylate 4 in 90% yield.
However, on the plant scale it was decided to use a slightly
lower-yielding but more practical option. The process
selected for scale-up used the second-best reaction solvent,
toluene, which was used to extract the Baylis-Hillman
product 13. The dried toluene extract was then used directly
in a highly streamlined process to the desired product 4, in
good yield and avoiding isolation of the lachrymatory
chloromethacrylate 14. Compound 4 was conveniently puri-
fied by 5-sulphosalicylic acid salt formation, providing an
expedient large-scale synthesis of a potentially useful chiral
synthon 4.
Once the chiral synthon 4 was obtained in a pure
crystalline form, the problems which had been associated
with the unpredictability of the asymmetric Michael addition
went away. The original medicinal chemistry procedure used
cryogenic conditions at -78 °C, but as can be seen in Table
1, very good selectivities can be obtained using plant-friendly
temperatures. The conditions highlighted in bold in Table 1
were selected for scale-up as these conditions gave the
maximum yield in standard chemical plant. It has been
proposed8 that the excellent stereoselectivity is due to a six-
membered lithium chelate in which the lithium counterion
is complexed between the enolate oxygen and the â-nitrogen
atom. This chelate facilitates transmission of stereochemical
information present in the two (S)-R-methylbenzyl protecting
groups to the newly developing carbon-hydrogen bond.
The coupling reaction to give the amide 16 was originally
scaled up in the pilot plant using 2-3 equiv of propane-
phosphonic acid cyclic anhydride (PPACA), but a cheaper
Figure 3. Reaction profile of Baylis-Hillman reaction.
The Baylis-Hillman reaction in nonaqueous solvents was
complex, leading to a range of ethers and acetals. However,
it was found that by using aqueous acetonitrile (or other
aqueous solvents) and 1 equiv of 3-quinuclidinol the reaction
was clean, and 85-90% yields were obtained. 3-Quinucli-
dinol was by far the most expensive component of this
reaction; hence, we looked at making the reaction catalytic.
This reduced the chemical yield to 80-85%, but overall it
was still preferred on an economic basis. A typical profile
is shown in Figure 3. As can be seen from the graph the
level of the ether by-product increases with time, and the
maximum yield of product is obtained after 2 h. On the plant,
the preferred reaction period was 90 min due to the longer
heat-up and sampling times. There was one last frustrating
problem to overcome. It was discovered that the Baylis-
Hillman product mixture did not have a sufficient margin
of safety for scale-up into the pilot plant. Its DSC showed
an exotherm starting at 110 °C, and we wanted to run the
reaction at reflux (77 °C) as we had found that lower-
temperature reactions were less clean. Exothermic polymeri-
sations are a well-known safety hazard, but strangely the
starting tert-butyl acrylate did not show evidence of exo-
(8) Barnish, I. T.; Corless, M.; Dunn, P. J.; Ellis, D.; Finn, P. W.; Hardstone,
D. J.; James, K. Tetrahedron Lett. 1993, 34, 1323-1326.
246
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

Figure 4.
alternative was sought. It was shown that the reaction of the
glutarate 5 with N,N′-carbonyldiimidazole (CDI) cleanly gave
the imidazolide 17 but that this imidazolide would not react
with di-tert-butyl tyrosine even at elevated temperatures.
Presumably the imidazolide is too hindered due to the
geminal cyclopentyl group. The addition of a catalytic
quantity of 1-hydroxybenzotriazole solved this problem, and
the imidazolide reacted cleanly with the tyrosine derivative
to give the desired amide 16. CDI is an extremely water-
sensitive, reagent and the solution of the glutarate 5 needed
to be dry following the salt-breaking operation. It was found
that on scale-up the glutarate 5 was exquisitely sensitive to
pH, heat, and especially both (see Figure 5). Heating a
solution of the glutarate 5 in toluene, washed with water at
pH 7, at reflux for 17 h gave a 70% conversion to the diacid
18. If the pH of the wash was pH 4, there was a 95%
conversion to the diacid 18 under the same conditions. One
solution to the problem, which worked on scale-up, was to
wash the toluene solution with brine which gave a low water
content and then charge an extra 2 or 3% of CDI to complete
the drying process.
It is proposed that the sensitivity of the glutarate 5 to pH
and heat is due to an intramolecular process such as that
shown in Figure 5.
Subsequent hydrogenation of 16 gave the common
intermediate 7. It was found that Johnson-Matthey catalyst
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
247

Table 1. Effect of Temperature on the Michael Addition
Scheme 4. Endgame chemistrya
Reaction to give 5
reaction
temperature
yielda
(%)
diasteroeselectivity
of reaction mixture
-78 °C
75
82
88
88
99:1
98:2
97:3b
95:5
-30 °C
-5 °C (-10 to 0 °C)
+20 °C
a The yield after isolation as salt. b Diasteroselectivity during the reaction.
The selectivity was improved to 99.2:0.8 after salt formation.
a Conditions: (i) WSCDI (1.4 equiv), HOBT (1.05 equiv), CH2Cl2, 10-20
°C. (ii) TFA (22 equiv), anisole (16 equiv), CH2Cl2, 35 °C. (iii) NaOH (2 equiv),
Pd/C, H2 (60 psi), 21-29 °C, then HCl (aq) to pH ) 4.
