The design and synthesis of a new tumor-selective fluoropyrimidine carbamate, Capecitabine

Article abstract of DOI:10.1016/S0968-0896(00)00087-0

To identify an orally available fluoropyrimidine having efficacy and safety profiles greatly improved over those of parenteral 5-fluorouracil (5-FU: 1), we designed a 5-FU prodrug that would pass intact through the intestinal mucisa and be sequentially converted to 5-FU by enzymes that are highly expressed in the human liver and then in tumors. Among various N⁴-substituted 5'-deoxy-5-fluorocytidine derivatives, a series of N⁴-alkoxycarbonyl derivatives were hydrolyzed to 5'-deoxy-5-fluorocytidine (5'-DFCR: 8) specifically by carboxylesterase, which exists preferentially in the liver in humans and monkeys. Particularly, derivatives having an N⁴-alkoxylcarbonyl moiety with a C4-C6 alkyl chain were the most susceptible to the human carboxylesterase. Those were then converted to 5'-deoxy-5-fluorouridine (5'-DFUR: 4) by cytidine deaminase highly expressed in the liver and solid tumors and finally to 5-FU by thymidine phosphorylase (dThdPase) preferentially located in tumors. When administered orally to monkeys, a derivative having the N⁴-alkoxylcarbonyl moiety with a C5 alkyl chain (capecitabine: 6) The highest AUC and Cmax for plasma 5'-DFUR. In tests with various human cancer xenograft models, capecitabine was more efficacious at wider dose ranges than either 5-FU or 5'-DFUR and was significantly less toxic to the intestinal tract than the others in monkeys. Copyright (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(00)00087-0

Bioorganic & Medicinal Chemistry 8 (2000) 1697±1706
The Design and Synthesis of a New Tumor-Selective
Fluoropyrimidine Carbamate, Capecitabine
Nobuo Shimma,a,* Isao Umeda,a Motohiro Arasaki,a Chikako Murasaki,a
Kazunao Masubuchi,a Yasunori Kohchi,a Masanori Miwa,b Masako Ura,b
Noriaki Sawada,b Hitoshi Tahara,c Isamu Kuruma,c Ikuo Horiid and Hideo Ishitsukab
aDepartment of Chemistry, Nippon Roche Research Center, 200 Kajiwara, Kamakura City, Kanagawa 247-8530, Japan
bDepartment of Oncology, Nippon Roche Research Center, 200 Kajiwara, Kamakura City, Kanagawa 247-8530, Japan
cDepartment of Pharmacokinetics, Nippon Roche Research Center, 200 Kajiwara, Kamakura City, Kanagawa 247-8530, Japan
dDepartment of Toxicology, Nippon Roche Research Center, 200 Kajiwara, Kamakura City, Kanagawa 247-8530, Japan
Received 19 January 2000; accepted 10 March 2000
AbstractÐTo identify an orally available ¯uoropyrimidine having ecacy and safety pro®les greatly improved over those of par-
enteral 5-¯uorouracil (5-FU: 1), we designed a 5-FU prodrug that would pass intact through the intestinal mucisa and be sequen-
tially converted to 5-FU by enzymes that are highly expressed in the human liver and then in tumors. Among various N4-
substituted 50-deoxy-5-¯uorocytidine derivatives, a series of N4-alkoxycarbonyl derivatives were hydrolyzed to 50-deoxy-5-¯uoro-
cytidine (50-DFCR: 8) speci®cally by carboxylesterase, which exists preferentially in the liver in humans and monkeys. Particularly,
derivatives having an N4-alkoxylcarbonyl moiety with a C4±C6 alkyl chain were the most susceptible to the human carboxylester-
ase. Those were then converted to 50-deoxy-5-¯uorouridine (50-DFUR: 4) by cytidine deaminase highly expressed in the liver and
solid tumors and ®nally to 5-FU by thymidine phosphorylase (dThdPase) preferentially located in tumors. When administered
orally to monkeys, a derivative having the N4-alkoxylcarbonyl moiety with a C5 alkyl chain (capecitabine: 6) The highest AUC and
Cmax for plasma 50-DFUR. In tests with various human cancer xenograft models, capecitabine was more ecacious at wider dose
ranges than either 5-FU or 50-DFUR and was signi®cantly less toxic to the intestinal tract than the others in monkeys. # 2000
Elsevier Science Ltd. All rights reserved.
Introduction
genous enzymes preferentially localized in human liver
and tumor tissues.
Most antitumor agents are cytotoxic and are associated
with severe side e€ects, including myelo-suppression,
intestinal toxicity and skin toxicity. There have been
several attempts to reduce such side e€ects by means of
prodrug strategies for the tumor-selective delivery of
cytotoxic agents,1 such as (a) prodrugs activated either
by endogenous enzymes preferentially localized in
tumor tissues or (b) exogenous enzymes speci®cally
introduced into tumor tissues by gene technology, and
(c) conjugates of cytotoxic drugs and molecules selec-
tively bound to tumors. To our knowledge, however,
none of them has yet to be used clinically. The present
study describes the tumor-selective delivery of 5-¯uoro-
uracil (5-FU: 1) by sequential conversion of the new
¯uoropyrimidine carbamates (7: R=alkoxy) by endo-
5-FU (1) is one of the antitumor agents most frequently
used for treating solid tumors, such as breast, color-
ectal, and gastric cancers, in either monotherapy or
combination therapy with various cytotoxic drugs and
modulators, such as leucovorin and methotrexate.2 To
optimize the ecacy of 5-FU (1), it is often adminis-
tered by continuous infusion,3 because of its short
plasma half-life, as well as in combination with other
cytotoxics or with biochemical modulators, such as leu-
covorin. 5-FU is poorly tumor selective, however, and
its therapy causes high incidences of toxicity in the bone
marrow, gastrointestinal tract, central nervous system
and skin. Several oral 5-FU derivatives, such as tegafur
(2), carmofur (3), and doxi¯uridine (50-DFUR: 4) (Fig.
