Independent synthesis and fate studies of impurities in process intermediates of the anti-AIDS drug d4T

Article abstract of DOI:10.1021/op970126p

Impurities in isolated intermediates in a process to prepare d4T were identified, independently synthesized, and then taken through the process to determine their ultimate fate. Some of the products from these fate studies were also independently synthesized and used in the validation of impurity assay methods.

Full text of DOI:10.1021/op970126p

Organic Process Research & Development 1998, 2, 203−207
Lecture Transcripts
Independent Synthesis and Fate Studies of Impurities in Process Intermediates
of the Anti-AIDS Drug d4T
Jayachandra P. Reddy,* J. Gregory Reid, Deborah A. Renner, Sandra L. Quinlan, Douglas G. Weaver,†
Bang-Chi Chen, and Derron R. Stark
Chemical DeVelopment Laboratories, Technical Operations, Bristol-Myers Squibb Company, P.O. Box 4755,
Syracuse, New York 13221-4755
Qi Gao‡
Analytical Research, Pharmaceutical Research Institute, Bristol-Myers Squibb Company, 5 Research Parkway,
Wallingford, Connecticut 06492
Scheme 1
Abstract:
Impurities in isolated intermediates in a process to prepare d4T
were identified, independently synthesized, and then taken
through the process to determine their ultimate fate. Some of
the products from these fate studies were also independently
synthesized and used in the validation of impurity assay
methods.
Introduction
d4T (stavudine, Zerit) is currently being marketed for the
treatment of AIDS in the United States and other countries.
Several syntheses of d4T have appeared in the literature.1,2
Among them is one route that was developed by us recently
for the production of d4T from 5-methyluridine (5-MU).2
During the initial scale-up of this new process (Scheme 1),
the levels of some impurities in isolated intermediates were
observed to increase. Investigations into the causes of the
higher levels revealed that these increases were generally
due to the longer processing times that were required on
larger scale, although in some cases there were other
* Phone: (315) 432-2414. Fax: (315) 432-4854. E-mail: jreddy@
usccmail.uscc.bms.com.
† Chemical Development Analytical Services (CDAS).
contributing factors. Independent synthesis of the impurities
not only confirmed the structure assignments but also
provided samples to carry out fate studies. In addition,
samples of observed impurities and potential impurities were
used to validate impurity assay methods.
‡ Correspondence regarding X-ray crystallography should be addressed to this
author.
(1) For recent syntheses of d4T see: (a) McDonald, F. E.; Gleason, M. M. J.
Am. Chem. Soc. 1996, 118, 6648. (b) Clive, D. L. J.; Wickens, P. L.; Sgarbi,
P. W. M. J. Org. Chem. 1996, 61, 7426. (c) Mustafin, A. G.; Gataullin, R.
R.; Spirikhin, L. V.; Abdrakhmanov, I. B.; Tolstikov, G. A. Zh. Org. Khim.
1996, 32, 1842. (d) Shiragami, H.; Ineyama, T.; Uchida, Y.; Izawa, K.
Nucleosides Nucleotides 1996, 15, 47. (e) Becouarn, S.; Czernecki, S.;
Valery, J.-M. Nucleosides Nucleotides 1995, 14, 1227. (f) Larsen, E.;
Kofoed, T.; Pedersen, E. B. Synthesis 1995, 1121. (g) Niihata, S.; Kuno,
H.; Ebata, T.; Matsushita, H. Bull. Chem. Soc. Jpn. 1995, 68, 2327. (h)
Lipshutz, B. H.; Stevens, K. L.; Lowe, R. F. Tetrahedron Lett. 1995, 36,
2711. (i) Luzzio, F. A.; Menes, M. E. J. Org. Chem. 1994, 59, 7267. (j)
Jung, M. E.; Gardiner, J. M. Tetrahedron Lett. 1992, 33, 3841. (k) Vargeese,
C.; Abushanab, E. Nucleosides Nucleotides 1992, 11, 1549. See also ref 2
and references therein.
Results and Discussion
The only significant process impurity in the first step (5-
MU f 1, Scheme 1) of this synthesis is 5′-chloro-5′-deoxy-
3′-O-(methylsulfonyl)-2,2′-anhydro-5-methyluridine, 3 (see
eq 1), which is presumably formed from chloride displace-
ment of the primary methylsulfonyl ester at C-5′ during the
reaction. Independent synthesis of chloride 3 was easily
accomplished by reaction of 1 with lithium chloride in DMF
(2) Chen, B.-C.; Quinlan, S. L.; Stark, D. R.; Reid, J. G.; Audia, V. H.; George,
J. G.; Eisenreich, E.; Brundidge, S. P.; Racha, S.; Spector, R. H. Tetrahedron
Lett. 1995, 36, 7957.
S1083-6160(97)00126-6 CCC: $15.00 © 1998 American Chemical Society and Royal Society of Chemistry
Published on Web 04/17/1998
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
203