Scheme 5. Polymorph interconversion before the discovery
of the new polymorph
Figure 5.
type 5R39 (5%Pd/C) was more effective that several other
Pd catalysts including Pearlman’s catalyst [20% Pd(OH)2/
C]. Pearlman’s catalyst is commonly used for difficult
debenzylation reactions but was clearly inferior for this
substrate. In our view the 4-fold lower palladium content of
the type 39 catalyst compared with that of Pearlman’s catalyst
also gave a greater margin of safety (with respect to
flammability) when the catalyst was filtered off.
Scheme 6. Polymorph interconversion after the discovery
of the new polymorph
End-Game Chemistry and a Striking New Polymorph.
All three routes conclude with the same final three steps
(Scheme 4). The first coupling step to give 19 was left
unchanged from the Medicinal Chemistry method. However,
on scale-up of the second step, the trifluoroacetic acid
induced deprotection of 19, severe problems were found
during the increased distillation time cycles used to remove
the trifluoroacetic acid. The increased exposure time led to
a variety of degradation reactions, and in some reactions less
than 50% of the desired product was obtained. The problem
was solved by the addition of water after the reaction was
complete to remove the TFA by washing rather than
distillation. This led to a three-layer system and a complex
extraction process, which, while unconventional, worked well
over many batches. Hydrogenolysis of the Cbz group was
originally achieved in an ethanol or ethyl acetate solution,
and the Medicinal Chemistry preparations involved isolating
sampatrilat as an amorphous freeze-dried solid. This was
replaced by the discovery of a crystalline hydrate which was
good for purification by recrystallisation from aqueous
methanol. The purified hydrate was dehydrated in acetone,
containing less than 3% water, to give an anhydrous
polymorph, which melted at 185 °C. This material was used
for early toxicology and clinical work, but after making
several batches of sampatrilat, a routine purification rework
using the aqueous methanol recrystallisation on a 25 L scale
set solid. It took 45 min to carefully chip out the thermometer
from the solid mass so that the material could be recovereds
a new polymorph had been formed!!
The new form melted at 256 °C and was designated P256.
This melting point was over 70 °C higher than that of the
previous anhydrous form that was designated P185. From
248
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

Table 2. Comparison of Polymorphs P185 and P256
polymorph
mp (DSC)
heat of fusion
solubility in water
P185 (original polymorph used in clinical trials)
P256 (new polymorph)
185 °C
256 °C
50-70 J/g
240 J/g
600 mg/mL
1 mg/mL
the melting point, heats of fusion and solubility data given
in Table 2 it was obvious that the new form was significantly
more stable. The interconversion diagram before the dis-
covery of the new form is shown in Scheme 5. On appear-
ance of the new form the interconversion diagram changed
dramatically and is shown in Scheme 6. Several of the
transformations in Scheme 5 are now historical even though
they were previously run on a multikilogram scale before
the discovery of the new form. It has been asserted that a
polymorph once formed can always be made again,9 and
indeed a new way of preparing P185 was found by freeze-
drying the more stable P256 to give amorphous material and
then reslurrying the amorphous material in 1-propanol or
acetonitrile. However, thus far we have been unable to find
a new way of making the hydrate. In the current intercon-
version diagram (Scheme 6) all arrows lead away from the
hydrate. Even though a 100 g sample of this material still
exists we have not yet identified a way of making further
material.
It was now found that the old process could no longer be
used to make sampatrilat, as the new form was so insoluble
that it crystallised out onto the catalyst, stopping the reaction.
The process was therefore redesigned by dissolving the
starting material 20 in aqueous sodium hydroxide for
hydrogenation to give the disodium salt of sampatrilat. After
the catalyst was removed by filtration, the resulting solution
was acidified to its isoeletric point (pH 4) to precipitate
sampatrilat (polymorph P256) which was collected by
filtration. The discovery of the new form had three advan-
tages.
undertook screening for polymorphs at an even earlier stage
in the development process and to use automated, high-
throughput screening to maximize the chances of finding all
polymorphs.
Experimental Section
Proton and carbon-13 NMR data were recorded on a
Varian Unity 300 spectrometer operating at 300 and 62.9
MHz, respectively, unless otherwise stated. Microanalytical
data were performed by Reginald English at Pfizer. Melting
points were determined on a Buchi melting point apparatus.
Ethanol refers to industrial methylated spirits (95% ethanol,
5% methanol).
1-[2-(tert-Butoxycarbonyl)-2-propenyl]-1-cyclopentane-
carboxylic acid (9). The glutarate 83 (as its R-methylben-
zylamine salt) (42 kg, 107.2 mol) was subjected to a salt-
break operation between hexane (100 L) and 2 M hydrochloric
acid solution (100 L). The hexane phase was washed with 2
M hydrochloric acid (2 × 50 L) followed by water (50 L).
The resulting hexane layer was concentrated to a yellow oil,
dissolved in tert-butyl alcohol (210 kg), and treated with
potassium tert-butoxide (24.18 kg, 215.5 mol). The reaction
was stirred for 20 h at 25 °C to give a very thick slurry.
Dichloromethane (437.5 L) and 1 M hydrochloric acid (267
L) were added, keeping the temperature below 30 °C. After
stirring for 30 min the phases were separated, and the organic
phase was washed with water (267 L). The dichloromethane
layer was stripped to a low volume and then stripped and
replaced with ethyl acetate so that the final volume was
around 200 L. The ethyl acetate solution was washed with
a solution of sodium bicarbonate (made by dissolving 200 g
of NaHCO3 in 32 L of water) to remove the diacid biproduct.
Finally, the organic layer was washed with water (32 L) and
concentrated to an oil which solidified on standing. Yield
of the title compound (27.35 kg, 99.8%). Purity by main
band HPLC ) 93.5%.