1), have been developed in Japan to improve such side
e€ects and are being used clinically in several countries.
Nevertheless, they still need to be improved in terms of
tumor selectively, ecacy, and safety.2
*Corresponding author. Tel.: +81-467-47-2280; fax: +81-467-45-
6824; e-mail: nobuo.shimma@roche.com
0968-0896/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(00)00087-0

1698
N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
Figure 1. 5-Fluorouracil and its prodrugs.
50-DFUR (4), which we previously developed for the
treatment of breast, colorectal and other cancers, shows
a tumor-selective activity as a result of its selective con-
version to 5-FU by pyrimidine nucleoside phosphor-
ylases (PyNPase) preferentially localized in tumors:
uridine phosphorylase in mice4 and thymidine phos-
phorylase (dThdPase) in humans.5,6 These enzymes,
however, also exist in the intestinal tract,4,5 and generate
5-FU there. As a consequence, 4 causes its dose-limiting
toxicity, diarrhea, when orally given at high doses for
long periods.7 To improve on the ecacy and safety
pro®les of the oral ¯uoropyrimidine 50-DFUR (4),
capecitabine (N4-pentyloxycarbonyl-50-deoxy-5-¯uoro-
cytidine; 6), a new orally available ¯uoropyrimidine
carbamate, was synthesized so that it passes intact
through the intestinal mucosa and then selectively
delivers 5-FU to tumor tissues following its enzymatic
conversion: ®rst to 50-deoxycytidine (50-DFCR: 8) by
carboxylesterase in the liver, then to 50-deoxy-5-¯uor-
ouridine (50-DFUR: 4) by cytidine deaminase,8 and
®nally to 5-FU by dThdPase. The latter two enzymes
are preferentially localized in tumor tissues.5 It has been
approved in the USA, Canada, and other countries for
the treatment of metastatic breast cancer that is resis-
tant to standard chemotherapy with paclitaxel and an
anthracycline-containing regimen.
cancer tissues.5 The enzyme activity was also reported
to be low in growing bone marrow cells.12 We, there-
fore, assumed that 50-DFCR (8) would be eciently
converted to 50-DFUR (4) in solid tumors, showing
tumor selective action, but only slightly converted in the
growing bone marrow cells, without severe bone mar-
row toxicity. In addition, acylation of the N4-amino
group of 50-DFCR could no longer render it a substrate
for dThdPase in the intestinal tissues, allowing it to pass
through the intestinal mucosa as an intact molecule
without causing intestinal toxicity. The key to the suc-
cess of this approach was to ®nd a liver-speci®c enzyme
for the hydrolysis of the N4-acyl group (Scheme 1).
Our strategy was to identify N4-(substituted)-50-DFCR
derivatives 7 that pass intact through the intestinal
mucosa and then are metabolized to 50-DFCR (8) in the
liver. First, we screened various N4-substituted 50-
DFCR derivatives to identify compounds susceptible
only to enzymes in the liver, but not to those in the
intestine or other organs, from humans and monkeys.
Among such compounds identi®ed, we then selected
compounds with high oral bioavailability in monkeys
while, in parallel, assessing their antitumor ecacy in
human cancer mouse xenograft models and their intest-
inal toxicity in monkeys. We selected human and mon-
key enzyme systems for this screening, because we had
previously found a signi®cant species di€erence in the
enzyme susceptibility of N4-substituted-50-DFCR deri-
vatives between rodents and primates. We previously
synthesized galocitabine (50-deoxy-5-¯uoro-N4-(3,4,5-
trimethoxybenzoyl)cytidine: 5) (Fig. 2), which was
selected from among various types of N4-acyl-50-DFCR
derivatives synthesized based on its ecacy and safety
in mouse tumor models. Galocitabine (5), however, had
low bioavailability in humans and monkeys; it was not
well converted to 50-DFCR in these species, although it
generated 50-DFCR and was highly e€ective and safe in
mice.13
In this paper, we describe our studies on the design,
synthesis, and assessment of N4-(substituted)-50-deoxy-
5-¯uorocytidine derivatives, including capecitabine.
Design and Synthesis of Prodrugs
Design
We set out to design 50-deoxycytidine (8)-type prodrugs
of doxi¯uridine (4) with high tumor-selective activity
based on our experience and knowledge of anticancer
drug development. It is well known that Ara-C (9), a
potent DNA polymerase a inhibitor, is active against
leukemia but inactive against solid tumors.9 This e-
cacy pro®le was suggested to result from the physio-
logical localization of cytidine deaminase, which
converts Ara-C to its inactive metabolite Ara-U (10),
whereby low levels are found in leukemic cells and high
levels in solid tumors.10,11 Moreover, we observed that
the enzyme activity was higher in various types of human
cancer tissues than in the healthy tissues adjacent to the
Synthesis
The synthesis of the new N4-(substituted)-50-DFCR (7)
is outlined in Scheme 2. Various N4-(acyl or alkoxy-
carbonyl)-50-DFCR (7) derivatives were synthesized
from 5-¯uorocytosine (11) in 3 steps: (i) glycosidation of
11 with 1,2,3-tri-O-acetyl-5-deoxyribose14 with stannic
tetrachloride15 in dichloromethane (at 15±20 ꢀC) or in
situ generated trimethysilyl iodide16 in acetonitrile (at

N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
1699
Scheme 1. Design of cytidine-type prodrugs of 5-FU.