at 87 °C in 68% yield.3
during the isolation of d4T, an optional rework procedure
was developed for benzoyl d4T, 2, a highly crystalline late
intermediate from the previous step. Eventually, a recrys-
tallization from CH3OH was developed that provided pure
benzoyl d4T, 2, with complete removal of impurities 4 and
5.
Since only a small quantity of 3 was removed during the
isolation of anhydro-intermediate 1, we were anxious to find
out the fate of 3 during the remainder of the process.
Subjecting 3 to the conditions of the second step of the
process shown in Scheme 1 produced two new impurities,
5′-chloro-2′,3′-didehydro-3′,5′-dideoxythymidine, 4, and the
corresponding 5′-bromo analogue, 5, as a 4:1 mixture of
products in 78% yield (eq 1). Close examination of HPLC
chromatograms of benzoyl d4T, 2, that had been prepared
from 1 containing appreciable amounts of 5′-chloro impurity
3 revealed the presence of 4 and 5, albeit in reduced amounts.
Pure samples of 4 and 5 were prepared independently in 84%
and 72% yield, respectively, by reaction of the mesylate of
d4T, 6 (eq 2), with lithium chloride or lithium bromide in
DMF at 80 °C. Mesylate 6 in turn was prepared in 87%
yield from d4T (MsCl, i-Pr2NEt, THF, 0 °C f rt).
While exo-olefin 7 could not be effectively removed from
d4T directly, we reasoned that, with acid catalysis, 7 could
be converted to methylfuran 9 (eq 4), which might prove
easier to remove. Thus, treatment of 7 with 0.15 equiv of
PPTS in CH2Cl2 at room temperature gave 9 in 95% yield
after flash chromatography. Unfortunately, d4T is itself
sensitive to acid, and no effective purification procedure
utilizing an acid treatment was found for d4T contaminated
with 7.
Impurity 10 (eq 5) was isolated from a batch of benzoyl
d4T, 2, that was produced on large scale. Its preparation
from 1 is shown in eq 5. Treatment of 1 with PhCOOK in
hot DMF followed by 57% aqueous HI gave 10 in 59% yield.
To determine the configuration of the C-2′ hydroxyl, 10 was
converted to 11 (eq 6). A comparison of NOE enhancements
of 11 and dibenzoate 13 (see Figure 1), which had been
isolated from an impure laboratory lot of 2, was then made
to allow the assignment of stereochemistry.
Treating 4 and/or 5 with n-butylamine, as in the third step
of the process (2 f d4T), provided 7 and 8 (HPLC ratio
9:1), formal products of â-elimination and displacement,
respectively, as a mixture that was separable by flash column
chromatography (eq 3).4
Unacceptably high levels of exo-olefin 7 and n-butyl-
amino derivative 8 were observed in some laboratory lots
of d4T, which were prepared from lots of benzoyl d4T, 2,
that contained high levels of impurities 4 and 5 (both sources
of 7 and 8). Since 7 and 8 were not consistently removed
1
Figure 1 summarizes the results from the H-1H NOE
experiments performed on 11 and 13.5 Irradiation of H1′ in
11 resulted in a 9.4% enhancement of H2′ and a 2%
enhancement of the signal due to H4′. In addition, irradiation
(3) None of the impurity syntheses described in this publication are optimized.
All new compounds were characterized by 1H/13C NMR, IR, UV and MS.
We thank Drs. H. Y. Lin (CDAS) and D. W. Dodsworth (AR&D) for
gathering MS data.
(4) Interestingly, treatment of mesylate 6 directly with n-butylamine gave 8
as the major product (73% HPLC area) along with 7 (12% HPLC area)
and d4T (8% HPLC area). Also, reaction of 6 with bases such as i-Pr2NEt
and i-Pr2NH gave only decomposition.
(5) All 1H and 13C chemical shift assignments were made from a combination
of several NMR experiments (DEPT, COSY, HMQC, HMBC) performed
on a Bruker DRX 360 MHz or a Bruker DRX 400 MHz spectrometer in
DMSO-d6 as solvent. The NMR spectra of 18 and 19 were obtained in
CD2Cl2.
204
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

