Process improvements in the synthesis of corticosteroid 9,11β-epoxides

Article abstract of DOI:10.1021/op0102013

Corticosteroid 9,11β-epoxides are key intermediates in the preparation of pharmaceutically important compounds such as betamethasone, mometasone, beclomethasone, and dexamethasone. A new process for the 9,11β-epoxide was developed using a PCl₅-mediated regioselective dehydration of 11α-hydroxy-steroid to form the corresponding Δ^9,11 double bond. The olefin is then converted into 9α,11β-bromoformate by treatment with 1,3-dibromo-5, 5-dimethyl hydantoin (DBH) in DMF and subsequently cyclized to produce the desired 9,11β-epoxide upon treatment with NaOH. Major process-related impurities such as 21-OH-Δ^9,11-triene, 21-OH-Δ^11,12-triene, 21-Cl-Δ^9,11-triene, and β-epoxide-21-cathylate as well as 11β-Cl are all eliminated or minimized. This new process has been implemented in our manufacturing facility in full-scale production and proved to raise the overall yield and the quality of the product dramatically compared to the existing process.

Full text of DOI:10.1021/op0102013

Organic Process Research & Development 2001, 5, 376−382
Process Improvements in the Synthesis of Corticosteroid 9,11â-Epoxides
Xiaoyong Fu,* Chou-Hong Tann, and T. K. Thiruvengadam
Synthetic Chemistry Department, Schering Plough Research Institute, 1011 Morris AVe. Union, New Jersey 07083, U.S.A.
Abstract:
Corticosteroid 9,11â-epoxides are key intermediates in the
preparation of pharmaceutically important compounds such as
betamethasone, mometasone, beclomethasone, and dexametha-
sone. A new process for the 9,11â-epoxide was developed using
a PCl5-mediated regioselective dehydration of 11r-hydroxy-
steroid to form the corresponding ∆9,11 double bond. The olefin
is then converted into 9r,11â- bromoformate by treatment with
1,3-dibromo-5, 5-dimethyl hydantoin (DBH) in DMF and
subsequently cyclized to produce the desired 9,11â-epoxide upon
treatment with NaOH. Major process-related impurities such
as 21-OH-∆9,11-triene, 21-OH-∆11,12-triene, 21-Cl-∆9,11-triene,
Figure 1.
and â-epoxide-21-cathylate as well as 11â-Cl are all eliminated
Scheme 1. Existing synthesis of â-epoxide (3a)
or minimized. This new process has been implemented in our
manufacturing facility in full-scale production and proved to
raise the overall yield and the quality of the product dramati-
cally compared to the existing process.
Pharmacologically important corticosteroid drugs such as
betamethasone (1), mometasone (2), beclomethasone (4), and
dexamethasone (5) have been well received since their
introduction to the worldwide market. And they continue to
play major roles in dermatological and allergy therapies.1
Their structural similarity presents an opportunity to develop
a general approach to synthesize these steroids (Figure 1).
The 9R,11â-halohydrin functionality, which is essential for
biological activity,2-5 is obtained by opening the correspond-
ing 9,11â-epoxide 3 with a suitable acid (HCl or HF).6
Therefore, it is critical to have a robust process to manu-
facture the penultimate intermediate epoxide 3 in high yield
and good quality. We wish to report here the discovery and
development of this new process strategy for â-epoxide 3.
I. 16â-Me-Epoxide 3a. Schering-Plough has a long
history of producing the above-described steroids.7 The
existing manufacturing synthesis1,3,7a for â-epoxide 3a from
Figure 2.
11R-hydroxy 6 is depicted in Scheme 1. Selective protection
of triol 6 with ethyl chloroformate in the presence of pyridine
provided 21-cathylate 7, which was subsequently converted
into 11R-mesylate 8. Elimination of the mesylate was carried
out at 115 °C to provide triene 9. The latter 9,11-double
bond was then reacted with 1,3-dibromo-5,5-dimethyl hy-
dantoin (DBH) to give bromohydrin 10, followed by
cyclization and hydrolysis of the 21-cathylate to form â-
epoxide 3a.
* Corresponding author. Fax: (908) 820-6620. E-mail: xiaoyong.fu@
spcorp.com.
(1) Herzog. H.; Oliveto E. P. Steroids 1992, 57, 617; Hogg, J. A. Steroids 1992,
57, 593; Hirschmann, R. Steroids 1992, 57, 579.
(2) Rausser, R.; Lyncheski, A. M.; Harris, H.; Grocela, R.; Murrill, N.; Bellamy,
E.; Ferchinger, D.; Gebert, W.; Herzog, H. L.; Hershberg, E. B.; Oliveto,
E. P. J. Org. Chem. 1966, 31, 26.
(3) Oliveto, E. P.; Rausser, R.; Herzog, H. L. J. Am. Chem. Soc. 1958, 80,
6687.
(4) Shapiro, E. L.; Gentles, M. J.; Tiberi, R. L.; Popper, T. L.; Berkenkopf, J.;
Lutsky, B.; Watnick, A. S. J. Med. Chem. 1987, 30(9), 1581.
(5) Popper, T. L.; Gentles, M. J.; Kung, T. T.; Berkenkopf, J.; Lutsky, B. N.;
Watnick, A. S.; Shapiro, E. L. J. Steroid Biochem. 1987, 27(4-6), 837.
(6) Graber, R. P.; Aedo, D. M. Span. Pat. ES 529530, 1985.
(7) (a). Kleemann, A.; Engel, J.; Kutscher, B.; Reichert, D. Pharmaceutical
Substances, Syntheses, Patents, Applications, 3rd ed.; Thieme: Stuttgart,
New York 1999. (b). Andrews, D. R.; Giusto, R. A.; Sudhakar, A. R.