(i) The new form was very insoluble, and this resulted in
a 15% yield improvement.
(ii) Biological performance over the old form was
unchanged, and we were able to file for a new polymorph
patent which has been granted.10
(iii) The new form (P256) was significantly less hygro-
scopic than P185.10
The crude product was recrytallised from isooctane
(crystallisation at -10 °C in a freezer) to give a white solid
mp 50-51 °C. Found C, 66.16; H, 8.89. C14H22O4 requires
Conclusions
The hypertension market is cost competitive; as a result
efficient, convergent, asymmetric chemistry was required for
sampatrilat. An atom-efficient Baylis-Hillman reaction was
devised and converted into an environmentally friendly
process. A scalable, efficient synthesis of the chiral ami-
nomethacryalte 4 was developed, and this material was used
in an asymmetric Michael addition reaction to prepare the
glutarate 5. The overall yield for preparing sampatrilat was
improved from 2 to 44.7%.
1
C, 66.11; H, 8.72%. H NMR (CDCl3) 1.48 (9H, s), 1.58-
1.80 (6H, m), 2.07-2.20 (2H, m), 2.70 (2H, s), 5.52 (1H,
s), 6.15 (1H, s).
Attempted Asymmetric Michael Addition of (S)-r-
Methylbenzylamine to the Acrylate (9). The acrylate 9 (508
mg, 2 mmol), (S)-R-methylbenzylamine (1.04 mL, 8 mmol),
and ethanol (2 mL) were heated at reflux for 14 h. The
reaction was stripped and the residue purifed by column
chromatography on silica; 10% ethyl acetate in hexane eluted
the lactam 10 (240 mg) as a 6:1 mixture of diastereoisomers.
This material was recrystallised from a small volume of ethyl
acetate hexane to give 10 as a 7.5:1 mixture of diastereo-
isomers as a white solid. Found C, 73.67; H, 8.55; N, 3.53.
This case history re-emphasises the importance of poly-
morph control. Following this project and others like it, Pfizer
(9) Dunitz, J. D.; Bernstein, J. Acc. Chem. Res. 1995, 28, 193-200.
(10) Dunn, P. J.; Hughes, M. L. World Patent WO 95/15308, 1994.
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
249

1
C22H31NO3 requires C, 73.92; H, 8.74; N, 3.92%. H NMR
(286 L). The organic layer was concentrated to an oil (55.9
kg, 96.4%) then dissolved in ethanol (45.8 kg) ready for the
next step.
(CDCl3, 250 MHz) 1.43 (9H, s), 1.56 (3H, d), 1.6-2.1 (2H,
m), 2.40-2.68 (2H, m), 3.05 (1H, d, d, d), 3.33 (1H, t), 6.17
(1H, q), 7.27-7.45 (5H, m). Further elution gave a (200 mg)
of oil shown by NMR to be a 10:12 ratio of diasteroisomers
favouring the minor diastereoisomer. Overall the combined
yield of the two isomers was 61.5% and an overall ratio of
2:1.
A small portion was purified for analysis by medium-
pressure chromatography over silica, eluting with ethyl
acetate/hexane mixtures. Ethyl acetate (20%) in hexane eluted
the title compound as a 1:1 mixture of diastereoisomers as
a colourless oil. Found C, 71.59; H, 8.77; N, 4.20. C38H56N2O5
requires C, 71.67; H, 8.86; N, 4.40%. 1H NMR (CDCl3) 1.35
(9H, s), 1.40 [9H, 2 × s, (two diasteroisomers)], 1.45 (9H,
s), 1.5-2.02 (10H, m), 2.32-3.0 (4H, m), 3.04 (2H, d, J )
8 Hz), 3.75 (2H, m), 4.68 (1H, m), 6.80 and 6.42 [1H, 2 ×
d, J ) 8 Hz, (two diastereoisomers)], 6.92 (2H, d, J ) 8
Hz), 7.06 (2H, d, J ) 8 Hz), 7.32 (5H, m). 13
tert-Butyl (S)-N-{1-[2-(tert-Butyloxycarbonyl)-2-pro-
penyl]-1-cyclopentylcarbonyl}-O4-tert-butyltyrosinate (11).
Dichloromethane (131.5 kg), water (66.8 L), acrylate 9 (17.65
kg, 93.5% purity, 54.5 mol), and tert-butyl O4-tert-butyl-
tyrosinate hydrochloride11 (21.4 kg, 64.9 mol) were charged
to a reactor equipped for pH monitoring and simultaneous
addition of two reagents. The pH was adjusted to pH 8.5 to
9.0 with 20% sodium hydroxide solution. A 50% solution
of propanephosphonic acid cyclic anhydride in ethyl acetate
(124 kg) was added over about 1 h with simultaneous
addition of 20% sodium hydroxide solution (approximately
132 kg) maintaining the pH in the range 8.5 to 9.0. During
these additions the temperature was maintained at 25 °C.
The reaction was stirred at 25 °C for 3 h. The phases were
allowed to separate (product containing phase on the top).
The aqueous layer was extracted twice with dichloromethane
(2 × 52 kg) and the combined organic phase washed with
water (40 L). The organic phase was concentrated to low
volume and ethyl acetate (55 kg) added along with hexane
(90.4 kg). The organic phase was washed with 1 M
hydrochloric acid (2 × 34.3 L) (to remove any unreacted
di-tert-butyl tyrosine), then with 10% potassium bicarbonate
solution (2 × 34.3 L), and finally with water (35 L). The
organic phase was concentrated to an oil (35.8 kg, 104.1%
weight yield) which was dissolved in 1-propanol (34.3 L)
ready for the next step.
