Determination of the source of the N-methyl impurity in the synthesis of pemetrexed disodium heptahydrate

Article abstract of DOI:10.1021/op0498013

The synthesis of Pemetrexed Disodium Heptahydrate has consistently resulted in a very low level (ca. 0.02%) unknown impurity. To ensure long-term control, the identity and source of the impurity were desired. Isolation and characterization identified the impurity as the N-methyl derivative. The source was identified as the methyl groups on the peptide coupling agent, 2,6-Dimethoxy-1,3,5-triazine (CDMT). Further work assured the current conditions provide adequate control.

Full text of DOI:10.1021/op0498013

Organic Process Research & Development 2005, 9, 738−742
Determination of the Source of the N-Methyl Impurity in the Synthesis of
Pemetrexed Disodium Heptahydrate
Douglas P. Kjell,* Dallas W. Hallberg, J. Michael Kalbfleisch, Cynthia K. McCurry, Michael J. Semo,
Edward M. Sheldon, Jeremy T. Spitler, and Ming Wang
Eli Lilly and Company Tippecanoe Laboratories, 1650 Lilly Road, Lafayette, Indiana 47909-9201, U.S.A.
Abstract:
The synthesis of Pemetrexed Disodium Heptahydrate has
consistently resulted in a very low level (ca. 0.02%) unknown
impurity. To ensure long-term control, the identity and source
of the impurity were desired. Isolation and characterization
identified the impurity as the N-methyl derivative. The source
was identified as the methyl groups on the peptide coupling
Figure 1. Structure of Pemetrexed Disodium Heptahydrate.
agent, 2,6-Dimethoxy-1,3,5-triazine (CDMT). Further work
assured the current conditions provide adequate control.
which it is cleaved to the aldehyde have been disclosed
previously.2 The route from the R-bromide 3 onward reflects
Introduction:
further process development on a previously disclosed route.3
The cyclocondensation of 3 with 2,4-diamino-6-hydroxypy-
rimidine 4 is a very efficient method to construct the pyrrolo-
[2,3-d] heterocycle. However the product 5 of this step is
relatively impure for a pharmaceutical intermediate.
The specifications for 5 allow up to 10 area % impurities
by HPLC, with typical levels approaching that limit.
Considerable effort failed to discover an effective recrystal-
lization of 5. The saponification to 6 only provides slight
purification, typically less than 1% reduction in total impuri-
ties. It is unusual for the specification of isolated intermedi-
ates within the pharmaceutical industry to allow these
amounts of impurities.4 However the specifications are
justified, as the remaining steps provide a highly effective
purification strategy that includes four isolations under
disparate conditions. The acid is activated for coupling by
reaction with 2-chloro-4,6-dimethoxytriazine (CDMT)5 to
form 7 and then reacted with L-glutamic acid diethyl ester
(LGADE) 8. The product of the peptide coupling 9 is isolated
as a p-toluenesulfonic acid salt 10 from ethanol and then
recrystallized from DMSO/EtOH. Compound 10 is then
saponified to produce the free acid form of the drug substance
(11), which is isolated. Finally, the pH is adjusted, and the
crystalline disodium salt is isolated as the heptahydrate 1.
The purification strategy has proven very effective; no
identified impurity expected in the drug substance (at the
outset of this work) was related to the method of synthesis.
All impurities expected were the result of trace degradation
of 1.
Pemetrexed Disodium Heptahydrate (LY231514*2Na*7-
H2O, active ingredient in Alimta) (1, Figure 1) is a multi-
targeted antifolate approved for treatment of mesothelioma
and for second-line treatment of nonsmall cell lung cancer
(NSCL). Alimta is also under investigation for multiple other
cancers. The current amount of Alimta administered to an
average size individual (1.8 m2) at the recommended dose
(500 mg/m2) is 0.9 g/day. Of particular concern to this work
is the possibility of dose escalation during the treatment of
future cancers. The International Conference on Harmoniza-
tion (ICH)1 sets a high standard for the purity of drug
substances. If the dose is less than 2 g /day, impurities over
0.10% are expected to be identified, justified by toxicology
data, and controlled to a specification. If the dose exceeds 2
g/day, the qualification threshold is lowered to 0.05%. The
synthesis of Pemetrexed Disodium Heptahydrate is very
capable of meeting the 0.10% threshold.