Figure 3 shows the initial screening results for suscept-
ibility of all the 50-DFCR derivatives to the crude
enzyme extract from human liver and intestine; the
detailed data for the N4-(C1 to C9)alkoxycarbonyl
derivatives are shown in Table 1. The carbamate deriv-
atives showed higher speci®city to human liver enzyme
over the intestinal enzyme, while most of the amide
derivatives showed very poor speci®city except for a few
derivatives. Namely, all the carbamate derivatives hav-
ing a normal alkyl, branched chain alkyl, cycloalkyl, or
aralkyl group in the structure were susceptible only to
the liver enzymes. In contrast, only the three amide
derivatives having a 3,5-disubstituted benzoyl group,
namely compounds 22 (N4-3,5-dimethoxybenzoyl), 23
(N4-3,5-dimethylbenzoyl), and 24 (N4-3,5-dichloro-
benzoyl) showed preferential susceptibility to the liver
enzymes; their susceptibilities (nmol/mg protein/h),
liver/intestine, were 32/<10, 270/22, and 38/<10,
respectively (Fig. 3, Table 1).
Figure 2.
0 ꢀC), (ii) acylation of the N4-amino group of the
resulting 50-DFCR derivative 12 with acid chlorides or
alkoxycarbonyl chlorides in 2 equiv pyridine/dichloro-
methane, and (iii) alkaline hydrolysis of the acetyl group
of 13 at 0 ꢀC.
Figure 4 shows the relationship between the alkyl chain
length of the carbamates (only normal alkyl: C1±C9)
and the susceptibility to the human and monkey car-
boxylesterases (crude enzymes from the liver and the
intestine). The carbamate derivatives showed extremely
high speci®city to the liver enzyme over the intestinal
enzyme for both humans and monkeys, suggesting their
ecient biotransformation to 50-DFCR in the liver after
passing intact through the intestinal mucosa when given
orally to humans.
Results and Discussion
Structure±enzyme susceptibility relationship
The susceptibility of N4-(substituted)-50-DFCR deriva-
tives to enzymes converting them to 50-DFCR was
measured as described in our previous report.5 Brie¯y,
the derivatives were incubated with extracts from
human and monkey tissue samples in the presence of a
cytidine deaminase inhibitor, tetrahydrouridine, at
37 ꢀC for 60 min, and the amount of the product 50-
DFCR was measured by HPLC.
The optimal chain lengths for the human liver enzyme
were C5 and C6, while that for the monkey enzyme was
C8.
As a reference, the susceptibilities of the same carbamate
derivatives to the mouse liver and intestinal enzymes are

1700
N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
Scheme 2. Synthesis of N4-(acyl or alkoxycarbonyl)-50-DFCR derivatives.
Figure 3. Susceptibility of 50-DFCR derivatives to human liver and intestinal carboxylesterase (crude enzyme).
shown in Figure 5, which emphasizes the species di€er-
ence in the enzyme susceptibility as compared with that
in humans and monkeys. Actually, no liver enzyme
selectivity was seen in mice. Therefore, it should be
emphasized that the carbamate type prodrugs could
never have been selected without human enzyme assay
systems (Figs 4 and 5).
Figure 6 and Table 1, N4-(alkoxycarbonyl)-50-DFCR
derivatives with the appropriate susceptibility to car-
boxylesterase showed better pharmacokinetic pro®les.
Namely, those having C4 (17), C5 (6), and C6 (18) alkyl
chains in the N4-alkoxycarbonyl moiety, whose enzyme
susceptibilities range from 20 to 50 nmol/mg protein/h,
gave high 50-DFUR AUC and Cmax. Among them, N4-
(n-pentyloxycarbonyl)-50-DFCR (capecitabine: 6) with
the highest AUC and Cmax of 50-DFUR was selected as
a candidate for further development (Fig. 6).
Pharmacokinetic pro®les in monkeys
Correlation between the Cmax or AUC of 50-DFUR (4)
and the susceptibility to the monkey liver carboxyl-
esterase as well as the alkyl chain length of the carbamates.
The enzyme susceptibilty and the PK parameters of
capecitabine (6) were compared with those of galocita-
bine (5) after single oral administration of 5 and 6
at 45.5 mmol/kg in monkeys (Table 2). Capecitabine
showed clearly better oral bioavailability than galocita-
bine: higher Cmax and AUC value of 50-DFUR (4) (5.3-
fold higher Cmax and 4.6-fold higher AUC as compared
with galocitabine administration) and lower levels of
intact molecule in the plasma and urine were achieved
with the oral capecitabine. The pharmacokinetic pro®les
of capecitabine indicate that it can pass through the
intestinal mucosa unchanged and then be hydrolyzed
into 50-DFCR (8) by the hepatic carboxylesterase.5 In
separate experiments, the higher tumor selectivity of
The N4-(alkoxycarbonyl)-50-DFCR derivatives 7 were
orally administered into groups of ®ve cynomolgous
monkeys. At various times after the administration,
blood was withdrawn for determination of plasma con-
centrations of intact molecules and their active metabo-
lite 50-DFUR by the HPLC method. Figure 6 shows the
2nd order correlation between the Cmax or AUC of 50-
DFUR (4) generated from the N4-(alkoxycarbonyl)-50-
DFCR derivatives in monkeys and their susceptibility
to the monkey liver carboxylesterase as well as that of
the alkyl chain length of the carbamates. As shown in

N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
1701
Table 1. Susceptibility of 50-DFCR derivatives to carboxylesterase and their pharmacokinetic pro®les in monkeys
Susceptibility to carboxylesterase (nmol/mg protein/h)
Mouse Monkey Human
Intestine Intestine Intestine
Pharmacokinetic pro®le in monkeysa
Drug concentration in plasma
Liver
Liver
Liver
Intact drug
50-DFUR (4)
AUCc
Cmax
AUCc
b
b
Compound no.