C NMR (CDCl3)
[Where splitting is observed, it is reported as 2 × CH (etc.),
for peaks in the region 33-55 ppm, diasteromeric pairs are
less obvious and all resonances are reported.] 23.6 and 23.9
(2 × CH2), 24.3 and 24.5 (2 × CH2), 27.6 (CH3), 28.6 (CH3),
33.3 (CH2), 33.9 (CH2), 36.8 (CH2), 37.1 (CH2), 37.8 (CH2),
38.0 (CH2), 39.1 (CH2), 44.1 (CH), 44.4 (CH), 52.1 (CH),
52.3 (CH2), 53.4 (CH2), 53.6 (CH), 53.9 (CH2), 77.9 (C),
79.9 and 80.0 (2 × C), 81.4 and 81.5 (2 × C), 123.9 (CH),
126.6 and 126.8 (2 × CH), 127.9-128.1 (multiple CH
resonances), 129.5 and 129.6 (2 × CH), 131.3 and 131.5
(2 × C), 140.0 and 140.1 (2 × C), 153.9 and 154.0 (2 × C),
170.6 and 171.2 (2 × C), 174.6 and 174.7 (2 × C), 176.1
and 176.3 (2 × C).
tert-Butyl (S,S)-N-{1-[3-amino-2-(tert-butoxycarbonyl)-
propyl]-1-cyclopentylcarbonyl}-O4-tert-butyltyrosinate (7)
The amine (22) (55.9 kg, 87.8 mol dissolved in ethanol (45.8
kg) was diluted with further ethanol (176.5 kg) and hydro-
genated over 5% palladium/carbon (60% water wet paste)
(11.2 kg) at 30-40 °C and 55-60 psi pressure for about 8
h. When the reaction was complete, the catalyst was removed
by filtration through filteraid and washed with ethanol, and
the resulting ethanol solution was concentrated to low volume
(approximately 70 L). Hexane (120 L) was added, the
mixture was cooled to 5-10 °C, and the product was
collected by filtration and washed with hexane (40 L). The
wet cake was reslurried with hexane (70 L), refiltered, and
washed with hexane (23 L). The resulting damp cake (17.8
kg as two crops) was recrystallised from cyclohexane (89
L), cooled to 5-10 °C, and granulated for 3 h. The title
product was collected by filtration, washed with cyclohexane
(18 L), and dried at 50 °C in vacuo. Yield 15.53 kg, 32.4%
(theoretical yield 50%) mp 129-130 °C (lit.12 122-127 °C).
1H NMR (CDCl3) 1.32 (9H, s), 1.40 (9H, s), 1.45 (9H, s),
1.55-1.75 (6H, m), 1.78-2.06 (4H, m), 2.23 (1H, m), 2.62
(1H, dd, J ) 12.8, 6.6 Hz), 2.64 (1H, dd, J ) 12.8, 6.8 Hz),
3.02 (2H, d, J ) 7 Hz), 4.73 (1H, q, J ) 7 Hz), 6.60 (1H,
d, J ) 8 Hz), 6.91 (2H, d, J ) 8.4 Hz), 7.08 (2H, d, J ) 8
Hz). The relative stereochemistry of (7) has previously been
established.8
A small portion was purified for analysis by medium-
pressure column chromatography over silica using hexane
as eluant. This gave a colourless oil which solidified on
standing, mp 66-68 °C. Found C, 70.24; H, 8.94; N, 2.87.
1
C31H47NO6 requires C, 70.29; H, 8.94; N, 2.64%. H NMR
(CDCl3, 250 MHz) 1.32 (s, 9H), 1.39 (s, 9H), 1.50 (s, 9H),
1.52-1.74 (6H, m), 1.79-1.98 (2H, m), 2.60 (2H, s), 3.02
(2H, s), 4.70 (1H, q, J ) 7 Hz), 5.41 [1H, s(br)], 6.06 (1H,
d, J ) 7 Hz), 6.08 [1H, s (br)], 6.90 (2H, d, J ) 8 Hz), 7.08
(2H, d, J ) 8 Hz).
tert-Butyl (S)-N-{(R,S)-2-tert-Butoxycarbonyl)-3-ben-
zylaminopropyl}-1-cyclopentylcarbonyl)-O4-tert-butylty-
rosinate (22). The acrylate 11 (48.2 kg, 91 mol) was
dissolved in 1-propanol (70.5 L in total). Benzylamine (42.2
kg, 393.8 mol) was added, followed cautiously by boron
trifluoride diethyl etherate (15.44kg, 108.8 mol). The mixture
was heated at reflux for 6.75 h then allowed to cool naturally
overnight. The reaction was concentrated under vacuum to
remove 85 L of distillate, cooled to ambient, and diluted
with ethyl acetate (171.6 kg) and hexane (125.9 kg). The
resulting solution was washed with 1 M hydrochloric acid
(2 × 286 L) [to remove excess benzylamine] followed by
5% aqueous sodium bicarbonate solution (114 L) then water
tert-Butyl 2-(hydroxymethyl)propenoate (13). (SAFETY
NOTE: Severe contact dermatitis has been reported after
exposure to reaction products from formaldehyde and methyl
(12) Danilewicz, J. C.; James, K.; Kobylecki, R. J. European Patent 0358398,
1990.