All impurities that could exceed that 0.10% threshold have
been identified, have been qualified, and are controlled. These
impurities are all the result of trace degradation of Pemet-
rexed Disodium Heptahydrate in the final isolation by
oxidative pathways. If Alimta, in the future, were to be
investigated at doses requiring the 0.05% threshold, an
impurity, which had not been identified at the outset of this
work, would require reinvestigation. Levels of this impurity
have consistently been ca. 0.02% in both plant and laboratory
samples. It was of interest to identify this impurity, determine
the source, and ensure that the current control strategy is
adequate or to modify the strategy if needed. This report
demonstrates the level of scientific understanding needed to
ensure the quality of a drug substance.
(2) Kjell, D. P.; Slattery, B. J.; Semo, M. J. J. Org. Chem. 1999, 64, 5722-
5724.
The commercial synthesis is shown in Scheme 1. The
preparation of bisulfite adduct 2 and the conditions under
(3) Barnett, C. J.; Wilson, T. M.; Kobierski, M. E. Org. Proc. Res. DeV. 1999,
3, 184-188.
* Corresponding author. E-mail: Kjell.DP@Lilly.com.
(1) International Conference on Harmonisation (ICH) Guidelines, Q3A(R),
Impurities in New Drug Substances (Revised Guideline), 2002.
(4) Based on an informal survey, and personal experience, <2-3% total related
substances would be a more typical specification for an early intermediate.
(5) Kaminski, Z. J. Tetrahedron Lett. 1985, 26, 2901-2904.
738
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development
10.1021/op0498013 CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/06/2005

Scheme 1. Commercial synthesis of Pemetrexed Disodium Heptahydrate
Results and Discussion
A key portion of the development of a commercial
synthesis is the identification of all impurities in the drug
substance and later intermediates. From the outset, identifica-
tion of the unknown impurity had proven problematic. No
samples of Pemetrexed with >0.05% of the impurity were
available. Therefore, only a weak, noisy spectrum could be
obtained by HPLC/MS. Based on this spectrum a molecular
weight of 504 was proposed, but we were unable to draw a
structure to fit this mass. Without a structure it is impossible
to determine the source of an impurity in a complex synthetic
route; the mass expected for earlier steps cannot be predicted.
Given the possibility of a higher dose requiring the lower
qualification threshold, identification was again attempted.
Improved isolation equipment and NMR techniques resulted
in the successful identification of the impurity.
Figure 2. Possible N-methyl derivatives.
Critical to the success of this identification attempt was
a change in strategy. Rather than attempt to identify the very
low level impurity in situ, the impurity was isolated by
preparative chromatography. This allowed an exact mass to
be determined, as well as 2D-NMR to be gathered. The initial
mass proposed was proven incorrect, and a correct exact mass
for the M + 1 of 442.1704 was determined. The structure
was assigned as the N-methyl derivative 12 (Figure 2) and
Figure 3.
attempted was to examine all isolated process intermediates
for an “M + 14” impurity. The results of this examination
were ambiguous at best. It was impossible to definitively
discern at what point the N-methylation occurred.
The second approach pursued was to start at the earliest
possible source. Perhaps the N-methyl arrives as an impurity
in the starting material, 2,4-diamino-6-hydroxypyrimidine 4.
An authentic sample of the N-methyl derivative 13 (Figure
3) was prepared6 and found to be completely unreactive with
the R-bromoaldehyde 3 under the standard conditions and
under all alternatives attempted. The limited solubility of 13
contributes to the lack of reactivity. Therefore, the presence
of 13 as an impurity in 4, while not rigorously disproven,
could not be causal for N-methylation.
confirmed by independent synthesis. The mass, 1H, and 13
C
NMR clearly indicate an N-methyl substitution. The HMBC
spectrum shows the long range 1H to 13C correlations which
support structure 12 over the other possible N-methyl
derivatives. Treatment of the drug substance 1 with an excess
of methyl iodide, saponification and preparative chromatog-
raphy prepared 12 in a 65% yield. The spectra of synthetic
and isolated 12 matched.
The identification of the N-methyl derivative 12 was
critical to identification of the source. The first method
(6) Munesada, K.; Suga, T. J. Org. Chem. 1987, 52, 5655-5662.
