Arabinosyl-5-azacytosine: A novel nucleoside entering clinical trials

Investigational New Drugs 5:315-328 (1987) C Martinus Nijhoff Publishers, Boston - Printed in the Netherlands

Arabinosyl-5-azacytosine: A novel nucleoside entering clinical trials Jean L. Greta 1, D. Dale Shoemaker 1, Daniel F. Hoth 1, Susan A. King 1, Jacqueline Plowman ~, Daniel Zaharko 2, Charles K. Grieshaber 2, Steadman D. Harrison Jr. 3, James C. Cradock 2 and Brian Leyland-Jones The Cancer Therapy Evaluation Program ~ and Developmental Therapeutics Program s, Division of Cancer Treatment, National Cancer Institute, Bethesda, MD 20892 USA; the Chemotherapy Division 3, Southern Research Institute, 2000 Ninth Avenue S. Birmingham, Alabama 35255, USA

Key words: arabinosyl-5-azacytosine, deoxycytidine kinase, cytosine arabinoside, 5-azacytidine

Abstract Arabinosyl-5-azacytosine is a new compound which has been selected by the Division of Cancer Treatment, National Cancer Institute for clinical development as an antineoplastic agent based on its high degree of activity against a broad range of tumor types in preclinical studies. Therapeutic activity has been observed against murine and human leukemias, transplantable murine solid tumors, and human tumor xenografts. Arabinosyl-5-azacytosine exhibited a broader spectrum of activity against human solid tumors than cytosine arabinoside. Arabinosyl-5-azacytosine is phosphorylated to the nucleotide level by deoxycytidine kinase. Upon further anabolism to the triphosphate level, it can be incorporated into DNA. The mechanism of cytotoxicity is thought to be related to inhibition of DNA synthesis. Leukemic and solid tumor cell lines that are resistant to cytosine arabinoside due to deletion of deoxycytidine kinase activity are cross-resistant to arabinosyl-5-azacytosine. Unlike cytosine arabinoside, arabinosyl-5-azacytosine does not readily undergo deamination. Schedule dependence has been demonstrated in mice bearing LI210 leukemia, with superior activity seen with multiple doses administered on each treatment day compared to administration of larger but less frequently administered doses. From preliminary data in solid tumor models, however, antitumor activity did not appear to be superior with continuous infusion compared to that observed on a bolus schedule. Preclinial toxicology studies indicated that the bone marrow and gastrointestinal tract were the main target organs. A single large dose of arabinosyl-5-azacytosine could be tolerated by both mice and dogs. When administered as a continuous infusion, the toxicity was related to both the dose and duration of exposure, suggesting that toxicity resulted from a critical time above a threshold concentration as opposed to the total area under the concentration-time curve. Phase I clinical trials have been initiated to determine the maximum tolerated dose on a low dose continuous infusion schedule for 72 hours and also on a high dose short infusion daily times five schedule.

Introduction 1 - beta- D- Arabinofuranosyl- 5- azacytosine (AraAC, Fazarabine; NSC-281272) is a novel compound synthesized by Beisler et al. at the National Cancer Institute (NCI) to combine the structural

features of two nucleosides currently used in the clinical management of acute leukemia [1, 2]. In preclinical screening, Ara-AC has shown a broad spectrum of antitumor activity against murine and human leukemias and solid tumors. The chemical and pharmaceutical data, preclinical antineoplastic

316 activity, mechanism of action, murine and canine toxicology and pharmacology of this interesting nucleoside will be reviewed, and the plans for its initial clinical evaluation will be described.

NH=

Structure

HO

NH2

OH

OH

Cyd

Ara-AC has the stereochemical inversion of the hydroxyl group at the 2' position of cytidine (as does cytosine arabinoside) plus the substitution of a nitrogen in place of the carbon 5 in the pyrimidine base (as does 5-azacytidine) (Fig. 1). The molecular formula of Ara-AC is C8H12N405, with a molecular weight of 244.2 gm/mole. Because of the lability of the triazine ring, Ara-AC is unstable in aqueous medium and plasma [3]. This instability can be explained by the nucleophilic addition of water to the 5 - 6 double bond of the azacytosine moiety accompanied by a proton transfer, followed by hydrolysis to a ring opened N-formyl product in the same manner as described for 5-azacytidine [4]. The stability of Ara-AC can be improved through the use of an organic solvent such as dimethylsulfoxide (DMSO) which cannot participate in the hydrolysis of the triazine ring.