R
Cmax
14
15
16
17
OCH3
10
10
24
41
38
44
55
43
36
280
10
28
93
160
99
76
100
42
34
<10
2
5
<10
<10
<10
<10
13
17
16
16
<10
<0.5
10
10
35
71
190
220
110
73
<10
<10
<10
<10
<10
<10
<10
<10
<10
<0.8
7.2
9.7
6.9
3.3
2.4
1.8
1.2
1.2
0.4
33.0
17.5
19.3
16.9
4.5
2.1
2.1
3.9
3.2
1.4
66.6
0.1
0.4
1.7
2.0
2.3
1.9
0.6
0.4
0.3
0.4
0.3
0.8
2.6
2.6
2.8
2.6
1.0
1.2
0.7
0.7
OCH2CH3
O(CH2)2CH3
O(CH2)3CH3
10
20
29
47
59
66
61
1.4
Capecitabine (6) O(CH2)4CH3
18
19
20
21
O(CH2)5CH3
O(CH2)6CH3
O(CH2)7CH3
O(CH2)8CH3
56
3.7
Galocitabine (5)
22
23
1300
1500
1800
7200
14
67
16
22
32
<10
22
10.0
6.1
20.9
10.0
0.9
1.5
1.7
2.2
270
24
270
82
18
<10
38
<10
3.2
6.0
1.0
2.8
aSingle oral administration of 5 and 6 (45.5 mmol/kg).
bCmax mg/mL.
cAUC: mg h/mL.
orally administered (po) capecitabine versus 5-FU
(ip) in the HCT116 human colon cancer xenograft
model in mice was actually con®rmed by measuring
5-FU levels in tumor, muscle, and plasma after a single
administration.17 In accordance with these results,
capecitabine showed only mild toxicity to both the
intestinal tract and bone marrow in monkeys. The
degree of toxicity, which will be reported in detail else-
where, was less than that of 50-DFUR administration
(po), even though capecitabine gave higher AUC and
Cmax of 50-DFUR in the plasma.18 Moreover, capecita-
bine exhibited higher ecacy than 50-DFUR and 5-FU in
various human cancer xenograft models in nude mice,
including colon cancer/CXF280, HCT116; gastric cancer/
GXF97, MKN-45; and breast cancer/MX-1, MAXF401.19
This sequential metabolic activation has also been
supported in humans by the results of clinical trials.20
The antitumor activity of capecitabine in mice depended
on the activity of thymidine phosphorylase (dThdPase)
in tumor tissues as well as the ratio of dThdPase and
dihydropyrimidine dehydrogenase, a catabolic enzyme
of 5-FU.21 Some in¯ammatory cytokines, such as
TNFa, IFNg, and IL-1a, up-regulated the expression of
dThdPase in tumor cells making the tumor cells more
susceptible to 50-DFUR.22 Consequently, cancer treat-
ments that up-regulate these cytokines, such as taxane
(taxotere/taxol) administration23 and X-ray irradia-
tion,24 enhanced the dThdPase levels in human cancer
xenografts and the ecacy of 50-DFUR and its pre-
cursor capecitabine. Thus, if this process were also the
case in humans, the ecacy of capecitabine would be
optimized by selecting appropriate patients whose
tumors highly express dThdPase or by combining it
with dThdPase up-regulators.

1702
N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
Figure 4. Correlation between the chain length and the susceptibility to carboxylesterase (monkey and human).
Figure 5. Correlation between the chain length and the susceptibility to carboxylestesterase (mice).
Figure 6. Correlation between enzyme susceptibility and Cmax or AUC of 50-DFUR (4) after oral administration of the C1±C9 normal alkyl car-
bamates (20 mg/kg) in monkeys.

N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
Table 2. Pharmacokinetic pro®le of capecitabine (6) and galocitabine (5)a
1703
Drug concentration in plasma
Urinary excretiond
<5-FUe
Intact
50-DFCR (8)
50-DFUR (4)
Intact drug
b
b
b
Compound no.
Cmax
AUCc
Cmax
AUCc
Cmax
AUCc
Capecitabine (6)
Galocitabine (5)
2.3
33
2.0
67
0.15
0.76
0.36
1.88
2.1
0.41
2.9
0.69
<0.2
27
66
34
aSingle oral administration of 5 or 6 (45.5 mmol/kg) in monkeys.
bCmax: (mg/mL).
cAUC: mg h/mL.
d% of dose.
e<5-FU: metabolites after 5-FU.
Figure 7. Conversion of capecitabine into 5-FU in humans.
Conclusion
and high resolution FAB±MS spectra (matrix: m-nitro-
benzoic acid) were obtained on a JEOL SX102/102.
Melting points were determined on a micro melting
point apparatus YANACO model MP. All products
were judged to be >98% pure by reverse-phase HPLC
analysis with diode array detection. 5-Fluorocytosine
was purchased from Aldrich.
We identi®ed a new ¯uoropyrimidine carbamate, cap-
ecitabine, which would be sequentially converted to
50-DFCR by carboxylesterase in the liver, to 50-DFUR
by cytidine deaminase highly expressed in the liver and
tumor tissues, and ®nally to 5-FU by dThdPase pref-
erentially located in various types of cancer tissues
(Fig. 7). As a result of the unique localization of these
enzymes in humans, capecitabine is expected to deliver
the active drug 5-FU selectively to tumor tissues and to
have better ecacy and safety pro®les than do 5-FU
and doxi¯uridine. This study thus demonstrates that the
approach utilizing enzymes that are speci®cally expressed
in certain tissues or tumors for prodrug activation is useful
for the targeting of cytotoxic drugs to tumor tissues.
20,30-Di-O-acetyl-50-deoxy-5-¯uorocytidine (12). Method
(a). 5-Fluorocytosine (25.8 g, 0.2 mol) was suspended in
toluene (103 mL) and hexyamethyl-disilazane (32.3 g,
0.2 mol). The mixture was heated at 100 ꢀC for 3 h. After
concentrating the reaction mixture under reduced pres-
sure, methylene chloride (330 mL) and 5-deoxy-1,2,3-tri-
O-acetyl-b-d-ribofuranoside (59.3 g, 0.23 mol) were
added to the residue. Then, anhydrous stannic chloride
(62.5 g, 0.24 mol) was added dropwise to the ice-cooled
reaction mixture over a period of 10 min. After stirring
the mixture at room temperature for an additional 2 h,
sodium bicarbonate (101 g, 1.2 mol) was added, fol-
lowed by the dropwise addition of water (35 mL) over a
period of 20 min. After stirring the resulting mixture at
Experimental
NMR spectra were obtained on a JEOL SX270 spectro-
meter and are referenced to Me4Si. FAB Mass spectra

1704
N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
room temperature for 3 h, insoluble material was ®ltered
o€ and the ®ltrate was washed with 4% sodium bicar-
bonate solution (100 mL). The solvent was removed
under reduced pressure, and the residue was recrys-
tallized from isopropanol (180 mL) to give 50-deoxy-
20,30-di-O-acetyl-5-¯uorocytidne (12) (49.9 g (y. 76%)):
mp 191.5±192.3 ꢀC. HRMS [FAB, MH+] calcd for
d, J=6.3 and 4.6 Hz), 5.80 (1H, d, J=4.6 Hz), 8.28 (1H,
br. s), 10.63 (1H, br. s).