(11) Beyerman, H. C.; Bontekoe, J. S. Receuil 1962, 81, 691-698.
250
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

acrylate.7 No issues were noted at Pfizer using the tert-butyl
ester, but given the structural similarity care should be taken.)
Water (459 kg) and 3-quinuclidinol (12.65 kg, 99.4 mol)
were charged to a reactor and stirred to give a clear solution.
Paraformaldehyde (19.1 kg, 636.7 mol) was added and the
solution heated to 40 °C before adding acetonitrile (201.5
kg). The resulting mixture was heated to reflux (77 °C) and
tert-butyl acrylate (51kg, 397.9 mol) added cautiously.
(SAFETY NOTE: The boiling point will drop during the
addition and vigorous boiling may occur if the addition rate
is too high.) The tert-butyl acrylate was washed in with
acetonitrile (50 L). The reaction was heated at reflux for 90
min, and then toluene (255 L) was cautiously added. On
completion of the toluene addition the mixture was cooled
to 30 °C, and the layers were separated. The bottom aqueous
layer was re-extracted with toluene (76.5 L) and then retained
for reuse of the 3-quinuclidinol. The combined toluene layers
were washed with saturated sodium chloride solution (2 ×
150 L). Water (100 L) was added and the pH of the aqueous
phase adjusted to pH ) 4 with 2 M citric acid solution. After
separating the phases, the toluene solution was washed with
saturated sodium chloride solution (2 × 100 L), and
monomethylhydroquinone (150 g) was added. (SAFETY
NOTE: It is important this stabiliser addition is made, see
text.) The toluene solution was stripped in vacuo at a
temperature of up to 50 °C to give a final volume of 120 L
(108.2 kg), which, by GC assay, was shown to contain 39.4
kg of the title compound (64.1% yield). A small portion of
the solution was purified by column chromatography on
silica, eluting with hexane and then 10% ethyl acetate in
hexane, to give the title compound 13 which was further
purifed by Kugelrohr distillation. Bp 80-100 °C/0.5 mmHg.
Found C, 60.74; H, 9.14. C8H14O3 requires C, 60.74; H,
mol) in toluene (83 L) was added over 1 h, keeping the
temperature below 15 °C. The resulting suspension was then
stirred for 1 h without cooling and then was degassed by
heating the mixture to 40-50 °C under vacuum to remove
any SO2. The mixture was cooled to 25 °C, and the toluene
solution of (S,S)-bis(R-methylbenzyl)amine (15) prepared
above was added over 30 min, keeping the temperature below
40 °C, followed by a toluene line wash (20 L). The reaction
was stirred overnight at 35-45 °C at which time an HPLC
sample showed 6% of the starting amine remained. Water
(520 kg) was added, and the phases were separated. The
aqueous layer was re-extracted with toluene (104 L), and
the combined toluene extracts were washed with water (125
L), adjusting to pH ) 4 with 2 M citric acid. Finally, the
toluene solution was washed with water (2 × 125 L).
Monomethylhydroquinone (5 g) was added (SAFETY
NOTE: It is important that this is done; see text.) and the
toluene solution concentrated to 360 L in vacuo. The toluene
solution was diluted with 2-butanone (317 L), and a solution
of 5-sulphosalicylic acid dihydrate (38.6 kg, 151.8 mol) in
2-butanone (106 L) was added followed by a seed of the
desired product. The resulting slurry was cooled to 0-5 °C
and granulated overnight. The title compound 4 was collected
by filtration, washed with 1:1 2-butanone:ethyl acetate (2 ×
30 L) and then ethyl acetate alone (2 × 50 L), and dried in
vacuo at 40 °C. Yield 79.83 kg, 81.1% of a white crystalline
solid, 97% pure by HPLC. A small sample was recrystallised
from ethyl acetate: 2-propanol (4:1) for microanalysis. Mp
139-141 °C. Found C, 63.83; H, 6.55; N, 2.48; S, 5.54.
C31H37NO8S requires C, 63.79; H, 6.39; N, 2.40; S, 5.49.
(S,S,S)-1-{3-[Bis(1-phenylethyl)aminomethyl]-2-(tert-
butoxycarbonyl)propyl}-1-cyclopentylcarboxylic Acid
5-Sulphosalicylic Acid Salt (5). Sodium hydroxide solution
(1 M, 7.784 L), the amine salt (2.273 kg, 3.894 mol) 4, and
hexane (5.15 L) were mixed and stirred until two clear layers
were obtained. The phases were separated, and the aqueous
phase was re-extracted with hexane (1.03 L) with the two
hexane extracts kept separate. The first hexane solution was
washed with water (2 × 1.5 L) and stirred with magnesium
sulphate (140 g) to remove entrained water. The magnesium
sulphate was filtered off and the second hexane extract passed
slowly through the magnesium sulphate so that it also was
dry.
1
8.92%. H NMR (CDCl3) 1.45 (9H, s), 2.9 [1H, m (br)],
4.21 (2H, s), 5.70 (1H, s), 6.10 (1H, s). 13C NMR 27.9 (CH3),
62.1 (CH2), 81.1 (C), 124.3 (CH), 140.8 (CH), 165,5 (C).
m/z 159, MH+, 103 (M - isobutylene).
tert-Butyl (S,S)-1-[2-Bis(1-phenylethyl)amino]propanoate
5-sulphosalicylate (4). Part A: Prepartion of Bis(R-meth-
ylbenzyl)amine (15) Free Base Solution. Water (82.8 L) was
charged to the reactor followed by toluene (82.8 L), (S,S)-
bis(R-methylbenzyl)amine hydrochloride (27.6 kg, 168.6
mol), and then 11.1 kg of 40% w/w sodium hydroxide
solution. When all the material had dissolved, the pH of the
mixture was approximately pH ) 10. The phases were
separated, and the aqueous phase was re-extracted with
toluene (27.6 L). The combined toluene phases were washed
with water (2 × 20 L) and then concentrated to ap-
proximately 60 L by atmospheric distillation to remove any
water present. The resulting amine solution in toluene was
used directly in the next procedure.