Pharmaceutical data

The investigational agent is supplied as a sterile white lyophilized powder in vials containing 250 mg of drug. The intact vials should be stored under refrigeration at 2 - 8 ~ Two methods of reconstitution of the freeze dried dosage form for intravenous (iv) administration are recommended depending on the duration of the infusion. For short term infusions of 3 hours or less, the drug is dissolved in 25 ml of Sterile Water for Injection, U.S.P. or Lactated Ringer's Injection, U.S.P. to yield a l0 mg/ml solution of Ara-AC. The drug should be further diluted immediately with Lactated Ringer's Injection, U.S.P. to a final Ara-AC concentration of 0.01 - 1.0 mg/ml. These solutions exhibit approximately 10070 decomposition in 3 hours at room temperature. For long term continuous infusion, the 250 mg

NH2

HO

NH=

HO

dCyd

OH

5-Aza-Cyd

NH=

HO

Ara-C

Ara-AC

Fig. 1. Structure of arabinosyl-5-azacytosine(Ara-AC), cytosine

arabinoside (Ara-C), 5-azacytidine(5-Aza-Cyd),and the natural nucleosides cytidine (Cyd) and deoxycytidine (dCyd).

vial should be reconstituted with 3.5 ml of sterile 70~ (v/v) DMSO to yield a 70 mg/ml solution. Further dilution with additional sterile 7007o DMSO should be performed to achieve the appropriate Ara-AC concentration. This solution should be administered slowly using a syringe pump delivery system into a side injection port of a running intravenous (iv) solution of 5070 Dextrose Injection, USP. A total volume of 12 ml with an infusion rate of 0.5 m l / h o u r is recommended to limit the amount of DMSO infused to 350 mg/hour. The accuracy of the syringe pump delivery system should be verified prior to drug administration. At Ara-AC concentrations ranging from 2 - 7 0 mg/ml, the drug exhibits approximately 1007o decomposition in 24 hours. The use of sterile microbore tubing having an inner lining o f polyolefin is recommended for connecting the syringe pump to the side port of the running 507o Dextrose in water iv infusion (e.g. Polyfin extension set, Mini Med Technologies, Sylmar, CA). The use of polyvinyl chloride iv tubing should be avoided because of the risk of extraction of plasticizer by the DMSO solven [3].

317 Table 1. Spectrum of preclinical antitumor activity of Ara-AC in NCI tumor models. The activity rating was based on an increase in life span in ip and ic implanted leukemias and solid tumors: + + = _> 50% ILS (T/C >_ 150%), except ILS _> 75% for P388 (T/C _> 175%); + = 25-49% ILS; - = < 25% ILS; and the cures represent the surviving tumored animals compared to the total number of animals on day 30. The activity rating in sc implanted tumors was based on tumor measurements: + + = >_ 90% inhibition of tumor growth (T/C _< 10%); - = < 80% and < 58% inhibition of growth for the CD8F~ mammary and colon 38 tumors, respectively. For the src models, values are based on changes in mean tumor weights; positive numbers = 100 x change in test tumor weight/change in control tumor weight; negative numbers (regression) = 100 x change in test tumor weight/initial test tumor weight. For the MX-1 mammary xenograft, the " c u r e s " represent the number of mice in which the tumors completely regressed. The T / C values represent the second best therapeutic effect in treated mice compared to control tumored mice achieved from a series of dose-response studies. All experiments were done more than once except for the ic P388 study. Tumor system

Treatment schedule

Activity

% T/C

Cures

310 > 341 > 297 > 303

0 3/6 6/10 5/6

rating Murine leukemias ip ic ip ic

L1210 leukemia L1210 leukemia P388 leukemia P388 leukemia

ip ip ip ip

q q q q

ld, ld, ld, ld,

days days days days

1-9 1-9 1-9 1-9

+ + + +

+ + + +

ip ip ip ip ip

q q q q q

ld, ld, 7d, ld, ld,

days 1-9 staging day days 2, 9 days 1-9 days 1-9

+ + + + +

141 --170 155

0 --0 0

Human tumor xenografts src CX-1 colon tumor src LX-1 lung tumor src MX-1 mammary tumor ic U-251 glioma

sc sc sc ip

q q q q

4d, 4d, 4d, ld,

days days days days

+ + + + + + -

7 -72 -81 --

0 0 3/6 --

ic TE-671 meduUoblastoma

ip q ld, days 1-9

+ +

200

0

Murine solid tumors ip sc sc iv ip

B16 melanoma CD8F~ mammary carcinoma colon adenocarcinoma 38 Lewis lung carcinoma M5076 sarcoma