Scheme 2
While we were pleased to find out that 10 was also
completely removed during the optional CH3OH recrystal-
lization of 2, we were interested in determining its fate
through the rest of the process. Thus, reaction of 10 with
n-BuNH2 at 70 °C (eq 7) gave a major product identified as
15, among other unisolable substances.
Figure 1. Results from NOE experiments on 11 and 13.
of H2′ in 11 gave enhancements of 9.3% and 1.6% of the
H1′ and H4′ signals, respectively. These results suggest a
cis substitution pattern across C-1′,2′ in 11. On the other
hand, irradiation of H1′ in 13 gave enhancements of the H2′,
H4′, and H6 signals of 1.7%, 3.1%, and 3.5%, respectively.
Enhancements of the signals due to H6 (4.9%), H1′ (2.7%),
and H3′ (6.7%) were noted when H2′ was irradiated. When
signal H3′ was irradiated, an enhancement of 6.4% was
observed on H2′. These results are consistent with the
stereochemistry assigned to C-2′ in 13. Further confirmation
of the stereochemistry of 10 came from X-ray crystal-
lographic analysis of the p-nitrobenzoate derivative, 12,
which was prepared from 10 in 75% yield (eq 6).
It is interesting to note that while 10 and 13 are formed
as impurities during the same reaction (1 f 2, Scheme 1),
their configuration at C-2′ is opposite. This can be explained
by the involvement of the common intermediate, 14 (Scheme
2), which is formed from 1 in the d4T process. Anhydro-
intermediate 14 can undergo acid-catalyzed hydrolysis at C-2
by adventitious water to yield 10 (Scheme 2, path A), while
displacement by PhCOOK at C-2′ can yield 13 (path B). It
should be noted that 2′â-benzoate 11 was not observed as
an in-process impurity or as an impurity in the isolated
intermediate, benzoyl d4T, 2. It was synthetically prepared
from 10 as described above.
The configurations at C-2′ and C-3′ of 15 were determined
1
from H-1H NOE experiments summarized in Figure 2.5
Thus, irradiation of H1′ resulted in an enhancement of 6.6%
of H2′ and 0.9% of H6, while irradiation of H2′ gave
enhancements of 13.7%, 8.3%, and 1.8% of the signals due
to H1′, H2′-OH, and H3′, respectively. Irradiation of H3′ caused
enhancement of signals from H6 (4.2%), H2′-OH (1.5%), and
H2′ (2%). In addition, irradiation of H6′ (CH2 next to N on
the n-butyl chain) gave enhancement of 4.1% and 4.4% of
the H2′ and H3′ signals, respectively.
Impurity 15 is likely formed by debenzoylation of 10 with
concomitant formation of a 2′,3′-â-epoxide, which in turn
could be opened by excess n-butylamine present in the
reaction.6 Fortunately, 15 and other by-products from this
reaction are completely removed during the isolation of d4T.
Another impurity present in some batches of 2 was
isolated and initially identified as 18 (Scheme 3). Furan 18
was prepared from 7 in two steps as shown in Scheme 3.
Epoxidation of the exocyclic double bond in 7 presumably
gave the corresponding epoxide, which isomerized readily
(6) Only one regioisomer was isolated from the ring opening of the putative
2′,3′-â-epoxide intermediate by n-butylamine. It is not clear why there is
a preference for the formation of 15 over the other regioisomer. More likely,
a minor regioisomer was formed in the reaction and it was one of the
unisolable substances (vide supra).
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
205