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
739

Table 1. HPLC data for area % N-methyl 16 to main peak
in standard reaction (6 to 9)
results were determined as area % N-methyl 16 relative to
the desired product. Each sample was injected once at a high
concentration to get the area of 16 and then diluted 1/100 to
determine the area of 9 without saturating the detector. The
results of the experiments using this methodology are
reported in Table 1. To ensure worst case results, the excess
of CDMT was increased from the typical 5% to 35%.
Control reactions under these conditions showed 0.02%
16 if the mixture was sampled after 1 h. Slightly more
N-methyl was observed on extremely prolonged stirring
(0.09% after 20 h).
conditions
16 (%) at 1 h
prolonged (%)
0.09 @ 20h
1.35 equiv of CDMT
1.35 equiv of CDMT, DBU
2.70 equiv of CDMT
0.02
detecteda
0.02
0.3 @ 5 days
0.6 @12 days
0.01 @ 20 h
0.18 @ 20 h
1.1 equiv of CDMT
2.70 equiv of Aldrich CDMT
0.01
0.02
a Yield of 9 poor, complicated mixture produced.
It is known that the complex of CDMT and NMM
decomposes to lose the methyl group originally attached to
the NMM, although a rate of decomposition is not reported
in DMF (eq 1).7
Next we introduced the methyl at the first possible
intermediate. Treatment of the cyclocondensation product 5
with methyl iodide again produced the N-methyl derivative
14 (Figure 2). An authentic sample allowed us to check
production samples of 5 for the impurity by HPLC/MS
retention time and mass match. This unambiguously estab-
lished that 14 was not present in 5 produced in the plant.
A sample of 5 with 5 mass % of the N-methyl 14 was
then prepared and carried forward through the synthesis in
the laboratory. Isolated intermediates (6, 10, 11, 1) were
examined for impurities with a mass and retention time
match. The first match was observed in the p-TsOH salt 10.
The reason for the ambiguity of the first attempt to locate
the source (by HPLC/MS without authentic samples) became
clear. The amount of the N-methyl impurity is very small
and was poorly resolved by the original HPLC/MS method
from a somewhat larger impurity.
Attention therefore focused on the chemistry converting
the acid 6 into the peptide-coupled product 10. Production
samples were used to further narrow the source. The first
step within this chemistry is reaction of a slight excess of
CDMT with the acid 6 catalyzed by N-methylmorpholine
(NMM) in DMF at 0 °C. This active ester intermediate 7 is
unstable, so it could not be sampled for this purpose.
L-Glutamic acid HCl (LGADE) 8 is then added, and the
mixture is warmed to 20-25 °C, stirred for 1 h to form 9,
and then sampled. Other sampling points chosen were as
follows: the organic phases during the EtOAc/water parti-
tions used to attenuate the DMF levels to allow crystalliza-
tion, the wet cake of 10 prior to recrystallization, the DMSO
solution of 10 as it is dissolved for the recrystallization, and
the isolated intermediate itself. The N-methyl 16 was detected
in all samples, indicating that N-methylation occurs during
the reaction rather than during isolation, dissolution into
DMSO, or recrystallization.
Therefore, the first possibility explored was delivery of
the methyl from NMM to form 16. Optimally, labeled NMM
would have been used. As this was not readily available an
alternative test was devised; NMM was removed from the
system, and DBU was used as the base.8 The yield of 9 with
this base was very poor, so quantitation as described above
is not appropriate, but it was clear that some amount of
N-methyl 16 was formed. To avoid the problem of poor yield
with DBU, a different test was devised. Isolated 10 was
treated with CDMT and >2 equiv of base under relatively
extreme conditions (100 °C). These results are reported in
Table 2. The control reaction with NMM produced 0.6%
Table 2. Treatment of Isolated 10 with 1.35 equiv of CDMT
and base at 100 °C
conditions
16 (%) at 1 h
prolonged (%)
CDMT, NMM
0.6
1.4 @ 20 h
1.8 @ 2 days
2.3 @ 5 days
4.1 @ 2 days
0.46 @ 2 days
CDMT, DBU
20, NMM
1.8
0.08
N-methyl 16 in 1 h. Replacement of NMM with DBU
increased the amount of methylation in 1 h to 1.8%.
Surprisingly, the methyl group is not derived from NMM.