1, 5, 9, 13 1, 5, 9 1, 5, 9 1-9

Preclinical antitumor activity Ara-AC given intraperitoneally (ip) was highly active against ip and intracranially (ic) implanted P388 and L1210 murine leukemias (Table 1) [5-7]. The optimal ip dose on a d x 9 schedule was 150 mg/m2/d for ip P388 and 600 m g / m V d for ip L1210 leukemia. Ara-AC was substantially more effective than Ara-AC against ic implanted P388 when given ip daily for 9 days [6]. Ara-AC at a dose of 600 m g / m V d x 9 had good activity against the iv implanted Lewis lung carcinoma and was moderately active against ip implanted B16 melanoma. Inactivity was demonstrated against subcutaneous (sc) implanted CD8FI mammary carcinoma on a single dose treatment schedule and colon 38 murine adenocarcinoma on a weekly schedule [5].

Ara-AC demonstrated significant activity against three human xenografts of the NCI tumor panel implanted under the renal capsule (src) of athymic mice [5]. In this assay, the drug is considered active if the percentage of average tumor weight change in the treated versus control animal (070 T / C ) is less than 20%, or if regression of the tumor in the treated animal occurs. Following sc administration every 4 days for 3 - 4 doses, Ara-AC inhibited the tumor growth o f the CX-I colon tumor by > 90~ (T/C < 10%), and caused regression of the LX-1 lung and MX-1 mammary xenografts. In contrast, cytosine arabinoside (Ara-C) and 5-azacytidine (5-Aza-Cyd) were inactive against the CX-1 and LX-1 human tumor xenografts. Ara-AC also demonstrated activity on a d x 9 schedule against an ic implanted medulloblastoma xenograft, but

318 Table 2. Effect of route and schedule of administration on the activity of A r a - A C in vivo against ip L 1210. CD2F~ mice were implanted ip with 105 L1210 cells on day 0 and treatment was started 24 hours later. An 84 m g / m l stock solution of Ara-AC in 70% DMSO was prepared, a n d treatment was administered iv, po or ip. Doses > 225 m g / k g were not completely in solution. The median survival time for control mice was 8 days. For the studies evaluating the effect of schedule, a 66 m g / m l stock solution of Ara-AC in 70% DMSO was prepared and diluted with saline for administration on the d 1 - 5 , d 1 - 9 , d I, 5, 9, and q 3h x 8 d 1, 5, 9 schedules. A 75 m g / m l solution was used for the d 1 only and q 3 h • 8 d 1 only schedules. Final concentrations of DMSO were _< 16%. Preparations were made fresh for every 8 hours for the q 3h x 8 schedules and just prior to injection for the once daily schedules. The ILS is based on the median survival times of dying mice and those surviving until day 60. The median survival of control mice was 8.2 days. Mice that were tumor free on day 60 were considered to be cured.

Route & schedule

Optimal dose (mg/m'-/day)

Total dose (mg/m")

ip d • 9 pod • 9 iv d • 5

150 300 540

1350 2700 2700

210 210 176

0/10 0/10 0/10

ip ip ip ip ip ip

1350 1800 540 150 338 28

1350 5400 2700 1350 2704 672

46 152 142 189 98 > 631

0/10 0/10 0/10 0/10 1/10 9/10

d d d d q q

• 1 1, 5, 9 1-5 1-9 3h • 8 d 1 3 h x 8 d 1, 5, 9

was inactive against an ic glioma model [7]. The activity o f Ara-AC against a panel of human tumor xenografts was documented in studies conducted by the American Cyanamid Co [8]. Groups of 5 to 6 Balb/nude athymic mice received five 2 mm 3 tumor fragments on day 0. Drug treatment given ip once daily for 9 days was initiated after tumors had reached an average weight of 100-350 mg. In this assay, a drug was considered to be negative if the % T / C was > 42%. In such an assay system, Ara-AC was highly active against 3/3 leukemia lines, 1/3 lung tumors (LX-1), 3/4 breast cancers (including MX-1), 3/4 colon cancers (including CX-1) and 1/1 ovarian carcinoma line, but was inactive against 2/2 melanoma lines. In comparison, Ara-C at the optimal dose was also active against all 3 leukemia lines, but was reproducibly active in only 1/4 breast cancers, 1/4 colon, and 1/1 ovarian lines, and was inactive against all 3 lung carcinomas and the 2 melanoma lines. Among the 18 xenograft models, Ara-AC induced actual tumor regression in six (3 leukemia, 1 lung, 1 breast, and 1 ovarian line), while Ara-C produced definite regression in only three models (2 leukemia and 1 ovarian line). Thus, Ara-AC appeared to have a broader spectrum of antitumor activity against