Tetrahedron Lett. 1996, 37, 3417. (c). Carruthers, N. I.; Garshasb, S.;
McPhail, A. T. J. Org. Chem. 1992, 57, 961.
This process produced epoxide 3a in about 60% overall
yield with a purity of about 92%. HPLC analysis of the
product revealed four major impurities (Figure 2): 21-OH-
∆
9,11-triene (11, ∼1%), 21-OH-∆11,12-triene (12, ∼1%),
(8) Crabbe, P.; Leon, C. J. Org. Chem. 1970, 35, 2594.
(9) Zehavi, U. J. Org. Chem. 1975, 40(26), 3870.
(10) Breslow, R.; Corcoran, R. J.; Snider, B. B.; Doll, R. J.; Khanna, P. L.;
Kaleya, R. J. Am. Chem. Soc. 1977, 99, 905.
376
•
Vol. 5, No. 4, 2001 / Organic Process Research & Development
10.1021/op0102013 CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 05/10/2001

Table 1. Solvent and temperature effects on the PCl5 dehydration
HPLC area %
solvent
temp (°C)
9 (∆9,11
)
15 (∆11,12
)
16 (11â-Cl)
CH2Cl2/pyridine
pyridine
CH2Cl2
THF
THF
THF
rt
rt
70.4
52.3
74.4
81.0
92.3
95.5
97.5
98.4
80.2
95.7
81.4
95.0
82.8
94.0
11.2
3.8
14.5
7.4
4.9
3.3
1.8
1.3
17.9
4.3
18.6
5.0
18.4
44.0
11.0
11.3
2.8
1.2
0.7
-20
rt
-20
-45
-78
-85
-20
-45
-45
-45
rt
THF
THF
0.3
1.9
EtOAc
tBuOMe
DME
toluene
CH3CN
i-propyl ether
(80% s.m. after 2 h)
(50% s.m. after 4 h)
3.1 (and new impurities)
(85% s.m. after 1 h)
1.8
6.0
-30
21-Cl-∆9,11-triene (13, ∼0.3%) and â-epoxide-21-cathylate
(14, ∼2%). As a penultimate intermediate, this existing
epoxide quality did not meet the minimum 96% purity
specifications for the starting material for drug substances
such as beclomethasone (4). Furthermore, the quality of
betamethasone drug substance needed to be upgraded to meet
the higher EP and USP specifications. Therefore, there was
a strong demand to eliminate or minimize the formation of
those impurities in aiming to increase both the product quality
and the overall chemical yield. Described below is our
systematic approach to achieve the above objectives.
21-OH-∆11,12-Triene (12). It was clear that impurity 12
was carried over from the elimination2,8-10 step (Scheme 1).
The conversion of 8 to 9 produced a considerable amount
of unwanted ∆11,12 isomer (15, ∼15%). It has been well
understood that the problem in the elimination step was
associated with the configuration of carbon C-11. With an
11â-OH,11-15 the base-catalyzed anti E2 elimination provided
the ∆9,11 double bond in greater than 90% yield. Obviously,
with the 11R-OH, the leaving group cannot be anti-coplanar
with the H-9 to form the ∆9,11 double bond exclusively. We
envisioned that this problem could be resolved by a syn-
elimination approach.
Alternatives to POCl3, such as PCl5 were used to optimize
the reaction yield. Shoppee et al.16b reported that reaction of
5R-androstan-11R-ol with PCl5 in CHCl3 at room temperature
gave 5R-androst-9(11)-ene in 65% isolated yield. When
compound 7 was reacted with PCl5 in CH2Cl2/py at room
temperature, HPLC analysis of the reaction mixture indicated
three main peaks with an 86:14 ratio of ∆9,11 to ∆11,12. The
third peak was isolated and identified as 11â-Cl 16 by mass
spectrum and NMR studies.
One can speculate that the ∆9,11 double bond was formed
by the base-catalyzed E2 elimination via the 11â-chloro
intermediate since the latter compound was detected in the
reaction. To minimize the level of 11â-Cl, the reaction
medium (mixture of CH2Cl2 and pyridine) was replaced with
pure pyridine so that all of the 11â-Cl formed would be
converted to 9. On the contrary, an even higher level of 11â-
Cl was obtained. These results clearly suggested that the
elimination did not go through the 11â-Cl intermediate.
Therefore, a better ratio (Table 1) of 9 to 15 might be
obtained as a result of a solvent effect. With this idea in
mind, the PCl5-mediated dehydration was carried out in
various solvents at different temperatures. After various
experiments, THF was found to be the solvent choice, and
-78 to -90 °C was found to be the optimum reaction
temperature (see Table 1 for results of solvent and temper-
ature effects). Under these conditions, ∆9,11 9 was isolated
in greater than 90% yield (vs ∼75% in the existing process)
with a ∆9,11 9 to ∆11,12 15 ratio at about 99:1. A syn
elimination mechanism as shown in Scheme 2 was proposed
for this reaction. The possibility of the formation of a free
carbonium ion reactive intermediate was excluded on the
basis of experimental observations.17 It should be emphasized
that anhydrous conditions are essential to ensure a reproduc-
ible high yield for the elimination, since moisture reacts with
PCl5 or 11R-chlorophosphate to prevent further dehydration.
The triene 9 prepared by the new reaction was converted to
the epoxide 3a. It was no surprise that impurity 12 was only
detected in much less than 0.1% in a purified 3a, since the
level of 15 in the olefin 9 was dramatically reduced.