50-Deoxy-5-¯uoro-N4-(pentyloxycarbonyl)cytidine, capeci-
tabine (6). The crude 20,30-di-O-acetyl-50-deoxy-5-¯uoro-
N4-(pentyloxycarbonyl)cytidine, 13a (300 g) obtained
above was dissolved in MeOH (300 mL). To the stirred
mixture on an ice±salt bath, a solution of NaOH
(108.4 g, 2.71 mol) in water (150 mL) was added drop-
wise over a period of 35 min so that the reaction tem-
perature was kept between 20 and 5 ꢀC. Then, to the
reaction mixture on the ice±salt bath, concentrated HCl
(ca. 226 mL) was added dropwise over a period of
25 min with stirring and cooling between 20 and
5 ꢀC. Then the pH of the reaction mixture was adjus-
ted from 4 to 5. The mixture was partitioned between
CH2Cl2 (1500 mL) and water (300 mL). The organic
layer was washed with water (300 mL), dried over
anhydrous Na2SO4 and ®ltered. The ®ltrate was evapor-
ated to dryness under reduced pressure. The oily resi-
due was dissolved in AcOEt (200 mL). To the solution,
n-hexane (400 mL) was added dropwise over a period of
20 min with stirring, and the solution was allowed to
stand at room temperature for 4 h. The colourless nee-
dle crystals precipitated were collected by ®ltration and
washed with a mixed solvent of n-hexane:AcOEt (5:2)
(900 mL) to give 50-deoxy-5-¯uoro-N4-(pentyloxycarb-
onyl)cytidine (6) (170 g: y. 75% from 20,30-di-O-acetyl-
50-deoxy-5-¯uorocytidine, (12)) as colorless needle crystals:
mp 119±121 ꢀC, [FAB, MH+] calcd for C15H23FN3O6
1
C13H17FN3O6 m/z 330.1097, found 330.1119; H NMR
(DMSO-d6) d 1.33 (3H, d, J=6.6 Hz), 2.04 (3H, s), 2.06
(3H, s), 4.04 (1H, d. q, J=6.6 Hz), 5.09 (1H, t,
J=6.6 Hz), 5.43 (1H, d. d, J=6.6 and 5.1 Hz), 5.76 (1H,
d, J=5.1 Hz), 7.67 (1H, br. s), 7.94 (1H, br. s), 8.01 (1H,
d, J=7.3 Hz).
Method (b). A solution of sodium iodide (3.6 g,
24 mmol) and chlorotrimethylsilane (3 mL, 24 mmol) in
dry acetonitrile (15 mL) was stirred with molecular
sieves 4A (200 mg) at 0 ꢀC for 5 min (colorless
sodium chloride deposited during stirring). 1,2,3-Tri-O-
acetyl-5-deoxy-b-d-ribofuranose (2.0 g, 7.7 mmol) was
added and the mixture was stirred at 0 ꢀC for 30 min.
Then, a solution of the trimethylsilylated 5-¯uoro-
cytosine, freshly prepared from 5-¯uorocytosine (1.12 g,
8.68 mmol), in dry acetonitrile (5 mL) was added at 0 ꢀC
and stirring was continued for 3 h at room temperature.
The mixture was ®ltered, the ®ltrate was concentrated in
vacuo, and the residue was partitioned between di-
chloromethane and saturated aq. sodium bicarbonate
solution. The aqueous layer was extracted with
CH2Cl2:MeOH (10:1). The combined organic layers
were dried over anhydrous sodium sulfate and evapo-
rated under reduced pressure. The residue was puri®ed
by silica gel chromatography using CH2Cl2:MeOH
(15:1) as an eluent, followed by recrystallization from
isopropanol to give 1.24 g (yield 49%) of 20,30-di-O-ace-
tyl-50-deoxy-5-¯uorocytidine.
1
m/z 360.1565, found 360.1592; H NMR (DMSO-d6) d
0.88 (3H, t, J=6.6 Hz), 1.22±1.42 (4H, m), 1.30 (3H, d,
J=6.3 Hz), 1.61 (2H, t. t, J=6.6 Hz), 3.67 (1H, q, J=
5.6 Hz), 3.87 (1H, d. q, J=6.3 Hz), 4.04±4.11 (3H, m),
5.02 (1H, d, J=5.6 Hz), 5.37 (1H, d, J=5.3 Hz), 5.66
(1H, d, J=3.6 Hz), 8.00 (1H, br. s), 10.54 (1H, br. s).
20,30-Di-O-acetyl-50-deoxy-5-¯uoro-N 4-(pentyloxycarbon-
yl)cytidine (13a). 20,30-Di-O-acetyl-50-deoxy-5-¯uoro-
cytidine (800 g, 2.43 mol) was dissolved in dry CH2Cl2
(2000 mL) and dry pyridine (394 mL). To the stirred
solution on an ice-salt bath, n-pentyl chloroformate
(493 mL, 3.38 mol) was added dropwise over a period of
90 min so that the reaction temperature was kept
between 20 and 5 ꢀC. After stirring the mixture for
an additional 30 min at room temperature, MeOH
(30 mL) was added in one portion, and the mixture was
evaporated to a heavy syrup under reduced pressure. To
the heavy syrup residue, Et2O (1500 mL) was added,
and the suspension was stirred for 10 min at room tem-
perature. The insoluble white powder (pyridine hydro-
chloride) was ®ltered o€ and washed with Et2O
(2000 mL). The ®ltrate and washings were combined,
dried over anhydrous sodium sulfate and ®ltered. The
solvent was removed under reduced pressure to give the
crude 20,30-di-O-acetyl-50-deoxy-5-¯uoro-N4-(pentyloxy-
carbonyl)cytidine, 13a (1311 g) as a pale yellow syrup.