% ILS

% Cured

solid tumors than Ara-C [5, 8]. The effect of route of administration has been examined using the ip implanted L1210 leukemia model (Table 2) [7]. Twenty-four hours after inoculation of the tumor burden, Ara-AC was administered either ip or orally (po) daily for 9 days or iv daily for 5 days. The median survival of the treated tumored mice was compared to the nontreated controls and expressed as a percentage increase in life span (ILS). Ara-AC was active whether administered by the ip, iv or oral route. Both the ip and po d • 9 treatments demonstrated equivalent activity at their optimal doses. The effect of schedule of administration has also been examined using the ip implanted L1210 leukemia model. Ara-AC produced equivalent increases in life span when given ip as a single dose of 600 m g / m 2 or as a total dose of 600 mg/m-' divided into 5 daily doses (120 mg/m2/d x 5), with an ILS of 114% and 99%, respectively [9]. An equal dose of Ara-AC (given ip once) was not therapeutic (ILS 12%) [9]. Further analysis of several schedules indicated that the activity of a solution of Ara-AC in DMSO was clearly dependent on the schedule employed (Table 2) [7]. Ara-AC administered ip every 3 hours for 8 doses on days 1,5 and 9 yielded an ILS

319 Table 3. Effect of dose and schedule of administration on the antitumor activity in vivo against the MX-1 m a m m a r y carcinoma. N C R - n u

female micereceiveda sc implant of MX-1 mammary carcinoma on day 0. Treatment began on stagingday 8 when median tumor weights for each group ranged from 88-126 mg. Alzet 2001 pumps were used for the sc infusion doses. All doses were soluble except for 1350 mg/m2 given on the q 3 hr • 2 dose schedule. Schedule

Dose ( m g / m 2)

Dose intensity (mg/m2/d)

Total dose ( m g / m 2)

Tumor growth delay: T - C (days) a

# with complete tumor regression b

Tumor free survivors c

Drug-deaths in tumored mice

control 72 hr sc infusion d 1, 2, 3 (70OToDMSO)

0 489 366 273

0 489 366 273

0 1467 1098 819

-15.0 12.6 15.5

-1/5 1/6 3/6

0/11 0/5 1/6 0/6

0/11 4/5 2/6 2/6

bolus sc q 3 hr • 2 doses d 1, 2, 3 (10~ DMSO)

489 366 273

978 732 546

2934 2196 1638

16.3 12.8 13.1

1/6 2/6 0/4

1/6 0/6 0/4

0/6 0/6 0/4

bolus sc q 3 hr x 2 doses d 1, 5, 9 (20OToDMSO)

1350 1014 756 0

2700 2028 1512 0

8100 6084 4536 0

14.8 15.9 13.9 1.4

0/6 0/6 0/6 0/6

0/6 0/6 0/6 0/6

0/6 0/6 0/6 0/6

8.6 6.1 7.9

l/6 0/6 0/6

0/6 1/6 0/6

0/6 0/6 0/6

bolus sc • 1 dose days 1 - 7 (3.3~ DMSO)

153.9 122.1 91.8

153.9 122.1 91.8

1077.3 854.7 642.6

a T - C : excludes tumor free survivors and any other animal whose t u m o r failed to attain the evaluation size. Times for control tumors to reach 1000 and 2000 mg were 15.2 and 18.2 days, respectively. The value given is the unweighted average of the differences of the median times postimplant for the treated and control groups to attain 1000 and 2000 mg; bTumor became unpalpable ( < 63 rag), but subsequently recurred; CTumor was not palpable on day 39.