Bernstein et al.16a have reported that reaction of a 3,17-
bisketal-11R-OH-steroid with phosphorus oxychloride in
pyridine gave ∆9,11 double bond in 90% yield. Under the
Bernstein conditions, ∆9,11 9 was formed from 7 with
excellent regioselectivity, 98:2 by HPLC area to the ∆11,12
15 isomer. However, the reaction yield was less than 15%.
(11) Hazen, G. G.; Rosenburg, D. W. J. Org. Chem. 1964, 29, 1930.
(12) Fried, J.; Florey, K.; Sabo, E. F.; Herz, J. E.; Restivo, A. R.; Borman, A.;
Singer, F. M. J. Am. Chem. Soc. 1955, 77, 4181. Arth, G. E.; Fried, J.;
Johnston, D. B. R.; Hoff, D. R.; Sarett, L. H.; Silber, R. H.; Stoerk, H. C.;
Winter, C. A. J. Am. Chem. Soc. 1958, 80, 3161.
(13) Chamberlin, E. M.; Tristram, E. W.; Utne, J.; Chemerda, J. M. J. Org. Chem.
1960, 25, 295.
(14) Hechter O.; Jacobsen R. P.; Jeanloz R.; Levy H.; Marshall C. W.; Pincus
G.; Schenker V. J. Am. Chem. Soc. 1949, 71, 3261. Collingsworth D. R.;
Brunner M. P.; Haines W. J. J. Am. Chem. Soc. 1952, 74, 2381
(15) Hanson, F. R.; Mann, K. M.; Nielson, E. D.; Anderson, H. V.; Brunner, M.
P.; Karnematt, J. N.; Collingsworth, D. R.; Haines, W. J. J. Am. Chem.
Soc. 1953, 75, 5369.
(16) (a) Bernstein, S.; Lenhard, R. H.; Williams, J. H. J. Org. Chem. 1954, 19,
41. (b) Shoppee, C. W.; Nemorin, J. J. Chem. Soc., Perkin Trans. 1 1973,
542.
(17) Fu, X.; Tann, C. H.; Thiruvengadam, T. K.; Lee, J.; Colon, C. Tetrahedron
Lett. 2001, 42, 2639.
Vol. 5, No. 4, 2001 / Organic Process Research & Development
•
377

Scheme 2. Synthesis of triene (9) by PCl5-mediated
Scheme 4. Selective protection of 21-hydroxyl
elimination of (7)
Scheme 5. Formation of 9,11-dibromo (19)
Scheme 3. 11â-Cl elimination
13 via the same chemistry described in Scheme 1. The
existing 21-cathylation (with pyridine) could generate a
significant amount (>10%) of 11,21-dicathylate 18. To avoid
the formation of the impurity in such a large quantity, the
reaction was quenched when the starting triol 6 was depleted
to about 0.5%. However, quenching the reaction with a
higher level of 6 would lead to a higher level of 13. It was
believed that the necessary base not only acted as an acid
scavenger but also served to further activate the acylation
reagent. The latter function of the base could significantly
affect the selectivity. Therefore, a more sterically hindered
base in place of pyridine would enhance the regioselectivity
of the cathylation. Results from fine-tuning of the reaction
by selecting different bases indicated that triethylamine was
the best choice. When triethylamine was employed as a base,
the reaction went to completion (<0.1% of 6) with less than
0.1% of 11,21-dicathylate (Scheme 4) by HPLC, hence
eliminating the source of impurity 13.
11â-Cl Impurity (17). The epoxide 3a resulting from 9
(Scheme 1) was further converted into betamethasone (1)
via the existing process to compare the impurity profile.
HPLC analysis indicated the presence of a new impurity,
11â-Cl-21-OH (17) arising from 16, present in ∼0.1% in
the final drug substance 1 obtained from the new process.
Since it was difficult to remove 16 by recrystallization of 9,
our efforts were directed to towards converting 16 into 9. It
was interesting to note that when 16 was heated at 80 °C in
a large volume mixture of acetonitrile and water, 9 was
detected as a predominate product. Since water was necessary
for the dehydrochlorination with a possible E1-type mech-
anism, a more polar solvent such as DMSO would push the
elimination at a much faster rate and in a smaller volume.
When 16 was heated at ∼100 °C in DMSO, ∆9,11 double
bond 9 was produced exclusively. Mechanistically, this may
be rationalized as ionization of the 11â-Cl group followed
by a rearrangement of secondary carbonium ion to a tertiary
ion (a similar mechanism was proposed for the dehydration
of 11â-OH steroids under the Burgess conditions8). Elimina-
tion of the 11-H would then provide the desired ∆9,11 function
(Scheme 3). This argument is also supported by the literature
reports that 9R-OH with strong acid18a and 9R-Cl18b with
AgNO3 were regioselectively converted into ∆9,11 respec-
tively.
∆
9,11-Triene-21-OH (11). The impurity ∆9,11-triene-21-
OH 11 could be formed by a simple hydrolysis of unreacted
9,11-triene-21-cathylate 9 in the epoxide formation step.
∆
Careful analysis by HPLC indicated that there was about
2% of impurity 11 present in the crude epoxide 3a. This
level of 11 was much higher than the amount one would
expect from the incompleteness of the bromohydrin reaction
(typically <0.5%). Hence, the alcohol 11 must come from
some other source. This triggered our interest to search for
the origin of this impurity in order to eliminate its formation.
We suspected that there could be a side reaction taking
place at the olefin double bond during the bromohydrin
formation. The side product could then convert back to the
double bond during the NaOH-catalyzed epoxidation step.