With NMM eliminated as the source, CDMT and DMF
remained. 13C2 and D6 labeled CDMT 19 (Figure 4) was
It was then of interest to determine if the N-methylation
occurs during the activation with CDMT at 0 °C, or during
the coupling with LGADE 8. Laboratory reactions were run
using typical and N-methyl spiked 6 and then analyzed
immediately. The N-methyl active ester 17 was detected in
the spiked sample but not in the reaction using typical 6.
Three potential sources of the methyl appear in the peptide
chemistry step. N-methylmorpholine (NMM), CDMT, and
DMF all possess methyl groups. To determine which of these
methyls was the source, HPLC of laboratory reaction mixture
samples was used. The mixtures are heterogeneous, which
makes consistent sampling difficult. To correct for this, the
Figure 4.
(7) Kunishima, M.; Kawachi, C.; Morita, J.; Terao, K.; Iwasaki, F.; Tani, S
Tetrahedron 1999, 55, 13159-13170.
(8) Upon review, use of N-ethylmorpholine instead of DBU was suggested.
740
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development

ambient, a precipitate forms immediately. Removal of an
aliquot after 30 s, filtration, and HPLC analysis shows a
barely detectable (below the quantitation limit) level of
CDMT in solution. When sampled after 30 min, all that is
observed in solution is the decomposition product 18 (eq
1). After 20 h the yield of 18 is near quantitative.
Figure 5.
Only one set of data remains to be understood at this
point: Tables 1 and 2 reveal that on prolonged exposure
further N-methylation occurs, long after the CDMT*NMM
complex has decomposed. At least one other species must
be capable of N-methylation at a very slow rate. The slower
N-methylation is sensitive to stoichiometry. With 1.1 equiv
of CDMT, no further methylation was observed after 20 h.
This suggests that decomposition of the excess CDMT*NMM
produces the weaker methylating agent. The labeling experi-
ment was only conducted for 1 h, so either 18 or the
sideproduct containing the methyl originally on NMM
(presumably methyl chloride) could be the second agent. As
these prolonged reaction times are irrelevant to the manu-
facturing process, and the identity of the second methylating
agent would not impact the control strategy, the identity of
the second agent was not determined.
prepared.9 Reaction under the standard conditions for 1 h
showed complete incorporation of the labels by HPLC/MS
into the N-methyl impurity 16.
This result led to more questions that needed to be
addressed. Clearly the methylating species is related to
CDMT; several possibilities remain. The source could be
the excess CDMT, the active ester 7, the byproduct(s) of
CDMT after the peptide coupling, or perhaps an impurity
common to all samples of CDMT (including the labeled 19).
To completely ensure the appropriateness of our control
strategy, this level of understanding is required. In particular,
if the true methylating agent was an impurity in CDMT, a
specification on the impurity would be established. We
consider it likely that all sources of CDMT, including 19,
are prepared from cyanuric chloride and methanol, so
common impurities are possible.
It is believed at this point that sufficient understanding
has been gained to ensure long-term control of N-methyla-
tion. The CDMT*NMM complex is formed at 0 °C in the
presence of 6. Solubility of the complex is very low but
sufficient to allow the very fast reaction with 6 to form the
active ester 7. The excess of CDMT*NMM is protected from
decomposition by insolubility and temperature. N-methyla-
tion of 6 or 7 does not occur at 0 °C during the time scale
of the process. L-GADE is added, and the mixture is warmed.
As the mixture warms, the excess CDMT*NMM gains
solubility and N-methylates competitively with decomposi-
tion. If held for excessive time, a decomposition product is
also capable of N-methylation.
Based on this understanding of the mechanism of N-
methylation, it is concluded that the current control strategy
is adequate. As long as the CDMT stoichiometry, specified
times, and temperatures during the peptide coupling are
maintained, the level of N-methyl impurity will not increase.
Even a massive excess of CDMT has no impact unless times
are extended or temperatures are dramatically increased.
Variation in quality of CDMT will have no impact, as CDMT
itself is the active agent.
Therefore, determining if an impurity in CDMT was the
actual problem was of primary importance. HPLC/MS of
the lot of CDMT used for these studies revealed (by both
UV and total ion count) only one impurity above the noise
level, 2,4,6-trimethoxy-1,3,5-triazine 20 (Figure 5).
A full equivalent of 2010 was tested under the more
strenuous (100 °C) conditions and found to produce minimal
16 after an hour (Table 2). To provide some assurance that
an impurity was not missed, perhaps due to decomposition
under the HPLC conditions, a sample of CDMT from a
second source (Aldrich) was tested and found to be indis-
tinguishable (Table 1).