The following compounds were obtained according to
the method similar to that described above.
50-Deoxy-5-¯uoro-N4-(methoxycarbonyl)cytidine (14).
HRMS [FAB, MH+] calcd for C11H15FN3O6 m/z
1
304.0911, found 304.0970; H NMR (DMSO-d6) d 1.31
(3H, d, J=6.6 Hz), 3.66±3.69 (4H, m), 3.89 (1H, m),
4.06 (1H, m), 5.05 (1H, J=5.9 Hz), 5.41 (1H, d, J=
5.1 Hz), 5.67 (1H, d, J=3.7 Hz), 8.03 (1H, br. s), 10.60
(1H, br. s).
50-Deoxy-5-¯uoro-N4-(ethoxycarbonyl)cytidine (15). Mp
119.5±121 ꢀC, HRMS [FAB, MH+] calcd for C12H17
FN3O6 m/z 318.1097, found 318.1143; 1H NMR (DMSO-
d6) d 1.23 (3H, t, J=6.9 Hz), 1.30 (3H, d, J=6.3 Hz), 3.67
(1H, m), 3.89 (1H, d.q, J=6.3 Hz), 4.06 (1H, d.d.d, J=
4.6 Hz), 4.13 (2H, q, J=6.9Hz), 5.03 (1H, d, J=5.9 Hz),
5.38 (1H, d, J=4.6 Hz), 5.66 (1H, d, J=4.0 Hz), 7.99 (1H,
br. s), 10.96 and 11.75 (total 1H, br. s each).
50-Deoxy-5-¯uoro-N4-(propoxycarbonyl)cytidine (16). Mp
125±126.5 ꢀC, HRMS [FAB, MH+] calcd for C13H19
FN3O6 m/z 332.1253, found 332.1280; 1H NMR
(DMSO-d6) d 0.91 (3H, t, J=7.3 Hz), 1.30 (3H, d, J=
6.3 Hz), 1.62 (2H, q. t, J=7.3 Hz), 3.67 (1H, q, J=
5.9 Hz), 3.89 (1H, d. q, J=5.9 Hz), 4.02±4.07 (3H, m),
HRMS [FAB, MH+] calcd for C19H27FN3O8 m/z
1
444.1775, found 444.1752; H NMR (DMSO-d6) d 0.88
(3H, t, J=7.3 Hz), 1.30±1.34 (4H, m), 1.36 (3H, d,
J=6.3 Hz), 1.61 (2H, m), 2.06 (3H, s), 2.07 (3H, s),
4.07±4.14 (3H, m), 5.11 (1H, t, J=6.3 Hz), 5.47 (1H, d.

N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
1705
5.02 (1H, d, J=5.9 Hz), 5.38 (1H, d, J=5.3 Hz), 5.66
(1H, d, J=3.6 Hz), 8.00 (1H, br. s), 10.58 (1H, br. s).
(1H, d, J=5.9 Hz), 5.41 (1H, d, J=4.9 Hz), 5.71 (1H, d,
J=3.3 Hz), 6.75 (1H, br. s), 7.14 (3H, s), 7.15 (3H, s),
8.10 (1H, br. s), 11.69 & 12.50 (total 1H, br. s each).
50-Deoxy-5-¯uoro-N4-(butoxycarbonyl)cytidine (17). Mp
119±120 ꢀC, HRMS [FAB, MH+] calcd for C14H21
FN3O6 m/z 346.1409, found 346.1418; 1H NMR
(DMSO-d6) d 0.91 (3H, t, J=7.3 Hz), 1.31 (3H, d, J=
5.9 Hz), 1.36 (2H, m), 1.59 (2H, t. t, J=7.3Hz), 3.68 (1H,
q, J=5.9 Hz), 3.89 (1H, m), 4.02±4.12 (3H, m), 5.06 (1H,
d, J=5.9 Hz), 5.41 (1H, br. d), 5.67 (1H, d, J=3.7Hz),
8.04 (1H, br. s), 10.53 and 11.67 (total 1H, br. s each).
50-Deoxy-N4-(3,5-dimethybenzoyl)cytidine (23). Mp 168±
169 ꢀC, MS [FAB, MH+] m/z 378; HRMS [FAB,
MH+] calcd for C18H21FN3O5 m/z 378.1460, found
378.1463; 1H NMR (DMSO-d6) d 1.32 (3H, d, J=
6.3 Hz), 2.34 (6H, s), 3.71 (1H, q, J=5.6 Hz), 3.91 (1H,
m), 4.12 (1H, m), 5.06 (1H, d, J=5.9 Hz), 5.42 (1H, br.
d,), 5.71 (1H, d, J=3.6 Hz), 7.27 (1H, s), 7.62 (2H, s),
8.09 (1H, br. s), 11.20 (1H, br. s).
50-Deoxy-5-¯uoro-N4-(hexyloxycarbonyl)cytidine (18).
Mp 114±116 ꢀC, HRMS [FAB, MH+] calcd for C16H25
FN3O6 m/z 374.1721, found 374.1721; 1H NMR
(DMSO-d6) d 0.87 (3H, t, J=6.3 Hz), 1.22±1.41 (6H,
m), 1.30 (3H, d, J=5.9 Hz), 1.60 (2H, t. t, J=6.3 Hz),
3.67 (1H, q, J=5.9 Hz), 3.89 (1H, d. q, J=5.9 Hz),
4.06±4.11 (3H, m), 5.02 (1H, d, J=5.9 Hz), 5.37 (1H, d,
J=5.3 Hz), 5.66 (1H, d, J=3.5 Hz), 8.00 (1H, br. s),
10.51 (1H, br. s).