To understand the suspected double bond conversions, a
9,11-dibromo byproduct 19 was identified from the bromo-
hydrin crude material. Its structure (Scheme 5) was confirmed
by an independent synthesis of 19 from olefin 9 with
bromine. The impurity from the bromohydrin 10 had an
identical retention time with the synthesized dibromo com-
pound by HPLC analysis. When dibromo 19 was subjected
to the ring closure conditions, ∆9,11-triene-21-OH 11 was
detected as a major product by HPLC although the reaction
was not very clean.
With the new dehydrochlorination process in hand, the
purity of the elimination product can easily be upgraded via
a conversion of 16 into 9 by heating the crude 9 at ∼100 °C
for 1 h in DMSO. HPLC analysis of the drug substance 1
from the DMSO-treated 9 did not show any impurity 17. It
should be noted that this clean up method was occasionally
being employed only when the level of 17 was higher than
0.5%.
â-Epoxide-21-Cl (13). The unprotected 21-OH 6 was
transformed to the corresponding 21-Cl in the existing
mesylation step, which was then subsequently converted into
(18) (a), Beaton, J. M.; Huber, J. E.; Padilla, A. G.; Breuer, M. E. U.S. Pat. No.
4,127,596, 1978. Bergstrom, C. G.; Dodson, R. B. Chem. Ind. 1961, 1530.
(b), Hyatt, J. A.; McCombs, C. A. U.S. Pat. No. 4,328,162. 1982.
The formation of 9,11-dibromo 19 can be rationalized by
what is depicted in Scheme 5. The â-nucleophilic attack of
378
•
Vol. 5, No. 4, 2001 / Organic Process Research & Development

Scheme 6. New preparation of epoxide (3a)
Scheme 7. Process for 16r-Me epoxide (3b)
96% purity specification. In view of an increase in the
requirement of mometasone furoate,22-25 there was a good
incentive to manufacture 3b in house for the production of
mometasone.
water to the bromonium ion produced the desired bromo-
hydrin 10, while the reaction with bromide ion gave the
unwanted dibromide 19. Due to its better nucleophilicity than
water, the bromide can give rise to an impurity such as 11
via 19 at a level that is not purgable by the existing
purification procedure.
Since a trace amount of bromide ion is always present in
the reaction mixture, increasing either the nucleophilicity of
the desired nucleophile or its concentration would favor the
product formation. With this idea in mind, DMF was selected
as the reaction solvent. The bromoformate19-21 formation,
similar to that of “path a”, should be a much more facile
reaction than the bromohydrin formation, since DMF is a
better nucleophile than water and it is also present in much
higher concentration (solvent). There was no surprise that
the 9,11-dibromo impurity was not detected from the
bromoformate reaction mixture by HPLC analysis.
The formation of bromoformate 20 is illustrated in
Scheme 6. Reaction of the ∆9,11 double bond with DBH in
the presence of perchloric acid provides an active bromonium
intermediate. The latter is then trapped by DMF to form an
iminium ion, which can be easily hydrolyzed to generate
the bromoformate during the work up. Bromoformate 20 was
isolated by a direct precipitation of the batch with water.
â-Epoxide-21-cathylate (14). In the existing process,
â-epoxide-21-cathylate (14) was present in epoxide 3a in
about 1 to 2%. This compound 14 was an intermediate in
the formation of 3a. Further addition of NaOH did not
resolve this problem, suggesting that intermediate 14 was
trapped in the crystals of epoxide since the crystallization
of 3a took place prior to the completion of the hydrolysis in
the solvent mixture of THF and MeOH. Replacing the
reaction medium with a better solvent system to maintain a
homogeneous mixture should ensure a complete hydrolysis
of the 21-cathylate. This was achieved by carrying out the
reaction in a mixture of CH2Cl2 and MeOH. HPLC analysis
indicated that there was no 21-cathylate 14 left in the reaction
mixture (<0.1). Epoxide 3a was obtained in 92% overall
yield from 9 with a purity of 99% after recrystallization.
II. 16r-Me Epoxide 3b. Epoxide 3b (16R-Me) is the
starting material for mometasone (2). Historically, this
epoxide was outsourced from a competitor. The existing
Schering epoxide process could not produce 3b to meet the
Application of the new chemistry to 16R-methyl series
would provide an efficient process for epoxide 3b. The
reaction of 16R-methyl triol (21) with ethyl chloroformate
in CH2Cl2 in the presence of TEA provided a regioselective
protection of 21-hydroxyl group. Elimination of 11R-hydroxy
22 with PCl5 was carried out in THF at -85 °C and generated
∆
9,11 23 in about 90% overall yield from 21 with a 99:1 ratio
to ∆11,12 isomer (Scheme 7). The triene 23 was converted
into bromoformate using DBH as reagent in DMF as
described in the case of 3a. Cyclization of the bromoformate
followed by hydrolysis of 21-cathylate yielded the desired
epoxide 3b in a greater than 90% overall yield with an
excellent purity (>98%). The material produced from this
process was transformed into mometasone (2), and no new
impurities were detected by HPLC analysis of the drug
substance.
In conclusion, a significantly improved general process
has been developed for the synthesis of â-epoxide 3 based
on systematic studies of the deficiencies of the existing
process. The overall yield from triol 6 to epoxide 3a was
improved to about 85% from 60%. The purity of the
penultimate intermediate 3a was enhanced to about 99%
from 92% to meet all specifications. The new processes were
implemented in commercial production. The drug substance
betamethasone derived from 3a showed a significant im-
provement in purity to meet all of the updated EP and USP
specifications. A regioselective PCl5-mediated 11R-OH
dehydration process was invented to circumvent the 11R-
mesylate elimination deficiency. A regiospecific dehydro-
chlorination process was also developed to eliminate a minor
new impurity, 11â-chloride (16), so that the new chemistry
could be implemented in the commercial process. Replacing
the existing bromohydrin process with the bromoformate
chemistry not only excluded the formation of dibromide
impurity (19) but also increased the reaction yield. The
(22) Kwok, D. A.; Tsai, D. J. S.; Tann, C. H.; Fu, X. U.S. Pat. No. 5,886,200,
1999.
