Serendipitous discovery of a zidovudine guanidine complex: A superior process for the production of zidovudine

Article abstract of DOI:10.1021/op2000805

A superior process for the commercial production of zidovudine (AZT) has been developed. It was discovered that an AZT-guanidine complex formed when a crude zidovudine solution was treated with guanidine. This readily precipitated from protic solvents resulting in the exclusion of impurities and permitted the development of a superior isolation and purification of AZT.

Full text of DOI:10.1021/op2000805

ARTICLE
pubs.acs.org/OPRD
Serendipitous Discovery of a Zidovudine Guanidine Complex:
A Superior Process for the Production of Zidovudine
Bruno K Radatus
Apotex Pharmachem, Inc., 34 Spalding Drive, Brantford, Ontario, N3T 6B8, Canada
S Supporting Information
b
ABSTRACT: A superior process for the commercial production of zidovudine (AZT) has been developed. It was discovered that an
AZTꢀguanidine complex formed when a crude zidovudine solution was treated with guanidine. This readily precipitated from
protic solvents resulting in the exclusion of impurities and permitted the development of a superior isolation and purification
of AZT.
’
INTRODUCTION
Zidovudine (AZT), first synthesized by Horwitz et al. in
964, was the first antiviral agent approved for the clinical treat-
alkaline phase was neutralized, extracted with MIBK, and crystal-
lized to give crude AZT 4 in an initial average yield of 67% and
eventually optimized to a yield of 81%. The crude AZT 4 was
recrystallized from aq acetone to >99.5% purity AZT 4 eventually
in 73% optimized recovery. The final optimized yield for pure
AZT 4 from thymidine was 50%.
1
1
2
ment of HIV infections. In their procedure, thymidine was
tritylated, mesylated, and reacted with excess sodium hydro-
xide to give, directly or via the corresponding 2,3 -anhydro
intermediate (TCT 2), 5 -O-trityl-threo-thymidine 5,
was mesylated, reacted with lithium azide, and deprotected
with hydrochloric acid to give AZT 4 in an overall yield of
3
,4
4
0
0
4ꢀ6
The process was efficient from an operational point of
view but it produced 3ꢀ7% of a major dimer side product
1
1
which1
0 00 000 000 00 0
[3 -N -(3 -azido-3 -deoxythymid-3 -yl)-3 -deoxythymidine]
9a,b
7
1
7
(Figure 1).
Two other major side products were also pro-
3
5.7% (Scheme 1). Glinski et al. reported a similar but shorter
0
duced: thymine 8 [3ꢀ7%] which is a degradation product
procedure (Scheme 1) using sodium azide and 5 -O-trityl-2,
0
0
0
0
of the thymidine derivatives present in the reaction mixture
3
-anhydrothymidine (TCT 2), to give 3 -azido-3 -deoxy-5 -O-
0
0
0
and 3-N-AZT 9 [2ꢀ5%, 3-(3 -azido-2 ,3 -dideoxy-β-D-erythro-
tritylthymidine (Trityl-AZT 3) which was deprotected with acetic
acid to AZT 4 in an overall yield of 17ꢀ18% from thymidine.
Although the yield was only about one-half of Horwitz’s process,
it was more efficient in terms of unit process operations, chemical
steps and the amounts of reagents used.
Our department was charged with designing a superior synth-
esis of AZT with the object of eventually establishing a lower-cost
large-scale manufacture of this important drug. After reviewing
pentofuranosyl)thymine] which is a degeneration and regenera-
1
0
tion product at the N3 from trityl AZT 3, however 7 was the
most difficult to separate, requiring large volumes of acetone
during the final recrystallization step. This also impacted the %
recovery.
To better understand the formation of this major impurity,
the dimer 7 was isolated by chromatography, and the structure
1
13
8
was confirmed by H NMR and C NMR spectral analysis,
elemental analysis, IR, UV and mass spectrometry. The proposed
mechanism for the formation of the 7 is shown in Scheme 2. Also
the knowledge of knowing the structure of 7 allowed for an early
improvement in its reduction from ∼7% to ∼4% by increasing
the amount of sodium azide from 1.5 to 2.5 equiv.
a number of other procedures, I decided to follow the approach
of Glinski in starting with TCT 2 for the introduction of the
azido group.
7
’
RESULTS AND DISCUSSION
TCT 2 reacts with sodium azide which may give the salt
sodium trityl AZT 10, which then attacks another molecule of
TCT 2 to furnish the salt sodium ditrityl dimer 11 which upon
deprotection gives 7. In Scheme 2 the sodium salts 10 and 11 are
shown with the negative charges on the C2-oxygens rather than
nitrogen because oxygen is more electronegative than nitrogen.