It was next determined if the byproduct(s) of CDMT from
the peptide coupling was the methylating reagent. This is
answered by the experiment under the strenuous conditions,
employing isolated 10 as the substrate. Under these condi-
tions CDMT byproducts could not be formed, yet the control
reaction showed substantial N-methyl 16 formation in an
hour. Therefore, CDMT byproducts from the peptide coup-
ling are not required for methylation.
Given that the CDMT*NMM complex is the methylating
agent, understanding the dependence on stoichiometry be-
comes important. Surprisingly, doubling the charges of
CDMT and NMM (to 2.7 equiv) produced no significant
change in the level of N-methyl 16 when sampled after 1 h.
While this constitutes a great result for one designing a
control strategy, it was difficult at first to understand. It is
now believed that a combination of the poor solubility of
the CDMT*NMM complex and the instability of the complex
are responsible for this lack of sensitivity to stoichiometry.
When NMM is added to a DMF solution of CDMT at
Experimental Section
Isolation and Characterization of N-Methyl Peme-
trexed (12). N-Methyl Pemetrexed (12) was isolated from a
production batch of Pemetrexed Disodium Heptahydrate (1)
by preparative reversed-phase HPLC. Repeated injections of
a 100 mg/mL solution of 1 in water were needed. Chroma-
tography was conducted on a Kromasil C18 (50 mm × 250
mm, 10 µm) column. Separation was achieved using am-
monium formate buffer (pH 3.5) and a gradient ramp from
12 to 25% acetonitrile over 25 min. 12 was isolated from
the combined fractions by rotary evaporation, followed by
lyophilization.
(9) Prepared by use of appropriately labeled methanol in the procedure described
in: Cronin, J. S.; Ginah, F. O.; Murray, A. R.; Copp, J. D. Synth. Commun.
1996, 26, 3491-3494.
(10) Obtained from Aldrich.
(11) Willker, W.; Leibfritz, D.; Kerssebaum, R.; Bermel, W. Magn. Reson. Chem.
1993, 31, 287-292.
HRMS found (M + 1) 442.1704, calculated for C21H23N5O6
441.1648 (calculated for M + 1 442.1726). All NMR spectra
Vol. 9, No. 6, 2005 / Organic Process Research & Development
•
741

Table 4. NMR assignments for 14
atom
number
(1H)
ppm
(13C)
ppm
type
confirmation
1
>CdO n/a
166.2 HMBC
1-OCH3 CH3
3.82
51.75 1H shift, HSQC, HMBC
127.13 HMBC
Figure 6. Numbering scheme for NMR assignments in Table
3.
2
>Cd n/a
3
arylCH 7.87
arylCH 7.36
>Cd n/a
arylCH 7.36
arylCH 7.87
138.85 1H shift, HSQC, HMBC
128.47 1H shift, HSQC, HMBC
148.05 HMBC
d
d
4
5
6
128.47 1H shift, HSQC, HMBC
128.85 1H shift, HSQC, HMBC
35.88 1H shift, HSQC, HMBC
26.88 1H shift, HSQC, HMBC
119.63 HMBC
d
d
m
m
7
8
CH2
CH2
2.98
2.92
9
10
11
12
>Cd n/a
arylCH 6.61
NH
114.46 1H shift, HSQC, HMBC
n/a
s
s
Figure 7. Numbering scheme for NMR assignments in Table
4.
not
observed
13
14
>Cd n/a
N
139.61 HMBC
n/a
Table 3. NMR assignments for N-methyl Pemetrexed (12)
n/a
3.52
32.18 1H shift, HSQC, HMBC
152.38 HMBC
n/a
atom
number type
(1H)
ppm
(13C)
ppm
14-NCH3 CH3
15
16
17
18
>Cd n/a
confirmation
N
n/a
>CdO n/a
163.16 HMBC
100.54 HMBC
1
2
3
>CdO n/a
166.22 13C shift, HMBC11
145.94 13C shift, HMBC
127.33 1H shift, 13C shift, HSQC11, d
HMBC
>Cd n/a
>Cd n/a
arylCH 7.8
4
arylCH 7.3
128.21 1H shift, 13C shift, HSQC,
HMBC
d
h. Additional methyl iodide (0.2 mL, 3.2 mmol) and
triethylamine (0.2 mL, 1.4 mmol) were added and the
resulting solution was allowed to stir at 24 °C for 17 h.