(23) Draper, R. W.; Hu, B.; McPhail, A. T.; Puar, M. S.; Vater, E. J.; Weber, L.
Tetrahedron 1999, 55, 3355.
(24) Smith C. L.; Kreutner, W. Arzneim.-Forsch. 1998, 48, 956. Chapman, R.
W.; Sehring S. J.; Garlisi, C. G.; Falcone A.; Kung, T. T.; Stelts D.;
Minnicozzi, M.; Jones H.; Umland, S.; Egan, R. W.; Kreutner W. Arzneim.-
Forsch. 1998, 48, 384
(25) Barton, B. E.; Jakway, J. P.; Smith, S. R.; Siegel, M. I. Immunopharmacol.
Immunotoxicol. 1991, 13, 251.
(19) Draper, R. W.; Vater, E. J. U.S. Pat. No. 5,602,248, 1997.
(20) Robinson, C. H.; Finckenor, L.; Kirtley, M.; Gould D.; Oliveto, E. P. J.
Am. Chem. Soc. 1959, 81, 2195.
(21) Devocelle L.; Mackiewicz, P. U.S. Pat. No. 5,310,896, 1994.
Vol. 5, No. 4, 2001 / Organic Process Research & Development
•
379

versatility of the new process was demonstrated by the
successful extension to the 16R-methyl series (3b).
analysis indicated the disappearance of 7, the mixture was
slowly transferred into stirred water (800 mL). The slurry
was filtered and washed with water (600 mL). The wet cake
was dried overnight under vacuum at 60 °C to provide 22.56
g of triene 9 (90.7% purity with a ratio of 98.7:1.3 to 15,
92% overall yield from 6).
Experimental Section
The starting materials triol 6 and 21 as well as all the
reference compounds were provided by Mr. Luis Gil of
Schering Plough Products, Manati, Puerto Rico. All other
chemicals or reagents were purchased commercially. The
ratios of ∆9,11 to ∆11,12 steroids and all of the purity
percentages were determined by HPLC analysis (µ-Bondapak
C-18 column, 1:1 CH3CN/H2O as mobile phase at 1-2 mL/
min, UV detector at 254 nm) of the products or reaction
mixtures. The product impurity profile comparisons were
performed by gradient HPLC analysis (Beckman Ultrasphere
C8 column, CH3CN/MeOH/H2O as solvent system with a
ratio of 15:20:65 and increment of CH3CN by a gradient
table, 2.5 mL/min. at 45 °C column temperature, UV detector
at 254 nm). Molar yields were calculated based on the
starting compound and product purities, as determined by
HPLC using the above conditions. NMR spectroscopic data
were recorded on a Bruker NMR spectrometer.
17r-Hydroxy-16â-methyl-21-ethoxycarbonyloxy-preg-
na-1,4,9(11)-triene-3,20-dione (9). The solid 16â-methyl-
11R,17R,21-trihydroxy-pregna-1,4-diene-3,20-dione (6) (20
g, 97% purity) was suspended in a mixture of CH2Cl2 (80
mL) and triethylamine (30 mL) and cooled to about -15
°C. A solution of ethyl chloroformate (6.5 mL) in CH2Cl2
(10 mL) was added slowly over a period of 1 h, while
maintaining the temperature at about -15 °C. The reaction
mixture was allowed to warm to room temperature, and the
progress was monitored by HPLC analysis. When the
reaction was judged complete (<0.2% of 6, about 2 to 4 h
agitation at about 25 °C), THF (40 mL) and water (80 mL)
were added to the reaction mixture. The pH of the mixture
was adjusted to below 2 by careful addition of concentrated
HCl (13 mL) at temperature below 25 °C. The resulting two
layers were separated, and the aqueous layer was extracted
with CH2Cl2 (50 mL). The combined organic layers were
concentrated, and the residual CH2Cl2 was replaced by
azeotropic distillation with THF (KF < 200 ppm). The
resulting THF solution was directly used for the dehydration
step without further purification.
The HPLC retention time (∼10.5 min) and NMR spectra
1
were identical with a reference compound. H NMR (400
MHz, CDCl3): 7.20 (d, J ) 10.2 Hz, 1H), 6.31 (dd, J )
10.2, 1.9 Hz, 1H), 6.10 (s, 1H), 5.59 (m, 1H), 4.99 (d, J )
17.9 Hz, 1H), 4.92 (d, J ) 17.9 Hz, 1H), 4.26 (q, J ) 7.1
Hz, 1H), 2.69 (m, 1H), 2.58 (m, 1H), 2.45 (m, 1H), 2.25
(m, 3H), 2.12 (m, 1H), 1.82 (m, 1H), 1.6 (m), 1.43 (s, 3H),
1.36 (t, J ) 7.1 Hz, 3H), 1.27 (m, 2H), 1.21 (d, J ) 7.3 Hz,
3H), 1.19 (m, 1H), 0.85 (s, 3H). 13C NMR (62.896 MHz,
CDCl3): 204.5, 186.5, 166.9, 154.8, 141.9, 127.2, 123.7,
120.8, 89.3, 71.6, 64.5, 49.6, 48.7, 47.0, 45.9, 36.4, 36.1,
35.0, 33.0, 32.2, 26.5, 19.6, 14.3, 14.2.