Part B: Chlorination and Alkylation. The toluene con-
centrate of the hydroxymethacrylate (13) (108.2 kg, contain-
ing 254.7 mol) prepared above was charged to the reactor
along with further toluene (354 L), pyridine (1.96 kg, 24.8
mol), and triethylamine (26.4 kg, 260.9 mol) washed in with
toluene (10 L). The resulting solution was cooled to 0 to
-5 °C, and a solution of thionyl chloride (26.7 kg, 224.4
THF (2.59 L) was charged to a 25-L vessel followed by
LDA solution (2895 g, 28.8%w/w in heptane ex FMC
lithium, 7.784 mol), and the resulting solution was cooled
to -30 °C. Cyclopentanecarboxylic acid (STENCH) (444.4
g, 3.895 mol) was added at -10 °C over 15 min and washed
in with THF (400 mL). The resulting slurry was stirred
overnight, allowing the temperature to rise to 20 °C to form
the dianion. The resulting red-ochre-coloured slurry was
cooled to -12 °C, and the hexane solution of the free base
(4) was added over 50 min, keeping the temperature between
-10 and 0 °C. The second hexane extract (see above) was
used to wash in the amine 4. The reaction was stirred at -5
°C for 30 min. The reaction was cooled to -15 °C and then
quenched with citric acid solution (2 M, 9.07 L) with
continuous cooling over 16 min (the resulting pH ) 4). The
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
251

layers were separated, and the aqueous phase was re-
extracted with hexane (1.5 L). The combined organics were
washed with water (3 × 2 L), and the organic phase was
concentrated to half volume in vacuo. 2-Butanone (6 L) was
added and the evaporation continued again to half volume.
The concentrate was diluted with 2-butanone (10 L) and
cooled to -10 °C. A solution of 5-sulphosalicylic acid
dihydrate (854.5 g, 3.362 mol) in 2-butanone (5 L) was added
over 30 min, keeping the temperature below 0 °C. The
resulting slurry was granulated at 0-5 °C for 2.5 h. The
title product 5 was collected by filtration, washed with
2-butanone, and dried in vacuo to give an off-white granular
solid (2.38 kg, 87.6%). Purity by HPLC was 98.4% with
0.8% of the diastereoisomer and 0.2% of the starting material
4 present. A small sample was recrystallised from ethyl
acetate/2-propanol to give material for microanalysis. Mp
172-174 °C. Found C, 63.45; H, 6.90; N, 2.07. C37H47NO10S
requires C, 63.68%; H, 6.79; N, 2.01%.
s), 1.4-1.7 (6H, m), 1.8-2.0 (4H, m), 2.55-2.80 (3H, m),
2.98 (2H, dd, J ) 13.5, 6.7 Hz), 3.11 (2H, dd, J ) 13.5, 4.8
Hz), 3.93 (2H, q, J ) 6.7 Hz), 4.70 (1H, q, 6 Hz), 6.10 (1H,
d, J ) 7.8 Hz), 6.89 (2H, d, J ) 6.9 Hz), 7.08 (2H, d, J )
6.9 Hz), 7.14-7.25 (10H, m).
tert-Butyl (S,S,S,S)-N-(1-{3-{Bis[1-phenylethyl)ami-
nomethyl]-2-(tert-butoxycarbonyl)propyl}-1-cyclopentyl-
carbonyl)-O4-tert-butyltyrosinate (16). Method 2 with N,N′-
Carbonyldiimidazole. The glutarate (5) (174.25 g, 0.25 mol),
sodium hydroxide (20.0 g, 0.5 mol) dissolved in water (700
mL) and toluene (870 mL) were mixed and stirred for 15
min. The aqueous layer was separated and the organic layer
washed with water (250 mL) then brine (520 mL). An
alternative method to drying by brine washing is stripping
out 1 mL/g of toluene, but this only worked well on a
laboratory scale (see text). 1-Hydroxybenzotriazole (low
water grade, 3.37 g, 0.025 mol) was added followed by N,N′-
carbonyldiimidazole (42.5 g, 0.2625 mol). The reaction was
stirred at 40 °C for 2 h to give a clear, pale-yellow solution.
tert-Butyl-O4-tert-butyltyrosine hydrochloride (86.6 g, 0.2625
mol) was added in one portion, and the reaction was heated
to 90-95 °C for 3 h to give a cloudy yellow solution. The
reaction was cooled to room temperature and washed with
hydrochloric acid (2 M, 175 mL). The two clear phases
separated, and the organic phase was washed with water (200
mL). The organic phase was evaporated to give the title
compound 16 as a pale yellow gum (191.6 g, 101.5%).
1H NMR (CDCl3) in agreement with that prepared by
method 1.
tert-Butyl (S,S,S,S)-N-(1-{3-{Bis[1-phenylethyl)ami-
nomethyl]-2-(tert-butoxycarbonyl)propyl}-1-cyclopentyl-
carbonyl)-O4-tert-butyltyrosinate (16). Method 1 with
1-Propanephosphonic Acid Cyclic Anhydride (PPACA).