The 1H NMR5 spectrum of the impurity indicated a methyl
group (1.83 ppm, d, J ) 1 Hz, 3H) which was coupled to a
vinyl CH (7.14 ppm, d, J ) 1 Hz, 1H). A broad singlet at
9.48 ppm (1H) indicated an NH, and multiplets at 8.17 (2H),
7.66 (1H), and 7.53 ppm (2H) indicated a phenyl group. Two
doublets at 7.41 and 6.61 ppm (J ) 2.3 Hz, 1H each) along
with a singlet at 4.92 ppm were also present in the spectrum.
The 13C NMR5 spectrum showed three carbonyl peaks at
163.9, 163.6, and 150.5 ppm. Ten more carbons, of which
four were quaternary carbons (DEPT), were evident in the
100-150 ppm range. Signals at 11.7 ppm (CH3) and 40.7
ppm (CH2) completed the carbon NMR spectrum. On the
basis of the molecular weight derived from the MS data,
and the NMR spectra, we initially deduced the structure of
the impurity to be 18. However, two aspects of the NMR
data troubled us about this structure. First, in the 1H NMR
spectrum, the doublets at 7.41 and 6.61 ppm were almost a
full part per million apart, and second, in the 13C NMR
spectrum, the CH2 was at 40.7 ppm (instead of the expected
∼60 ppm).8 The NMR data from the synthetic sample
confirmed our suspicion that our assigned structure may be
wrong. While most other signals were comparable, the
doublets at 6.61 and 6.48 ppm (J ) 3.6 Hz, 1H each)
Figure 2. Results from NOE experiments on 15.
Scheme 3
(7) The formation of lactone 17 may be explained by the mechanism shown
below. Ring opening of the spiro epoxide i by m-CPBA could give rise to
intermediate ii and/or iii. Fragmentation of ii/iii can then yield 17 from the
formal loss of m-chlorobenzoic acid and formaldehyde.
under the reaction conditions to provide 16 in 41% yield.
Lactone 17, which was difficult to separate by flash
chromatography from 16, was also formed during this
reaction.7 Benzoylation of (hydroxymethyl)furan 16 (still
containing lactone 17) provided benzoate 18 as expected.
Compounds 17 and 18 were easily separated at this stage
by flash chromatography. Upon examination of the 1H and
13C NMR spectra, as well as HPLC retention times and UV
spectra, it was clear that the synthetic material, 18, was
different from the impurity that was isolated from 2. The
molecular weight of both these substances was found to be
the same (326) as determined by LRMS.
(8) Nevertheless, 18 made good sense in that we saw how this could be easily
formed by elimination of the elements of HBr and methanesulfonic acid
from iv, an intermediate in the preparation of 2 (resulting also in
aromatization of the furan ring).
206
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

1
highlighted the difference in the H NMR of the synthetic
HBr/HOAc in hot DMF. An equilibrium between 14 and
20 may even be possible. Neighboring-group participation
with displacement of the C-3′ mesylate by the benzoate group
at C-5′ could give the stabilized carbocation 21. Bond
formation across N-1 and C-5′ along with cleavage of the
oxygen-C5′ bond would yield 22. Protonation of the
pyrimidine ring followed by elimination across C-2′ and C-3′
to give 23 and then a second elimination (of HX) would
yield 19.
As in the case of 10, furan 19 is converted to products
which are removed during the isolation of d4T in the
subsequent step. It is also removed from 2 during the
optional CH3OH recrystallization. Exposure of a sample of
19 to the conditions of step 3 (2 f d4T, Scheme 1) led to
its conversion to a large number of minor products, none of
which were identified.
sample. Additionally, a CH2 signal at 57.9 ppm was present
in the 13C NMR spectrum of the synthetic sample while all
the other signals were comparable to the spectrum of the
impurity sample.
We speculated then that the impurity may be 19 instead
of 18, purely on the basis of the NMR chemical shift data
of the impurity. The identity of the isolated impurity was
unequivocally established by X-ray crystallography to be 19.
The synthetic substance, on the other hand, was shown to
be 18, again by X-ray crystallography.
An independent preparation of lactone 17 was achieved
by bubbling oxygen through a hot solution of d4T in
CH3CN/N-methyl-2-pyrrolidinone in the presence of catalytic
AIBN (Scheme 5) in 16% yield. This reaction is presumed
to go through hydroperoxide 24, which upon fragmentation
would provide 17.
The formation of 19 during the preparation of 2 (from 1)
may be explained as shown in Scheme 4. Intermediates 20
Scheme 4
Scheme 5
Conclusions
In conclusion, many impurities observed during the
development of the 5-MU process for d4T were isolated and
definitively identified. In addition, these impurities were
synthesized independently, and fate studies were performed.
Process modifications have been made in order to avoid the
formation of these impurities, and this process now delivers
very pure d4T in high overall yield.
Received for review January 13, 1998.
OP970126P
(X ) Br or OAc) could result from 14 by the reaction of
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
207