17r-Hydroxy-16r-methyl-21-ethoxycarbonyloxy-preg-
na-1,4,9(11)-triene-3,20-dione (23). Following essentially
the same procedures as described for 9, 16R-methyl triene
(23) was prepared from 16R-methyl triol (21) via 21-
cathylate (22) in an overall yield of 94%. The ratio of ∆9,11
to ∆11,12 was 99:1 by HPLC analysis.
22: 1H NMR (400 MHz, CDCl3): 7.76 (d, J ) 10.3 Hz,
1H), 6.30 (dd, J ) 10.3, 1.9 Hz, 1H), 6.07 (s, 1H), 4.99 (s,
2H), 4.23 (q, J ) 7.1 Hz, 2H), 4.06 (dt, J ) 10.3, 5.1 Hz,
1H), 2.50 (dt, J ) 13.2, 4.6 Hz, 1H), 2.38 (m, 1H), 2.22 (m,
1H), 2.01 (m, 2H), 1.92 (dd, J ) 12.0, 5.1 Hz, 1H), 1.75
(m, 2H), 1.58 (m, 1H), 1.34 (t, J ) 7.1 Hz, 3H), 1.30 (s,
3H), 1.13 (d, J ) 7.3 Hz, 3H), 1.02 (m, 3H), 0.89 (s, 3H).
13C NMR (62.896 MHz, CDCl3): 205.4, 187.8, 169.8, 160.6,
155.4, 124.9, 124.5, 89.3, 72.3, 68.2, 64.9, 60.7, 50.3, 49.5,
48.2, 44.6, 43.0, 35.5, 34.4, 34.3, 33.5, 20.0, 19.0, 16.1, 14.6.
23: 1H NMR (400 MHz, CDCl3): 7.18 (d, J ) 10.1 Hz,
1H), 6.30 (dd, J ) 10.1, 1.7 Hz, 1H), 6.08 (s 1H), 5.56 (br.
d, J ) 5.7 Hz, 1H), 5.03 (d, J ) 17.7 Hz, 1H), 4.84 (d, J )
17.7 Hz, 1H), 4.25 (q, J ) 7.1 Hz, 2H), 3.13 (m, 1H), 2.66
(m, 2H), 2.43 (m, 1H), 2.31 (m, 1H), 2.14 (m, 1H), 1.81
(m, 3H), 1.46 (m, 1H), 1.42 (s, 3H), 1.36 (t, J ) 7.1 Hz,
3H), 1.21 (m, 1H), 0.95 (d, J ) 7.2 Hz, 3H), 0.79 (s 3H).
13C NMR (62.896 MHz, CDCl3): 205.2, 186.8, 167.4, 155.4,
155.2, 142.9, 127.6, 124.2, 121.1, 91.4, 71.0, 65.0, 48.5, 48.1,
46.3, 37.2, 35.2, 33.4, 32.8, 32.6, 27.1, 14.9, 14.8, 14.6.
11â-Chloro-17r-hydroxy-16â-methyl-21-ethoxycar-
bonyloxy-pregna-1,4-diene-3,20-dione (16). The starting
material (7, 2 g) was dissolved in a mixture of CH2Cl2 (30
mL) and pyridine (6 mL) at room temperature. To the stirring
solution, PCl5 (2 g) was added portionwise over a period of
15 min. The resulting mixture was stirred for an additional
30 min at room temperature, quenched with water (20 mL)
and then diluted with CH2Cl2 (100 mL). The mixture was
washed with aqueous HCl solution (6 N, 50 mL). The
aqueous layer was extracted with CH2Cl2 (2 × 50 mL). The
combined organic layers were washed with water, brine, and
dried over anhydrous Na2SO4. The solvent was removed
under vacuum to provide a residue of 16 (0.15 g), which
was purified by silica gel chromatography (30/70 then 50/
An analytical sample of 21-cathylate 7 was prepared by
removing solvents under vacuum and recrystallizing 7 from
a mixture of CH2Cl2 and THF. The HPLC retention time
(∼4.5 min) and NMR spectrum were identical with that of
1
a reference compound. H NMR (400 MHz, CDCl3): 7.76
(d, J ) 10.3 Hz, 1H), 6.20 (dd, J ) 10.3, 1.9 Hz, 1H), 6.10
(s, 1H), 4.99 (s, 2H), 4.22 (q, J ) 7.1 Hz, 2H), 4.09 (dt,
J ) 10.4, 5.1 Hz, 1H), 2.50 (m, 1H), 2.40 (m, 1H), 2.20 (m,
1H), 2.05 (m, 2H), 1.92 (m, 1H), 1.75 (m, 2H), 1.57(m, 1H),
1.35 (t, J ) 7.2 Hz, 3H), 1.32 (s, 3H), 1.15 (d, J ) 7.3 Hz,
3H), 1.06 (m, 2H), 0.90 (s, 3H).
The reaction volume was adjusted to about 150 mL with
THF and cooled to about -85 °C. Phosphorus pentachloride
(PCl5, 20 g) was charged slowly over a 30 min period while
keeping the reaction temperature below -83 °C. The mixture
was agitated at about -85 °C for about 1 h. When HPLC
380
•
Vol. 5, No. 4, 2001 / Organic Process Research & Development

50 EtOAc/hexane). Mass spectrum (FAB): 467 (M+ + 3),
The resulting slurry was cooled to about 0 °C and filtered.
The wet cake was washed with water and dried under
vacuum at about 50 °C to provide 41.2 g (purity 90.0%
against standard) of the crude epoxide 3a (97% overall yield
from 9).