It is also possible to put the negative charge on the C4-oxygen,
The preparation of TCT 2 was improved to an average yield of
8
5% starting from thymidine by an approach which was similar
5
to the one developed by Horwitz. The azidation reaction was
modeled on Glinski’s approach in that the starting material was
TCT 2, but the solvent was changed to dimethyl sulfoxide
(
DMSO) because DMF gave blackish solutions and the resultant
AZT 4 was difficult to crystallize. Initially the amount of sodium
azide was reduced from 3.7 equiv to 1.5, but after an impurity,
which is discussed below, had been identified, the amount was
increased to 2.5 equiv, and the impurity was reduced from ∼7%
to ∼4%, the deprotection was done with aq HCl in methyliso-
butyl ketone (MIBK) instead of 80% acetic acid, and the AZT 4
was separated from the trityl alcohol by rendering the mixture
alkaline, driving sodium AZT 4 into the aqueous phase. The
13
but the C NMR chemical shifts of sodium trityl AZT 10 for
C2 at δ 157.84 ppm and C4 at δ 173.79 ppm fit very closely
11
to the simulated spectral values at 159.62 and 173.75, respec-
tively, when the negative charge is placed on the C2-oxygen.
Received: March 29, 2011
Published: August 27, 2011
r 2011 American Chemical Society
1281
dx.doi.org/10.1021/op2000805
|Org. Process Res. Dev. 2011, 15, 1281–1286

Organic Process Research & Development
ARTICLE
Scheme 1. Processes by Glinski (a) and Horwitz (b) for the preparation AZT 4
modify 7, thereby forming products which may be more readily
separated from crude AZT 4. A number of ultimately unsuccess-
ful approaches were studied, including degradation by either acid
or base, and complexation or salt formation. Derivatization was
not considered since this would entail at least one additional
1
2,13
chemical step. However, the literature
approach in that of uracil, 1-methyluracil, 3-methyluracil and 1,
-dimethyluracil, only 1,3-dimethyluracil would react with gua-
suggested a different
3
1
2
nidine (Scheme 3 based on description given in reference ).
The explanation for the nonreactivity of the other three
compounds (see ref 12) was that in the presence of a strong
base such as guanidine the presence of acidic protons would
deprotonate the bases, rendering nucleophilic attack more
12
difficult or, in other words, an electron-rich nucleophile such
as guanidine would be repulsed by an electron-rich pyrimidine
anion. An examination of the structures of AZT 4 and 7 revealed
that, for the former, the base moiety was monoalkylated and
would be predicted, according to the above explanation, not to be
susceptible to attack by guanidine because the nucleo-base would
become anionic; for 7 the middle base unit was dialkylated and
thus would be predicted to be susceptible to attack by guanidine
because the nucleo-base would remain neutral. A possible out-
come is shown in Scheme 4. The resulting urea or further
degradation products may be more easily removed from the
product mixture than 7. This analysis provided sufficient justifi-
cation for us to examine this approach.
Some dimer 7 was combined with a methanolic solution of
guanidine, evaporated to a thin syrup, and heated. After 2.5 h,
TLC showed only a trace of 7. This result seemed to support the
proposed mechanism of Scheme 4. In another experiment, AZT
4 syrup containing 83.2 HPLC area % of AZT and 9.38% of 7,
was dissolved in ethanolic guanidine at reflux. After 29 h, it
showed little if any change by TLC. However, when the mixture
was evaporated to give a syrup, heated at 80 °C for 6.5 h to
replicate fusion conditions, and allowed to cool, it had solidified
by the next day. In spite of further manipulation by adding more
Figure 1. Major impurities in the production of AZT 4.
The corresponding simulated values for when the negative
charge is placed on the C4-oxygen and N3 are 161.99 and
1
77.60 and 163.00 and 180.00, respectively. Chemical support
for this mechanism comes from the reaction between pure trityl
AZT 3 and TCT 2 in DMSO in the presence of potassium
carbonate at 108 °C for 24 h. The reaction mixture was worked
up, treated with acid, and worked up to a syrup. HPLC of
the syrup indicated 7 (27 area %), thymine (16%), and AZT 4
(
31%). The syrup was purified to furnish 7 in 8.6% yield.