Water (177 L), sodium hydroxide solution (40% w/w, 19
kg) the glutarate (5) (64.4 kg, 92.1 mol) and dichloromethane
(275.6 kg) were stirred to give two clear layers. The aqueous
layer was run off, and the dichloromethane layer was washed
with brine (2 × 65 L). The resulting glutarate solution was
transferred to a reactor equipped for simultaneous and
synchronous addition of two reagents with on-line pH
monitoring, and water was added (168 L). tert-Butyl O4-
tert-butyltyrosine hydrochloride (30.4 kg, 92.1 mol) was
added and the pH adjusted to 8.5-9.0 with 20% sodium
hydroxide solution. PPACA (50% in ethyl acetate, 83.4 kg)
was added, whilst maintaining the pH in the range 8.5 to
9.0 by the simultaneous addition of 20% sodium hydroxide
solution (approximately 75 kg added). During the additions
of PPACA and sodium hydroxide the temperature was
maintained below 25 °C. The reaction was stirred for 30
min before adding the second portion of PPACA (50% in
ethyl acetate, 83.4 kg), again keeping the pH in the region
8.5-9.0. The reaction was stirred for 5.5 h and then a further
66.5 kg of PPACA solution added, again keeping the pH
8.5-9.0.13 The reaction was left to separate overnight and
the bottom aqueous layer run off. Hexane (300 L) was added
and the organic solution washed with 1 M hydrochloric acid
(2 × 84 L) (to wash out any unreacted tyrosine derivative)
followed by water (2 × 84L with adjustment to pH 7 with
sodium bicarbonate). The organic layer was concentrated to
low volume by atmospheric distillation. Ethanol (55.2 kg)
was added to give 127.4 kg of concentrated solution
containing 72.2 kg of crude title product. Crude yield
104.3%.
tert-Butyl (S,S)-N-{1-[3-Amino-2-tert-butoxycarbonyl)-
propyl]-1-cyclopentylcarbonyl}-O4-tert-butyltyrosinate (7).
A portion of the above ethanol solution of the amine 16,
prepared by method 1, 107 kg containing 60.5 kg of crude
product was diluted with further ethanol (400 L) and
hydrogenated over 5% Pd/C type 5R39 at 50-60 psi and
25-35 °C until hydrogenation was complete. The catalyst
was filtered off and washed with ethanol. The filtrate was
concentrated to low volume (approximately 70 L) and hexane
(106.5 L) added. The resulting slurry was cooled to 5-10
°C, granulated overnight then collected by filtration, and
washed with hexane (43 L). The resulting damp cake (34.5
kg) was recrystallised from cyclohexane (174 L), granulated
for 3 h at 5-10 °C, collected by filtration, and dried in vacuo
to give the title compound 7 27.2 kg [70.2% yield from the
1
glutarate 5] mp 129-130 °C (lit.12 122-127 °C). H NMR
(CDCl3) identical to that prepared by CRD route 1 (see
above).
tert-Butyl (S,S,S)-N-{1-[3-(N6-Benzyloxycarbonyl-N2-
mesyllysylamino)-2-(tert-butoxycarbonyl)propyl]-1-cyclo-
pentylcarbonyl}-O4-tert-butyltyrosine (19). The amine 7
(20.81 kg, 38.06 mol) and the acid 2110 (13.64 kg, 38.06
mol) were dissolved in dichloromethane (166.2 kg). 1-Hy-
droxybenzotriazole hydrate (6.19 kg, 40.46 mol) was added
and the reaction cooled to 10 °C. 1-(3-Dimethylaminopro-
pyl)-3-ethylcarbodiimide hydrochloride (10.25 kg, 53.47 mol)
was added over about 15 min allowing the temperature to
rise to 20 °C and the reaction stirred overnight. The reaction
solution was washed with saturated sodium bicarbonate
A sample was stripped for NMR spectroscopy. 1H NMR
(CDCl3) 1.31 (9H, s), 1.33 (6H, d), 1.36 (9H, s), 1.37 (9H,
(13) The number of equivalents of PPACA required for complete reaction in
these heterogeneous amide coupling reactions depends heavily on the
efficiency of mixing in the reactor. For reactors with efficient stirring 2
equiv of PPACA can be used, and the third charge used in this preparation
would be unnecessary.
252
•
Vol. 7, No. 3, 2003 / Organic Process Research & Development

solution (2 × 29 L), followed by 1 M hydrochloric acid
(2 × 29 L), and finally with water (29 L, adjusting the pH
to 7.0). Dichloromethane (100 L) was stripped out, and
further dichloromethane (50 L) was added and then stripped
out again to give the title product 19 in dry dichloromethane
(107.2 kg of solution that was used directly in the next step).
A stripped aliquot indicated the yield was quantitative.
(S,S,S) -N-{1-[3-(N6-Benzyloxycarbonyl-N2-mesyllysyl-
amino)-2-carboxypropyl]-1-cyclopentylcarbonyl}tyro-
sine (20). The product from the last reaction (107.2 kg of
dichlormethane solution containing 19 (38.06 mol) was
charged to a reactor and washed in with dichloromethane
(16.8 kg) followed by anisole (64.5 kg). Trifluoroacetic acid
(97 kg, 850.7mol) was added to the stirred solution over a
period of 1 h allowing, the temperature to rise to (but not
exceed) 35 °C. When the addition was complete, the reaction
was stirred for 6 h at 30-35 °C and then allowed to cool
naturally overnight. Water (84.4 L) was added to the reaction
to give three phases. The phases were separated, and the
bottom oil phase was combined with the upper organic phase
and diluted with ethyl acetate (169 L). Saturated brine
solution (67.5 L) was added, and the bottom aqueous phase
was run off. Saturated brine solution (47 L) was added
followed by sodium bicarbonate (47 L, 10% w/w). The pH
of the aqueous phase was equal to 3. The mixture was stirred
for 15 min, and the three phases were allowed to separate.