The crude epoxide 3a was heated in a mixture of CH2Cl2
(700 mL) and MeOH (200 mL) until dissolved. The resulting
solution was filtered and concentrated to a volume of
about 160 mL. The resulting slurry was cooled to about
0 °C, filtered, and washed with cold MeOH. The wet cake
was dried under vacuum at about 50 °C to provide 36.0 g
of purified epoxide 3a (99% purity, 93% overall yield
from 9).
1
465 (M+ + 1), and 429 (M+ + 1 - HCl). H NMR (400
MHz, CDCl3): 7.25 (d, J ) 10.0 Hz, 1H), 6.30 (dd, J )
10.0, 1.7 Hz, 1H), 5.99 (s, 1H), 5.00 (d, J ) 17.8 Hz, 1H),
4.91 (d, J ) 17.8 Hz, 1H), 4.66 (br s, 1H), 4.24 (q, J ) 7.1
Hz, 2H), 2.58 (dt, J ) 13.4, 4.5 Hz, 1H), 2.40 (dd, J ) 14.5,
4.7 Hz, 1H), 2.34 (dd, J ) 13.1, 3.6 Hz, 1H), 2.10 (m, 5H),
1.55 (m, 1H), 1.52 (s, 3H), 1.38 (dd, J ) 10.8, 3.7 Hz, 1H),
1.34 (t, J ) 7.1 Hz, 3H), 1.23 (m, 2H), 1.18 (d, J ) 6.6 Hz,
3H), 1.16 (s, 3H). 13C NMR (100.6 MHz, CDCl3): 204.8,
186.8, 170.4, 155.6, 128.9, 122.4, 89.3, 72.1, 64.9, 59.8, 56.6,
51.1, 49.7, 49.4, 44.5, 40.9, 35.3, 34.4, 32.0, 30.9, 21.0, 20.2,
18.2, 14.6.
Dehydrochlorination of 11â-chloro-21-cathylate (16).
11â-Chloro-21-cathylate (16, 95 mg) was dissolved in
DMSO (2 mL). The solution was heated to 90 °C for about
3 h. The reaction progress was monitored by HPLC analysis.
The starting compound (retention time ∼15.5 min.) was
gradually converted into the triene product (retention time
∼10.5 min.) over the course of the reaction. The triene 9
was formed in greater than 97% yield as determined by
HPLC. Essentially no ∆11,12 was detected.
HPLC retention time and 1H NMR of 3a are identical to
those of a reference compound. 1H NMR (400 MHz, DMSO-
d6): 6.62 (d, J ) 10 Hz, 1H), 6.10 (br s, 1H), 6.09 (dd, J )
10, 1.8 Hz, 1H), 5.26 (s, 1H), 4.52 (t, J ) 5.8 Hz, 1H), 4.37
(dd, J ) 19.4, 5.9 Hz, 1H), 4.10 (dd, J ) 19.3, 5.8 Hz, 1H),
3.19 (br s, 1H), 2.67 (m, 1H), 2.45 (m, 2H), 2.36 (m, 2H),
2.21 (m, 2H), 2.05 (m, 2H), 1.59 (m, 1H), 1.52 (m, 1H),
1.37 (s, 3H), 1.33 (m 1H), 1.01 (d, J ) 6.6 Hz, 3H) and
0.86 (s, 3H).
9â,11â-epoxy-17r,21-Dihydroxy-16â-methylpregna-
1,4-diene-3,20-dione (3a). The crude 16â-methyl-triene-21-
cathylate 9 (50 g, 90% purity) was dissolved in DMF (175
mL) at room temperature. The resulting solution was cooled
to about 10 °C, and then 70% HClO4 (6.25 mL) was slowly
added. DBH (25.6 g) was slowly added to the mixture over
a period of about a 15 min while keeping the reaction
temperature below 20 °C. The reaction progress was
monitored by HPLC analysis. When HPLC indicated that
the reaction was complete (about 3 h), the mixture was
diluted with MeOH (175 mL). The bromoformate 20 was
precipitated in 1500 mL of water, and cooled to below 10
°C. The product was isolated by filtration and washed with
water (about 1000 mL). The wet cake was used directly in
the next step without further purification.
9r,11â-Dibromo-17r-hydroxy-16â-methyl-21-ethoxy-
carbonyloxy-pregna-1,4-diene-3,20-dione (19). 16â-Meth-
yl-triene (9, 2 g) was dissolved in CHCl3 (20 mL) and cooled
to 0 to 5 °C with an ice bath. Bromine (0.45 mL, 2 eq) was
added slowly. The reaction was monitored by HPLC analysis
for completion. The reaction was quenched with aqueous
Na2SO3. The organic layer was separated and washed with
water, dried over anhydrous Na2SO4. The solvent was
removed under vacuum to provide the crude dibromide 19
(greater than 92% by HPLC analysis), which was directly
subjected to the next reaction. MS (FAB): 591 (M+ + 5),
1
589 (M+ + 3), 587 (M+ + 1), 429 (M+ + 1 - 2Br). H
NMR (250.1 MHz, CDCl3): 7.3 (d, 1H), 7.4 (d, 1H), 6.0 (s,
1H), 5.1 (m, 1H), 4.9 (dd, 2H), 4.2 (q, 2H), 3.4 (m, 1H), 2.4
(m, 5H), 2.0 (m, 6H), 1.8 (s, 3H), 1.7 (m), 1.3 (t, 3H), 1.2
(s, 3H), 1.15 (d, 3H). 13C NMR (62.896 MHz, CDCl3):
204.0, 186.0, 166.0, 154.8, 151.2, 129.8, 124.2, 88.9, 87.1,
77.2, 71.6, 64.5, 50.9, 50.5, 49.3, 45.3, 38.7, 35.1, 34.1, 30.0,
29.4, 25.5, 19.7, 18.2, 14.2.