The separation of 7 from AZT 4 could be achieved by
crystallization using a number of different solvent combinations
but with a significant loss of yield. A number of process variations
before crystallization were tried, such as acid/base liquid/liquid
extractions that also proved to be unsuccessful in removing this
impurity. In a search to find means to manage the problem posed
by this side product in the reaction and its propensity to remain
with the product AZT 4, the focus was shifted to finding a way to
1
282
dx.doi.org/10.1021/op2000805 |Org. Process Res. Dev. 2011, 15, 1281–1286

Organic Process Research & Development
ARTICLE
Scheme 2. Possible mechanism for the formation of dimer 7 during AZT 4 production
Scheme 3. Mechanism for the modification of dimethyl uracil upon treatment with guanidine
0
Scheme 4. Expected outcome of the treatment of dimer 7
with guanidine
singlet integrating for seven protons between the H-1 and H-6
signals of AZT 4. I believed that AZT 4 had formed a salt or
complex with guanidine that, unlike the salt AZT sodium 19,
precipitated readily from the solvent. If AZT guanidine 20 is a
salt, then the question arises as to where the negative charge
should is located. The negative charge could be on the C2ꢀO or
N3 or C4ꢀO, and C2ꢀO is chosen for the same reasons given
13
earlier for the case of the C NMR downfield shifts of the C2 and
C4 resonance when going from trityl AZT 3 to sodium trityl AZT
10. The C2 and C4 resonances of 150.42 and 163.73, respec-
tively, for AZT 4 shift downfield to 157.86 and 173.96, respec-
tively, for AZT guanidine 20 which is in line with the simulated
11
spectra where the corresponding values are 150.33 and 163.33
for AZT 4 going to 159.62 and 173.75 ppm for AZT guanidine
13
2
0. Also the actual C NMR of sodium AZT 19 shows downfield
shifts to 157.82 and 173.68 ppm, respectively, which are very
close to the actual shifts for AZT guanidine 20. The discovery of
this AZT guanidine salt 20 was very exciting and changed the
course of the study from base degradation to salt formation.
These initial studies demonstrated that the content of 7 could
be lowered from 19% to 7% along with a 75% recovery of AZT.
However, I wished to determine if such purification could be
achieved at an earlier stage in the process. An attractive point
would be when AZT 4 had been separated from the trityl alcohol
by the water/MIBK extraction, which would avoid the additional
steps of neutralization, extraction, and concentration. Solid
guanidine hydrochloride was added to basic (pH ∼12) AZT 4
in aqueous solution, and the pH was adjusted to 12.2ꢀ12.3 with
guanidine and exchanging the solvent to other alcohols such as
methanol or isopropanol, the solid could not be dissolved.
Interestingly, much solid still remained even at reflux, an
observation which was surprising since it was known that the
solvents should have dissolved all of the available AZT 4 at reflux.
A white solid was obtained after cooling and washing with
1
isopropanol whose H NMR spectrum showed the characteristic
AZT 4 peaks, although slightly shifted, and a broad exchangeable
1
283
dx.doi.org/10.1021/op2000805 |Org. Process Res. Dev. 2011, 15, 1281–1286

Organic Process Research & Development
ARTICLE
a
Scheme 5. Improved process for the production of AZT 4
been optimized, the yield of the first crop of pure AZT 4 was
∼
55%, and the second crop increased the overall yield to ∼62%.
’
CONCLUSION
The commercial process for the production of AZT 4 has been
improved by the discovery that AZT 4 forms a salt with guanidine
which precipitates from aqueous solutions quite readily while an
otherwise difficult-to-remove dimer remained in solution. The
AZT guanidine salt 20 could be easily converted to crude AZT 4
in high recovery. This development significantly improves the
purification of AZT and is currently used in our company’s
commercial production of AZT.
’
EXPERIMENTAL SECTION
The Initial Commercial Process for the Production of
0
0
3
-Azido-3 -deoxythymidine (AZT) 4. Step A: Preparation of
Crude AZT 4. Tritylcyclothymidine (TCT) 2 (163 kg, 349.4 mol)
was reacted with sodium azide (56 kg, 862.5 mol, 2.46 equiv) in
DMSO (410 kg) at ∼110 °C for 24 h and was cooled to about
6
0 °C; MIBK (647 kg) preheated to ∼60 °C was added, followed
a
Reagents and conditions: a) MIBK, 77 °C, 23% aq HCl, then RT, aq
NaOH, pH 12, separate; b) aqueous phase, 70 °C, solid guanidine HCl,
by brine (195 kg) preheated to ∼60 °C and water (195 kg)
preheated to ∼60 °C (preheating prevented the formation
of taffy-like yellow globs); the phases were separated. Water
aq NaOH, pH 12, cool, filter 78ꢀ82%; c) H
2
O, HCl, or AcOH, 75 °C,
then 20, cool, filter, 81ꢀ85%; d) H O, acetone, carbon, filter, partly
2
(
130 kg) and brine (130 kg) were added to the MIBK phase,
distill acetone at atm, 35 °C, seeds, 23 °C, 12 h, vacuum distill acetone,
5
°C, filter, dry, 88ꢀ92%.
cooled to about 35 °C, and separated. The combined aqueous
phases were further extracted with MIBK (2 ꢁ 107 kg), and the
combined MIBK phases were washed once with brine (195 kg)
to give trityl AZT 3 as an MIBK solution (1230 L). The MIBK
solution was heated to 77 °C, and 23% hydrochloric acid
(56.4 kg) was added. The reaction was monitored by TLC
(ethyl acetate/hexane, 3:2), and after 1.5ꢀ2.5 h there was only
a trace of trityl AZT 3 remaining. Heating was stopped, water
(225 kg) was added, and the pH was adjusted to 12.05 with 50%
NaOH (∼40 kg). The phases were separated, and the MIBK
phase was extracted twice with water (64 and 33 kg). The
combined aqueous phases were diluted with MIBK (445 kg),
and the pH was adjusted to 7.02 with 16% hydrochloric acid
(87.6 kg). The phases were separated, and the aqueous phase was
vacuum-distilled until 330 L of distillate was collected and
extracted with MIBK (multiple times with a total of 330 kg).