50-Deoxy-N4-(3,5-dichlorobenzoyl)cytidine (24). Mp 156±
157 ꢀC, MS [FAB, MH+] m/z 418; HRMS [FAB,
MH+] calcd for C16H1535Cl 37ClFN3O5 m/z 420.0340,
35
found 420.0443; calcd for m/z C16H15 Cl2FN3O5
1
418.0370, found 418.0428; H NMR (DMSO-d6) d 1.32
(3H, d, J=6.3 Hz), 3.71 (1H, q, J=5.6 Hz), 3.91 (1H, d.
q, J=5.9 Hz), 4.12 (1H, m), 5.07 (1H, d, J=5.6 Hz),
5.40 (1H, d, J=4.9 Hz), 5.70 (1H, d, J=3.6 Hz), 7.90
(1H, s), 7.96 (2H, s), 8.09 (1H, br. s), 12.51 (1H, br. s).
50-Deoxy-5-¯uoro-N4-(heptyloxycarbonyl)cytidine (19).
Mp 115.5±117.5 ꢀC; HRMS [FAB, MH+] calcd for
Enzyme assay
1
C17H27FN3O6 m/z 388.1877, found 388.1883; H NMR
(DMSO-d6) d 0.86 (3H, t, J=6.1 Hz), 1.21±1.40 (11H,
m), 1.60 (2H, t. t, J=6.6 Hz), 3.67 (1H, q, J=5.9 Hz),
3.89 (1H, d. q, J=5.9 Hz), 4.04±4.10 (3H, m), 5.02 (1H,
d, J=5.9 Hz), 5.38 (1H, d, J=4.9 Hz), 5.66 (1H, d,
J=4.0 Hz), 7.99 (1H, br.s), 10.52 (1H, br. s).
The enzyme assay method as well as the method for the
preparation of crude enzyme extracts of the liver and
intestinal tissues from mice, monkeys, and humans are
described in ref 5.
Pharmacokinetic studies
50-Deoxy-5-¯uoro-N4-(octyloxycarbonyl)cytidine (20).
HRMS [FAB, MH+] calcd for C18H29FN3O6 m/z
402.2033, found 402.2057; H NMR (DMSO-d6) d 0.86
The pharmacokinetic studies of N4-(alkoxycarbonyl)-50-
DFCR derivatives were carried out in cynomolgus
monkeys (3±4kg; ®ve monkeys in each group). The blood
samples were taken at various time points after single
oral administration of the test compounds (20 mg/kg).
The intact drug levels in plasma were measured by
the reverse-phase HPLC-UV method (column: YMC-
Pack AM-212 C-8 (150Â6 mm I.D., 5 mM, YMC K.K.,
Tokyo Japan); mobil phase; acetonitrile:citrate bu€er
(30:70±40:60, v/v); Flow rate: 1.0 mL/min; detection:
310 nM). The 50-DFCR (8) and 50-DFUR (4) levels in
the plasma were measured by the reverse-phase HPLC-
UV method (column: Inertsil ODS-2 (250Â4.6 mm I.D.,
5 mM, GL Science, Tokyo, Japan); mobil phase; aceto-
nitrile:methanol:citrate bu€er (6:20:174, v/v/v); ¯ow
rate: 0.8 mL/min; detection: 267 nM; retention time:
9.3 min for 50-DFCR, 12.8 min for 50-DFUR, 16.5 min
for tegafur (internal standard)) with quanti®cation limits
of 0.02±0.1 mg/mL. The urinary concentration (0±24 h) of
the intact drug, 5-FU, and a range of further metabolites
(5,6-dihydro-5-¯uorouracil, a-¯uoro-b-ureido propionic
acid, a-¯uoro-b-alanine (FBAL), N-carboxyl-FBAL)
were determined simultaneously by 19F NMR with
quanti®cation limits of 6.5±20mg/mL for 10mL of urine.
1
(3H, t, J=6.4 Hz), 1.26±1.32 (10H, m), 1.30 (3H, d,
J=6.3 Hz), 1.60 (2H, t. t, J=6.3 Hz), 3.67 (1H, q, J=
5.6 Hz), 3.89 (1H, d. q, J=5.9 Hz), 4.04±4.10 (3H, m),
5.02 (1H, d, J=6.3 Hz), 5.37 (1H, d, J=5.3 Hz ), 5.66
(1H, d, J=2.6 Hz ), 8.00 (1H, br. s), 10.50 (1H, br. s).
50-Deoxy-5-¯uoro-N4-(nonyloxycarbonyl)cytidine (21).
Mp 120±122 ꢀC, HRMS [FAB, MH+] calcd for C19H31
FN3O6 m/z 416.2189, found 416.2203; 1H NMR
(DMSO-d6) d 0.85 (3H, t, J=6.3 Hz), 1.25±1.32 (12H,
m), 1.30 (3H, d, J=6.3Hz), 1.60 (2H, t. t, J=6.3 Hz), 3.67
(1H, q, J=5.6 Hz), 3.89 (1H, d. q, J=5.9 Hz), 4.04±4.10
(3H, m), 5.02 (1H, d, J=5.9 Hz), 5.37 (1H, d, J=5.3Hz),
5.66 (1H, d, J=4.0Hz), 8.00 (1H, br. s), 10.53 (1H, br. s).