The 9,11-dibromide 9 was dissolved in a mixture of THF
(15 mL), MeOH (15 mL) and water (2 mL). The solution
was cooled to about 3 °C with an ice bath. A chilled solution
of NaOH (0.85 g) in water (7 mL) was slowly added. The
mixture was agitated overnight while the reaction temperature
was allowed to warm to room temperature. HPLC analysis
indicated that the major peak (53%) had the same retention
time as a reference sample of 11.
A small sample of the wet cake was taken and dried under
1
vacuum for NMR analysis. H NMR (400 MHz, CDCl3):
8.12 (s, 1H), 6.85 (d, J ) 10 Hz, 1H), 6.34 (dd, J ) 10, 1.7
Hz, 1H), 6.10 (d, J ) 1.6 Hz, 1H), 5.90 (br s, 1H), 5.02 (d,
J ) 18 Hz, 1H), 4.86 (d, J ) 18 Hz, 1H), 4.22 (q, J ) 7 Hz,
2H), 2.89 (dd, J ) 14.8, 3.6 Hz, 1H), 2.60 (m, 1H), 2.40 (m
2H), 2.15 (m 3H), 1.95 (m, 1H), 1.75 (m, 2H), 1.58 (s 3H),
1.33 (t, J ) 7 Hz, 3H), 1.25 (m, 1H), 1.19 (d, J ) 7 Hz, 3H)
and 1.00 (s, 3H). 13C NMR (100.6 MHz, CDCl3): 204.8,
186.3, 165.4, 159.6, 155.2, 151.5, 130.1, 125.6, 89.2, 82.2,
75.1, 72.0, 64.9, 49.8, 49.7, 49.5, 44.6, 35.9, 34.5, 33.8, 31.0,
29.1, 25.1, 20.1, 17.9, 14.6.
The bromoformate 20 wet cake was dissolved in a mixture
of CH2Cl2 (350 mL) and MeOH (325 mL). The resulting
mixture was cooled to about -5 °C, vacuum degassed three
times under nitrogen. A chilled solution of NaOH (15 g) in
water (15 mL) was added slowly over about 1 h period while
maintaining the reaction temperature between -5 and 0 °C.
The resulting mixture was stirred for about 2 h. The reaction
was quenched with acetic acid (40 mL). The solvents were
removed by distillation and replaced with water (500 mL).
9â,11â-Epoxy-17,21-dihydroxy-16r-methylpregna-1,4-
diene-3,20-dione (3b). The 16R-methyl-triene-21-cathylate
23 (50 g, 86.9% purity) was suspended in DMF (175 mL)
at room temperature. Then 70% HClO4 (6.5 mL) was added
in the mixture, which was followed by addition of DBH (25
g) at room temperature over about a 10 min time period.
The resulting mixture was agitated for about 1 h. MeOH
(150 mL) was charged, and the resulting solution was slowly
transferred into a water solution (4000 mL) that contained
Vol. 5, No. 4, 2001 / Organic Process Research & Development
•
381

MeOH (500 mL) and Na2SO3 (6 g). The slurry was cooled
to about 5 °C with agitation. The bromoformate 24 was
filtered and washed with water.
MHz, DMSO-d6): 6.47 (d, J ) 10 Hz, 1H), 5.96 (br s, 1H),
5.93 (dd, J ) 10, 1.7 Hz, 1H), 4.89 (s 1H), 4.54 (t, J ) 5.9
Hz, 1H), 4.31 (dd, J ) 19.3, 6.4 Hz, 1H), 3.88 (dd, J )
19.3, 6.4 Hz, 1H), 3.05 (br s, 1H), 2.74 (m, 1H), 2.51 (m,
1H), 2.32 (m, 1H), 2.10 (m, 3H), 1.61 (m, 1H), 1.47 (m,
1H), 1.34 (m, 1H), 1.22 (s, 3H), 1.14 (m, 1H), 1.05 (m, 1H),
0.60 (d, J ) 6.8 Hz, 3H), 0.59 (s, 3H).
The bromoformate 24 wet cake was dissolved in a mixture
of CH2Cl2 (300 mL) and MeOH (25 mL). The water layer
was separated and extracted with CH2Cl2 (100 mL). The
organic layers were combined with MeOH (100 mL). The
chilled and degassed solution was reacted with NaOH (15
g) solution in water (30 mL) following the same procedure
as described for 3a. The reaction was quenched with an
aqueous acetic acid solution (20 mL, 50%). The two layers
were separated, and the aqueous layer was extracted with
CH2Cl2 (2 × 125 mL). The combined organic layers were
concentrated to remove CH2Cl2 and replaced with MeOH
to a final volume of 150 mL. The slurry obtained was cooled
to about -2 °C, filtered, and washed sequentially with cold
MeOH (30 mL) and a mixture of MeOH and water. The
wet cake was dried under vacuum at 60 °C to yield 35.5 g
epoxide 3b with a purity of 99.2% (93.3% overall yield
from 24).
Acknowledgment
We thank Dr. Derek Walker for his helpful discussions
and support, Mr. Luis Gil for providing all of the reference
samples as well as converting the epoxides to betamethasone
and mometasone for impurity profile comparison. We are
also indebted to Mr. Danny Angeles for experimental
assistance, Dr. Birendra Pramanik for providing mass spectra,
and Dr. Tze-Ming Chan for NMR analysis.
Received for review January 4, 2001.
OP0102013
1
The HPLC retention time and H NMR spectrum are
identical with those of a reference compound. 1H NMR (400
382
•
Vol. 5, No. 4, 2001 / Organic Process Research & Development