Celite (4.7 kg) was added to the combined MIBK phases and
filtered. The MIBK filtrate was vacuum concentrated until about
765 L of distillate was collected, cooled to 25 °C, seeded, and
maintained for 6 h. The mixture was further cooled to 5 °C and
maintained for 24 h. The crystals were filtered, washed with
MIBK (65 kg), and dried to give 78.6 kg (294.1 mol, 84.2%) of
crude AZT 4, (HPLC purity = 94.3 area %, dimer 7 content =
4.5%).
Step B: Preparation of Pure AZT 4. Crude AZT 4 (290 kg,
HPLC purity = 93.8%, dimer = 4.7%) was dissolved in a mixture
of water (870 kg) and acetone (540 kg) at 50 °C. Active carbon
(7.25 kg) and Celite (20 kg) were added, the mixture was filtered
over a Celite bed at 30 °C, and the bed was washed with
additional acetone (3 ꢁ 45 kg). Acetone (∼380 L) was distilled
at atmospheric pressure, and a sample was checked by NMR to
determine the water/acetone ratio which was 2.2 and which was
in the acceptable range (2.1ꢀ2.5). The mixture was cooled to
35 °C, seeded with pure AZT 4, and agitated for 12 h. The
mixture was further cooled to 4 °C and maintained for 1 h; a
sample was filtered and checked by HPLC for the dimer content
5
0% NaOH. AZT guanidine salt 20 readily precipitated from
solution with a dimer 7 content of usually <1%. The modification
achieved the desired purity result, eliminating three steps and
avoided the need to separately prepare guanidine in alcohol from
guanidine hydrochloride even though it introduced an additional
step of releasing AZT 4 from the salt.
Releasing the AZT 4 from the guanidine salt was achieved
by adding the salt to hot water containing one equivalent of
hydrochloric acid or acetic acid, whereupon the precipitate
dissolved and then another precipitate formed which was AZT
4
largely free of guanidine and dimer 7 (typically <1%). The
resulting crude AZT 4 could then be purified by recrystallization
in ∼90% recovery from water/acetone as previously described
with an additional operation of vacuum distilling most of the
acetone. This sequence of precipitating the AZT guanidine salt
2
0 was incorporated into the existing process and became the
9a
new commercial process (Scheme 5).
This new technology ultimately had a number of advantages
over that of the original process. First, the AZT 4 is initially
separated from the reaction mixture as AZT-guanidine 20 by
precipitation, thus avoiding the tedious, highly dilute, and time-
consuming liquid/liquid extraction, followed by concentration
to obtain crystals. Second, the resultant AZT-guanidine 20 had
a low dimer 7 content relative to the original formulation of this
process which also led to crude AZT 4 of low dimer 7 content.
Third, the purer crude AZT 4 allowed a more facile removal of
the acetone during the final purification step leading to an
improved recovery. The fourth advantage was that the water-
damp AZT guanidine 20 salt and the crude crystalline AZT 4 did
not have to be dried before going to the next step. Finally, the
facile precipitation of AZT guanidine 20 allowed the recovery of
a second crop of AZT 4 from the mother liquors in spite of the
high levels of dimer 7 present. Eventually, once the process had
1
284
dx.doi.org/10.1021/op2000805 |Org. Process Res. Dev. 2011, 15, 1281–1286

Organic Process Research & Development
ARTICLE
which was 0.15% (required end point was <0.2%). The mixture
was dissolved in acetone (15 mL) and water (100 mL) and
agitated overnight. The trityl alcohol was filtered and washed
with water (15 mL); the aqueous filtrate was vacuum evaporated
to a syrup (4.45 g). The syrup was preabsorbed on silica gel
(10 g), applied to a silica column (60 g), and eluted with ethyl
acetate. Appropriate fractions were combined and evaporated,
and the residue was crystallized from ethanol to give 0.42 g
(8.6%) of dimer 7.
Preparation of Trityl AZT 3 and Sodium Trityl AZT 10 for
Use in NMR Comparisons. The trityl AZT 3 which was used to
obtain NMR data and to convert to sodium trityl AZT 10 was
prepared from AZT 4 (26.7 g) and trityl chloride (35.4 g) in
pyridine (100 mL) heated at 70ꢀ75 °C for 1 h and then kept at
room temperature for 20 h. A standard workup gave a syrup
was filtered, washed with water (3 ꢁ 35 kg), vacuum-distilled
(
25ꢀ30 in.), and dried at 55ꢀ65 °C for 12ꢀ14 h to yield 210.7
kg of AZT 4: HPLC purity = 99.7%, dimer 7 content = 0.15%.