50-Deoxy-5-¯uoro-N4-(3,4,5-trimethoxybenzoyl)cytidine,
Galocitabine (5). Mp 170±171 ꢀC, HRMS [FAB, MH+]
calcd for C19H23FN3O8 m/z 440.1463, found 440.1475;
1H NMR (DMSO-d6) d 1.33 (3H, d, J=6.3 Hz), 3.74
(1H, q, J=5.6 Hz), 3.75 (3H, s), 3.85 (6H, s), 3.91 (1H.
m), 4.12 (1H m), 5.06 (1H, d, J=5.9 Hz), 5.41 (1H, d,
J=4.9 Hz), 5.72 (1H, d, J=3.6 Hz), 7.35 (2H, s), 8.12
(1H, br. s), 11.75 and 12.42 (total 1H, br. s each).
50-Deoxy-N4-(3,5-dimethoxybenzoyl)cytidine (22). Mp
159±161 ꢀC, HRMS [FAB, MH+] calcd for C18H21
FN3O7 m/z 410.1358, found 410.1372; 1H NMR
(DMSO-d6) d 1.32 (3H, d, J=6.3 Hz), 3.71 (1H, q, J=
5.6 Hz), 3.80 (6H, s), 3.91 (1H, m), 4.12 (1H, m), 5.06
Acknowledgement
We would like to thank Mr. Noboru Nakayama and
1
Ms. Yoshiko Itezono for measuring MS, H NMR and
19F NMR spectra.

1706
N. Shimma et al. / Bioorg. Med. Chem. 8 (2000) 1697±1706
12. Chabner, B. A.; Johns, D. G.; Coleman, C. N.; Drake, J.
References
C.; Evans, W. H. J. Clin. Invest. 1974, 53, 922.
13. Ninomiya, Y.; Miwa, M.; Eda, H.; Sahara, H.; Fujimoto,
K.; Ishida, M.; Umeda, I.; Yokose, K.; Ishitsuka, H. Jpn. J.
Cancer Res. 1990, 81, 188.
14. D'Sousa, R.; Kiss, J. EP 21231, 7 January 1981.
15. (a) Niedballa U.; Vorbruggen H. J. Org. Chem. 1974, 39,
3654. (b) Saneyoshi M.; Inomata M.; Fukuoka F. Chem.
Pharm. Bull. 1978, 26, 2990.
16. Matsuda, A.; Kurasawa, Y.; Watanabe, K. A. Synthesis
1981, 748.
17. Ishikawa, T.; Utoh, M.; Sawada, N.; Nishida, M.; Fukase,
Y.; Sekiguchi, F.; Ishitsuka, H. Biochem. Pharmacol. 1998, 55,
1091.
18. Kawashima, A.; Horii, I. In Proceedings of the 53rd Annual
Meeting of the Japanese Cancer Association, October, 1994,
Nagoya. Abstract No. 2411.
19. Ishikawa, T.; Fukase, Y.; Yamamoto, T.; Sekiguchi, F.;
Ishitsuka, H. Biol. Pharm. Bull. 1998, 21, 713.
20. Mackean, M.; Planting, A.; Twelves, C.; Schellens, J.; All-
man, D.; Osterwalder, B.; Reigner, B.; Grin, T.; Kaye, S.;
Verweij, J. J. Clin. Oncol. 1998, 16, 2977.
1. (a) Sinhababu, A. K.; Thakker, D. R. Advanced Drug
Delivery 1996, 19, 241. Jungheim, L. N.; Shepherd, T. A.
Chem. Rev. 1994, 94, 1553. (c) Deonarain, M. P.; Spooner, R.
A.; Epenetos, A. A. Gene Ther. 1995, 2, 235.
2. Chabner, B. A.; Longo, D. L. In Cancer Chemotherapy and
Biology: Principles and Practice, 2nd ed; Lippincott-Raven:
Philadelphia, New York, 1996; pp 149±211.
3. De Gramont, A.; Louvet, C.; Andre, T.; Tournigand, C.;
Krulik, M. Eur. J. Cancer 1998, 34, 619.
4. Ishitsuka, H.; Miwa, M.; Takemoto, K.; Fukuoka, K.;
Itoga, A.; Maruyama, H. Gann (Jpn. J. Cancer Res.) 1980, 71,
112.
5. Miwa, M.; Ura, M.; Nishida, M.; Sawada, N.; Ishikawa, T.;
Mori, K.; Shimma, N.; Umeda, I.; Ishitsuka, H. Eur. J. Can-
cer 1998, 34, 1274.
6. Kono, A.; Hara, Y.; Sugata, S.; Karube, Y.; Matsushima,
Y.; Ishitsuka, H. Chem. Pharm. Bull. 1983, 31, 175.
7. Taguchi, T.; Terasawa, T.; Abe, O.; Yoshida, Y.; Tomi-
naga, T.; Ogawa, N. Jpn. J. Cancer Chemother. 1985, 12, 2052.
8. Kreis, W.; Watanabe, K. A.; Fox, J. J. Helv. Chim. Acta.
1978, 61, 1011.
21. Ishikawa, T.; Sekiguchi, F.; Fukase, Y.; Sawada, N.; Ish-
itsuka, H. Cancer Res. 1998, 58, 685.
9. Chabner, B. A.; Longo, D. L. In Cancer Chemotherapy and
Biology: Principles and Practice, 2nd ed. Lippincott-Raven:
Philadelphia, New York, 1996; pp 213±233.
22. Eda, H.; Fujimoto, K.; Watanabe, S.; Ura, M.; Hino, A.;
Tanaka, Y.; Wada, K.; Ishitsuka, H. Cancer Chemother.
Pharmacol. 1993, 32, 333.
10. Camiener, G. W.; Smith, C. G. Biochem. Pharmcol. 1965,
14, 1405.
11. Giusti, G.; Mangoni, C.; De Petrocellis, B.; Scarano, E.
Enzym. Biol. Clin. 1970, 11, 375.
23. Sawada, N.; Ishikawa, T.; Fukase, Y.; Nishida, M.;
Yoshikubo, T.; Ishitsuka, H. Clin. Cancer Res. 1998, 4, 1013.
24. Sawada, N.; Ishikawa, T.; Sekiguchi, F.; Tanaka, Y.;
Ishitsuka, H. Clin. Cancer Res. 1999, 5, 2948.