The bottom oil phase was run off and dissolved in ethyl
acetate (47 L). The middle aqueous phase was run off and
the ethyl acetate solution of the oil added back to the reactor.
The product was extracted out of the aqueous layer as the
disodium salt by adding brine (47 L) and sodium bicarbonate
solution (210 L, 10%w/w). The layers were separated, and
the product containing aqueous phase was washed with ethyl
acetate (2 × 33 L). Finally, the product containing aqueous
phase was adjusted to pH 3 by the addition of 28 L of
concentrated hydrochloric acid and the product extracted
back into ethyl acetate (2 × 27.5 L). The ethyl acetate layer
was washed once with brine (23.5 L) to give 120.8 kg of
product 20 containing solution. An aliquot was stripped to
show the crude yield was 100%.
Polymorph Preparations and Interconversions. His-
torical Preparation of the Hydrate. A solution of Cbz-
Sampatrilat (20) (351 g, 0.49 mol) in ethyl acetate (1300
mL) was added to water (385 mL) and the two-phase mixture
hydrogenated at 60 psi and room temperature over a 5%
palladium-on-carbon catalyst (35 g) for 20 h. The catalyst
was filtered off, and the aqueous phase was separated and
concentrated to low volume under reduced pressure. The
viscous solution was poured into methanol (2.85 L) and
stirred at room temperature for 18 h during which time there
was a slow precipitation of a solid. The solid was granulated
at 5-10 °C for 2 h, filtered, washed with methanol, and dried
to give the title compound as a white solid (178.1 g, 60.5%)
mp 168-171 °C. Found C, 51.37; H, 7.47; N, 9.06.
C26H40N4O9S‚xH2O (where x ) 1) requires C, 51.81; H, 7.02;
N, 9.30%. Water content 3.6 wt % as determined by Karl
Fischer analysis (x ) 1 requires 3.0 wt %).
Historical CoVersion of the Hydrate to Polymorph (P185).
Sampatrilat hydrate (847.0 g, 1.41 mol) was dissolved in
water (762 mL) and diluted with acetone (1.0 L). This
solution was added to vigorously stirred acetone (18.05 L)
at room temperature, and a white solid precipitated. The
mixture was stirred at room temperature for 18 h, and the
solid was collected by filtration, washed with acetone, and
dried to give the title compound as a white solid (775 g,
94%), mp 179-181 °C (DSC 185 °C). Found C, 53.42; H,
6.88; N, 9.37; S, 5.49. C26H40N4O9S requires C, 53.41; H,
6.90; N, 9.58; S, 5.48%. This preparation had been scaled
up to 5.4 kg before the discovery of the new form made it
invalid.
Preparation of the Amorphous Form. Sampatrilat (P256)
(4 g, 6.84 mmol) was added to water (200 mL) and the
mixture stirred at 90-95 °C for 30 min. Insoluble material
was filtered off and the filtrate diluted further with water
(50 mL) and cooled to room temperature, After filtration to
remove a slight haze, the clear filtrate was frozen by using
a solid carbon dioxide/acetone bath. The solid mass obtained
was freeze-dried to yield the title compound as a white solid
(3.0 g, 75%). PXRD showed the material was amorphous.
ConVersion of the Amorphous Form to Polymorph (P185).
The amorphous form prepared above (0.3 g) was stirred in
1-propanol (10 mL) for 5 days. The resulting white solid
was collected by filtration and dried under reduced pressure
to give the title compound (0.26 g) mp 175-180 °C.
Further data (PXRD, IR) for all four physical forms can
be found in ref 10.
Sampatrilat (2). The product containing solution from
the last reaction (120.8 kg) was extracted with a sodium
hydroxide solution (3.01 kg of sodium hydroxide in 56 L of
water) followed by a water wash (2 L). The combined
organic layers were hydrogenated over 5% Pd/C (1.5 kg,
60% wet) at 60 psi and 21-29 °C for 7 h. The catalyst was
removed by filtration and washed with water (40 L). The
solution of the disodium salt was then neutralised with 5 M
hydrochloric acid to the isoelectric point, pH ) 4, and
granulated for 24 h at room temperature. The product was
collected by filtration, washed with water (4 × 15 L) and
then acetone (15 L), and dried in vacuo to give the title
compound 2 as a white crystalline solid as polymorph (P256)
(19.12 kg, 85.9% from compound 7. Mp 250 °C (lit.10 248-
250 °C).
Acknowledgment
We thank Jim Stichbury and Andrew Pearce for Pilot
Plant work and chemical development ideas, Gary Nichols
and Christopher Dallman for their work on the physical form,
and Michael Williams for his chemical input and support
for this project over several years.
An earlier preparation performed on a 0.3 mol scale10 gave
microanalysis as follows C, 53.47; H. 7.25, N, 9.50.
C26H40N4O6S requires C, 53.41; H, 6.90; N, 9.58%.
Received for review October 31, 2002.
OP020091F
Vol. 7, No. 3, 2003 / Organic Process Research & Development
•
253