Calcd for C H N 0 : C, 44.94; H, 4.90; N, 26.21. Found: C,
1
0 13 5 4
4
4.85; H, 4.87; N, 26.15. UV: λ (H O) 208, 268 nm. IR (KBr
max 2
1 +
ꢀ
disk): 2115 cm (azide). MS (EI) m/z, 267 (M ), 142 (sugar
moiety), 126 (thymine). H NMR (300 MHz DMSO-d ) δ 1.78
1
6
0
0
(
3
s, 3H, CH ), 2.23ꢀ2.43 (m, 2H, H-2 s), 3.62 (m, 2H, H-5 s),
3
0
.82 (dd, 1H, J = 4.0, 8.5 Hz, H-4 ), 4.41 (dd, 1H, J = 5.2, 12.0 Hz,
0
0
H-3 ), 5.23 (bs, 1H, exchangeable, OH), 6.10 (t, 1H, J = 6.4 Hz,
H-1 ), 7.69 (s, 1H, H-6), 11.33 (bs, 1H, NH). C NMR (75 MHz
DMSO-d ) δ 12.24 (CH3), 36.22 (C-2 ), 60.19 (C-3 ), 60.83
13
0
0
6
0 0 0
(C-5 ), 83.43 (C-1 ), 84.01 (C-4 ), 109.54 (C-5), 136.06 (C-6),
1
50.42 (C-2), 163.73 (C-4).
Isolation of 3 -N -(3 -Azido-3 -deoxythymid-3 -yl)-
-deoxythymidine (7). The MIBK mother liquor produced
which was purified by column chromatography to yield 41.18 g
0
00
000
000
00
1
of trityl AZT 3 as a white foam. H NMR (400 MHz DMSO-d
)
6
0
0
3
δ 1.57 (s, 3H, CH ), 2.37 and 2.50 (2m, 2H, 2 ꢁ H-2 ), 3.27
3
0
0
from processing 13.3 kg of tritylcyclothymidine was concen-
trated to 2.5 L, cooled, seeded, and held for 10 d. The crystals
were filtered, washed with MIBK, and dried to give about 600 g of
crystals containing ∼15% 7. The crystals were dissolved in ethyl
acetate (1 L) and applied to a silica gel column (1 kg) and eluted
with ethyl acetate/hexanes (7:3). Appropriate fractions were
combined and recrystallized from ethanol to give about 40 g of 7.
Mp 215.2ꢀ216.7 °C. Anal. Calcd for C H N 0 : C, 48.86; H,
(m, 2H, H-5 s), 3.89 (m, 1H, H-4 ), 4.60 (dd, 1H, J = 6.4, 13.3 Hz,
0
0
H-3 ), 6.15 (t, 1H, J = 6.4 Hz, H-1 ), 7.26ꢀ7.42 (m, 15H, trityl),
13
7.54 (s, 1H, H-6), 11.39 (s, 1H, NH). C NMR (100 MHz
0
0
DMSO-d ) δ 11.86 (CH ), 36.98 (C-2 ), 59.72 (C-3 ), 63.11
(C-5 ), 81.94 (C-1 ), 83.31 (C-4 ), 86.49 (Ph C), 109.73 (C-5),
6
3
0
0
0
3
127.19, 127.97, 128.23, and 143.35 (Ph C), 135.85 (C-6), 150.35
3
(C-2), 163.64 (C-4).
Trityl AZT 3 (0.509 g) and 50% sodium hydroxide (0.080 g,
1 equiv) were dissolved in methanol (20 mL) and evaporated in
20 25 7 8
5
.13; N, 19.96. Found: C, 48.54; H, 5.06; N, 19.75. UV: λ
max
1
ꢀ
1
(H O) 207, 268 nm. IR (KBr disk): 2101 cm (azide). MS
vacuo to a solid foam to give sodium trityl AZT 10. H NMR
2
+
+
1
(ESI ) m/z, 514.19 (67.8%)(M + Na ). H NMR (500 MHz
(400 MHz DMSO-d ) δ 1.51 (s, 3H, CH ), 2.32 and 2.50 (2 ꢁ
6
3
0
0
0
0
0
DMSO-d ) δ 1.78 (s, 3H, CH3 ), 1.84 (s, 3H, CH ), 2.15 (m, 1H,
H-2 ), 2.33 (m, 1H, H-2 ), 2.43 (m, 2H, H-2 , H-2 ), 3.49
m, 2H, 2 ꢁ H-2 ), 3.23 (m, 2H, H-5 s), 3.82 (m, 1H, H-4 ), 4.49
13
6
3
0
000
0
000
0
0
(m, 1H, H-3 ), 6.29 (t, 1H, J = 6.8 Hz, H-1 ), 7.21ꢀ7.40 (m, 15H,
1
1
2
2
0
0
000
(
(
(
m, 1H, H-5 ), 3.56 (m, 1H, H-5 ), 3.59 (m, 1H, H-5 ), 3.67
m, 1H, H-5 ), 3.83 (m, 1H, H-4 ), 4.11 (m, 1H, H-4 ), 4.38
m, 1H, H-3 ), 4.97 (t, 1H, exchangeable, 5 -OH), 5.23 (t, 1H,
trityl), 7.43 (s, 1H, H-6). C NMR 100 MHz DMSO-d ) δ
1
2
1
6
0
00
000
0
0
0
0
13.46 (CH ), 36.32 (C-2 ), 60.50 (C-3 ), 63.55 (C-5 ), 81.22
2
00
3
0
0
0
0
(C-1 ), 83.11 (C-4 ), 86.45 (Ph C), 109.77 (C-5), 127.15,
3
0
00
0
0
000
exchangeable, 5 -OH), 5.57 (m, 1H, H-3 ), 6.14 (t, 1H, H-1 ),
6
exchangeable, NH). The H NMR assignments were made by
comparison with the spectrum for 3 -azido-3 -deoxythymidine 4
and by analysis of a two-dimensional (COSY) experiment which
allowed the assignment of the sugar protons to the appropriate
127.96, 128.26, and 143.45 (Ph C), 133.32 (C-6), 157.84 (C-2),
3
00
.56 (t, 1H, H-1 ), 7.78 (s, 2H, H-6, and H-6 ), 11.27 (s, 1H,
173.79 (C-4).