Additional methyl iodide (0.2 mL, 3.2 mmol) and triethyl-
amine (0.2 mL, 1.4 mmol) were added, and the result-
ing solution was stirred at 24 °C for 20 h. At this point
HPLC analysis showed no further reaction occurred
from the last addition. Water (40 mL) was added to the
reaction solution and extracted with dichloromethane (3 ×
40 mL). The organic layer was washed with saturated brine
(50 mL), dried (Na2SO4), filtered, and concentrated to an
oil. Sodium hydroxide (1 N, 13.4 mL, 13.4 mmol) was added
to the oil at 24 °C under nitrogen, and the resulting slurry
was stirred for 30 min. To the resulting solution, water (13.4
mL) and ethanol (13.4 mL) were added. The pH of the
solution was adjusted to 3.5 with HCl (3.2 N). The resulting
slurry was heated to 65 °C, and dissolution occurred. The
hot solution was allowed to cool to 24 °C. The resulting
slurry was stirred at 24 °C for 30 min and then at 0-5 °C
for 60 min. The green slurry was filtered and dried (ambient
pressure and 24 °C) to afford 12 (0.97 g, 65%). Preparative
chromatography was used to prepare the analytical sample.
Preparation of N-Methyl Intermediates (14, 16). Meth-
odology similar to that described above for 12, omitting the
saponification step, was used to prepare 14. Assignments
are made against the numbering scheme shown below (Figure
7). Spectra, assignments, and confirming experiments are
summarized in Table 4. A DMF solution of intermediate
9enriched in 16 was prepared by treatment of 10 under
similar conditions.
5
6
>Cd n/a
arylCH 7.3
131.57 13C shift, HMBC
128.21 1H shift, 13C shift, HSQC,
HMBC
d
7
8
9
arylCH 7.8
127.33 1H shift, 13C shift, HSQC,
HMBC
d
CH2
CH2
2.98
2.9
36.12 1H shift, 13C shift, HSQC,
HMBC
m
m
27.52 1H shift, 13C shift, HSQC,
HMBC
10
11
>Cd n/a
arylCH 6.49
119.61 13C shift, HMBC
113.58 1H shift, 13C shift, HSQC,
HMBC
s
12
NH not
n/a
observed
13
14
>Cd n/a
139.25 13C shift, HMBC
n/a
N
n/a
3.55
14-CH3 CH3
32.06 1H shift, 13C shift, HSQC,
HMBC
s
15
16
17
18
1′
>Cd n/a
152.44 13C shift, HMBC
n/a
N
n/a
>CdO n/a
>Cd n/a
>CdO n/a
164.17 13C shift, HMBC
101.19 12C shift, HMBC
174.2 13C shift, HMBCa
52.25 1H shift, 13C shift, HSQC,
HMBC
2′
CH1
CH2
CH2
4.36
m
m
m
3′
4′
5′
2.02, 1.92 26.44 1H shift, 13C shift, HSQC,
HMBC
2.3
30.83 1H shift, 13C shift, HSQC,
HMBC
>CdO n/a
174.04 13C shift, HMBCa
a Insufficient resolution to distinguish.
were collected on an Avance 300 spectrometer equipped with
a 5 mm broad-band inverse (BBI) detection probe, with the
exception of the Survey 13C spectrum which was collected
with a QNP probe. All spectra were collected in DMSO-d6.
Assignments are made against the numbering scheme shown
in Figure 6. Spectra, assignments, and confirming experi-
ments are summarized in Table 3.
Independent Synthesis of N-Methyl Pemetrexed (12,
LSN2071949). Methyl iodide (0.75 mL, 12.1 mmol) was
added to a stirred slurry of Pemetrexed Disodium Heptahy-
drate 1 (2.0 g, 3.35 mmol) and triethylamine (0.75 mL, 5.38
mmol) in DMF (25 mL) at 24 °C under a nitrogen
atmosphere. The resulting slurry was stirred at 24 °C for 3
Acknowledgment
The authors thank Shengming Ma of the Shaghai Institute
of Organic Chemistry for the preparation of 13. We thank
Professors William R. Roush and Marvin J. Miller for their
consultation.
Received for review November 3, 2004.
OP0498013
742
•
Vol. 9, No. 6, 2005 / Organic Process Research & Development