1
Preparation of Sodium AZT 19 for Use in NMR Com-
parisons. AZT 4 (0.267 g) and 50% sodium hydroxide (0.080 g,
1 equiv) were dissolved in methanol (20 mL) and evaporated in
0
0
1
vacuo to a solid foam to give sodium AZT 19. H NMR (400 MHz
1
3
0
furanose ring. C NMR (75 MHz DMSO-d ) δ 12.20 and 12.97
DMSO-d ) δ1.57 (s, 3H, CH ), 2.37 and 2.50 (2m, 2H, 2 ꢁ H-2 ),
6
6
3
00
0
000
0
0
(
CH and CH ), 34.84 and 36.57 (C-2 and -2 ), 50.76 and
9.68 (C-3 and -3 ) 60.55 and 62.03 (C-5 and -5 ), 81.71 and
4.23 (C-4 and -4 ), 84.45 and 84.77 (H-1 and -1 ), 108.82 and
3.27 (m, 2H, H-5 s), 3.82 (dd, 1H, J = 4.0, 8.5 Hz, H-4 ), 4.41 (dd,
1H, J = 5.2, 12.0 Hz, H-3 ), 5.23 (bs, 1H, exchangeable, OH), 6.10
3
3
0
0
000
0
000
0
5
8
1
000
0
000
0
(t, 1H, J = 6.4 Hz, H-1 ), 7.69 (s, 1H, H-6), 11.33 (bs, 1H, NH).
0
0
00
13
0
09.66 (H-5 and -5 ), 134.92 and 136.78 (C-6 and -6 ), 150.30
C NMR (100 MHz DMSO-d ) δ 13.69 (CH ), 36.08 (C-2 ),
6
3
00
00
0
0
0
0
and 150.51 (C-2 and -2 ), 162.87 and 163.75 (C-4 and -4 ).
61.03 (C-3 ), 61.34 (C-5 ), 83.35 (C-1 ), 83.68 (C-4 ), 109.57
(C-5), 134.04 (C-6), 157.82 (C-2), 173.68 (C-4).
New Commercial Process for the Production of 3 -Azido-
0
00
000
000
00
Synthesis of 3 -N -(3 -Azido-3 -deoxythymid-3 -yl)-
-deoxythymidine (7). Trityl AZT 3 (5.1 g, 0.01 mol, prepared
0
0
3
0
from pure AZT 4 to ensure no contamination with dimer 7) was
combined with TCT 2 (9.33 g, 0.02 mol, 2 equiv) in DMSO
3 -deoxythymidine (AZT) 4. Step A: Preparation of AZT
Guanidine Salt 20. Tritylcyclothymidine (TCT) 2 (250 kg,
535.9 mol) was reacted with azide and further processed as
described in Step A of the first example of the Experimental
Section to give 900 L of a solution (pH 12ꢀ13) containing
sodium AZT 19. The solution was heated to 70 °C, solid
guanidine hydrochloride (61.3 kg, 641.9 mol, 1.2 equiv) was
added, a precipitate formed and the pH was adjusted to
12.2ꢀ12.3 with 50% NaOH (∼6 kg). The mixture was va-
cuum-distilled (350ꢀ550 mmHg) at ∼60 °C to a volume of
about 300 L to give a pastelike mixture, to which was added
methanol (200 kg). The mixture was heated to reflux, maintained
for 15 min (crystals remained), cooled to 5 °C, maintained for 6 h,
filtered, and washed with water (2 ꢁ 50 kg) to give 186 kg of
water-damp AZT guanidine 20. A loss on drying determination
(
25 mL) and heated to dissolve the solids. Potassium carbonate
(2.8 g, 2 equiv) was added to the mixture, heated to 108 °C, and
maintained for 24 h. The mixture was diluted with MIBK
(
75 mL), brine (50 mL) and water (50 mL), the pH was adjusted
to about 7 with aq HCl and the phases were separated. The
aqueous phase was extracted with MIBK (3 ꢁ 50 mL). The
combined MIBK phases were vacuum evaporated on a rotary
evaporator to a solid foam (14.78 g). This was dissolved in
ethanol (50 mL), chloroform (25 mL), and trifluoroacetic acid
(
25 mL). After 75 min at ambient temperature, TLC showed
that the reaction was complete, and the reaction mixture was
vacuum evaporated (rotary evaporator) at ∼40 °C. Methanol
(3 ꢁ 50 mL) was added and evaporated three times. The syrup
1
285
dx.doi.org/10.1021/op2000805 |Org. Process Res. Dev. 2011, 15, 1281–1286

Organic Process Research & Development
ARTICLE
(
LOD = 24.1%) gave a dry weight of 141.2 kg (433.1 mol, 80.8%).
’ ACKNOWLEDGMENT
An HPLC analysis: 94.54%/0.34% for AZT 4/dimer 7.
I thank the management of Apotex Pharmachem Inc. for
supporting this work. In particular I acknowledge the helpful
discussions and suggestions for the preparation of this manu-
script from Dr. Keshava Murthy, Dr. Stephen Horne, and Dr.
Allan Rey.
0
0
An analytical sample of 3 -azido-3 -deoxythymidine guanidine
0
0
salt 20 was prepared by suspending clean 3 azido-3 -deoxythy-
midine 4 (53.4 g. 0.2 mol) in water (250 mL) and adding 50%
NaOH (17.6 g, 0.22 mol) to pH 13.0. The mixture was heated to
7
0 °C. Thereafter, guanidine hydrochloride (23 g, 0.24 mol) was
added and heated to 97 °C to dissolve the emerging precipitate.
The mixture was set aside to cool with agitation to crystallize for
’
REFERENCES
1
6 h. The crystals were filtered, washed with water (50 mL) and
(1) Horwitz, J. P.; Chua, J.; Noel, M. J. Org. Chem. 1964, 29, 2076.
ethanol (50 mL), and dried to 58.76 g (0.18 mol, 90%) of white
(2) Nickel, W.; Austermann, S.; Bialek, G.; Grosse, F. J. Biol. Chem.
crystals: mp 210.9ꢀ212 °C. Anal. Calcd for C H N O : C,
11 18 8 4
1
992, 267, 848.
3) Horwitz, J. P.; Urbanski, J. A.; Chua, J. J. Org. Chem. 1962,
4
0.49; H, 5.56; N, 34.34. Found: C, 40.92; H, 5.45; N, 34.17.UV:
(
ꢀ
1
1
λmax (H O) 268 nm. IR (KBr disk): 2095 cm (azide). HNMR
2
27, 3300.
0
(500 MHz DMSO-d ) δ 1.71 (s, 3H, CH ), 2.12 (m, 1H, H-21 ),
6
3
(4) Michelson, A. M.; Todd, A. R. J. Chem. Soc. 1955, 816.
0
0
0
0
0
0
2
4
(
(
(
(
.25 (m, 1H, H-22 ), 3.57 (m, 2H, H-51 , -52 ), 3.75 (m, 1H, H-4 ),
.33 (m, 1H, H-3 ), 6.12 (t, 1H, H-1 ), 7.29 (s, 1H, H-6), 7.38
bs, 7H, 3-NH, 5 -OH, exchangeable guanidine). C NMR
100 MHz DMSO-d ) δ 13.66 (CH ), 36.24 (C-2 ), 60.87
C-3 ), 62.32 (C-5 ), 83.47 (C-1 ), 83.70 (C-4 ), 109.68
C-5), 134.52 (C-6), 157.86 (C-2), 158.99 (guanidinium),
(5) Horwitz, J. P.; Chua, J.; Urbanski, J. A.; Noel., M. J. Org. Chem.
0
1963, 28, 942.
13
(6) Fox, J. J.; Miller, N. C. J. Org. Chem. 1963, 28, 936.
(7) Glinski, R. P.; Khan, M. S.; Kalamas, R. L.; Sporn, M. B. J. Org.
Chem. 1973, 38, 4299.
0
6
3
0
0
0
0
(8) (a) Fleet, G. W. J.; Son, J. C.; Derome, A. E. Tetrahedron 1988,
44, 625. (b) Chu, C. K.; Beach, J. W.; Ullas, G. V.; Kosugi, Y. Tetrahedron.
1
73.96 (C-4).
Step B: Isolation of Crude AZT 4 from AZT Guanidine Salt 20.
Lett. 1988, 29, 5349. (c) Kowollik, G.; Gaertner, K.; Langen, P.
Tetrahedron. Lett. 1969, 10, 3863. (d) Rao, T. S.; Reese, C. B. J. Chem.
Soc., Chem. Commun. 1989, 997.(e) Czernecki, S.; Viile, G.; Valery, J.-M.
Water (about 137 kg) was combined with 1 equiv of conc HCl
and heated to 75 °C. Then water-damp AZT guanidine 20 cake
(
0
0
Process for the synthesis of 3 -azido-3 -deoxythymidine and the like. EP
about 263 kg; dry weight is 200 kg) was added; first a solution
0325537, 1989, CAN 112:56586. (f) Agyei-Aye, K.; Yan, S.; Hebbler,
forms, and soon after a precipitate began to form. The mixture
was cooled to 5 °C, maintained for 6 h, filtered, and washed with
water (3 ꢁ 20 kg) to yield 200.5 kg of water-damp crude AZT 4
A. K. Nucleosides Nucleotides 1989, 8, 327. (g) Verheyden, J. P. H.;
Moffat, J. G. J. Org. Chem. 1970, 35, 2868.(h) Hayauchi, Y.; Lockhoff, O.
Process for the preparation of 1-(3-azido-2,3-dideoxy-β-D-erythro-pen-
tofuranoxyl)pyrimidines, and intermediates. DE 3705794, 1988, CAN
(dry weight of 136.5 kg, 510.8 mol, 83.3%). HPLC indicated
109:211404. (i) Zaitseva, V. E.; Dyatkina, N. B.; Kraevskii, A. A.;
9
8.9% AZT 4 and 0.52% dimer 7.
Skaptsova, N. V.; Turina, O. V.; Gnuchev, N. V.; Gottikh, B. P.; Azhaev,
A. V. Bioorg. Khim. 1984, 10, 670. (j) Matsuda, A.; Watanabe, K. A.; Fox,
J. J. J. Org. Chem. 1980, 45, 3274. (k) Secrist, J. A., III. Carbohydr. Res.
Step C: Preparation of Pure AZT 4. Water-damp crude AZT 4
176.2 kg, dry weight 120 kg,) from Step B was dissolved in
(
water (304 kg) and acetone (190 kg) at 50 °C. Active carbon
1
975, 42, 379.
9) (a) Radatus, B. K.; Karimian, K. Recovery of AZT by Guanidine
(
3 kg) was added and cooled to 30 °C; Celite (9 kg) was
(
added; the mixture was filtered over a Celite bed, and the
bed was washed with additional 40 °C acetone (3 ꢁ 25 kg).
About 74 kg of acetone was distilled at atmospheric pressure
Precipitation. U.S. Patent 5,387,677, 1995, CAN 122:248299. (b)
Adams, D. R.; Perez, C.; Maillard, M.; Florent, J.-C.; Evers, M.; H ꢀe nin,
Y.; Litvak, S.; Litvak, L.; Monneret, C.; Grierson, D. S. J. Med. Chem.
1997, 40, 1550.
1
and, until H NMR determination, produced a water/acet-
(
10) Zeidler, J. M.; Kim, M. H.; Watanabe., K. A. Nucleosides and
Nucleotides 1990, 9, 629.
11) ACD/C+H NMR Predictor and DB, version 10.02; Advanced
one ratio of 2.55 (acceptable range is 2.0ꢀ3.0). The mix-
ture was cooled to 35 °C, seeded with pure AZT 3, cooled
to 23 °C, and agitated for 12 h. The mixture was vacuum-
distilled (400ꢀ550 mmHg) at 38 °C until 81 kg (little observ-
able distillate) had been collected. The mixture was cooled
to 5 °C, maintained for 1 h, a small sample of crystals was
filtered, and HPLC determined the dimer content to be
(
Chemistry Development, Inc.: Toronto, ON, Canada, 2006, www.
acdlabs.com.
(
12) Hirota, K.; Watanabe, K. A.; Fox, J. J. J. Org. Chem. 1978,
3, 1193 and references cited therein.
13) Hirota, K.; Watanabe, K. A.; Fox, J. J. J. Heterocycl. Chem. 1977,
14, 537.
4
(
0
.16% (end point <0.2%). The mixture was filtered, washed
with water (2 ꢁ 30 kg) to 137 kg of water damp crystals and
vacuum-dried at 55ꢀ65 °C to 106.2 kg (88.5%) of AZT 4;
9
9.7% HPLC A%, 7 content = 0.15%.
’
ASSOCIATED CONTENT
S
Supporting Information. Experimental and calculated
b
1
3
C NMR spectra for neutral and ionized forms of trityl AZT,
AZT, and guanidine. This material is available free of charge via
the Internet at http://pubs.acs.org.
’
AUTHOR INFORMATION
Corresponding Author
bradatus@apotexpharmachem.com
1
286
dx.doi.org/10.1021/op2000805 |Org. Process Res. Dev. 2011, 15, 1281–1286