> 600070 with 9/10 mice surviving t u m o r free at 60 days. In comparison, a single injection of the drug on the same treatment days produced an ILS of 15207o at the highest soluble dose with no long term survivors. Zaharko et al. compared the antitumor efficacy of Ara-AC in mice bearing iv and ip L1210 leukemia on a sc continuous infusion (CI) schedule and an intermittent ip dose schedule, respectively [10]. A dose rate of 20.4 m g / m 2 / h r for 24 and 48 hours (total dose 490 m g / m 2 and 980 mg/m% respectively was estimated to produce a 4-log and 7-log cell kill with acceptable toxicity to the host. The ILS in these mice was 21707o and 283070, respectively. In contrast, a dose of 2400 m g / m 2 given ip every 3 hours for 4 doses (total dose 9600 m g / m 2) produced only a 3-log cell kill and an ILS o f 167~ These data indicated that longer infusion times resulted in longer median life spans up to limiting lethal toxici-

ty occurring _> 96 hours of infusion. The effect of Ara-AC on viability in vitro appears to be a function of both dose and duration of exposure. A 24 hour exposure to 1/zg/ml (4.1 #M) or 10/zg/ml (41/zM) reduced the cloning efficiency of L1210 cells to 7~ and 0.2~ of control, respectively [10]. Increasing the duration of exposure to 48 or 72 hours enhanced the lethality of a given dose of Ara-AC; for example, 10 /~g/ml resulted in a colony survival of 0.09070 and 0.0150/0 of control, respectively [10]. Thus, a 72 hour exposure to 10 # g / m l produced approximately a 4-log reduction in viability compared to control. No increase in lethality was observed by increasing the dose to 100 #g/ml. A 24 hour exposure to 2.4/zg/ml (10/zM) reduced the cloning efficiency of a h u m a n colon carcinoma cell line (HT 29) to 307o of control, approximately a 2-log reduction in viability [11]. Exposure of Molt-4 cells (a T lymphoblast line) to 0.24

320 Table 4. Activity of Ara-AC against drug-resistant P388 and L1210 leukemia. Cohorts of ten CD2F~ mice were implanted ip with either the parental leukemic line which retained sensitivity to the anticancer agent tested or with a subline that had acquired resistance to the agent. Treatment was begun the following day. Resistant

Optimal q d • 9

Sensitive leukemia

tum or tested

treatment (mg /m2 /day)

Cures/

% ILS

total

Resistant leukemia Log~o c h ~ g e

Cures/

in tum or burden after last Rx a

total

% ILS

Loglo change

C omme nt s

in t u m o r burden after last Rx a

P388/Dox

450 b 450 b

0/10 0/10

143 202

-4.5 - 7

0/10 0/ 10

177 196

-6 - 7

no crossresistance

P388/VCR

450 b 450 b

0/10 0/10

212 206

-7 -7

0/10 0/ 10

134 161

-7 -7

no crossresistance

P388/L-PAM

450 b 300

0/10 1/10

202 211

-7 - 7

1/10 1/10

107 158

-5 - 7

no crossresistance

P388/CDDP

300

0/10

171

- 7

0/ 10

84

- 3

possible cross-

resistance P388/Ara-C

300

1/10

190

- 7

0/ 10

- 20

+2

cross-resistance

P388/MTX

450 b 300

0/10 0/10

175 216

-7 -7

0/10 0/10

215 209

-6 -6

no crossresistance

L1210/5-FU

300

0/10

237

-6

8/10

83

-7

no cross-

201

0/10

226

-6

9/10

231

-7

resistance

athe data represent the estimated Log10 change in the viable tumor stem cell population at the end of therapy as compared to that at the start of therapy, based on the median day of death among the animals that died. In the instances where _ 50% of mice survived, the estimate of the number of viable cells remaining after treatment was based on the % survivors, assuming that the number of cells surviving treatment in each animal obeys a Poisson distribution and that those animals still alive at 60 days had no t u m o r cells that survived treatment, bThe dose was the highest tested.

~ g / m l (1 ~M) Ara-AC for 24 hours decreased the cloning efficiency to 2% of control [12]. The effect of schedule on the cytotoxicity of AraAC in vivo against a h u m a n m a m m a r y carcinoma has also been evaluated (Table 3). Activity, as evidenced by the delay in time for treated tumors to reach a certain size when compared to tumor growth in control animals, was observed with all four schedules. If evaluation of activity also included complete tumor regressions and tumor-free survivors, then the two best treatments appeared to be 273 m g / m 2 / d administered as a 72 hour CI and 489 m g / m ~ given as two bolus injections 3 hours apart on three sequential days. Whether either of the latter two schedules is more therapeutic in hum a n solid t u m o r xenograft models awaits further evaluation, since a maximally tolerated dose was apparently not attained with the bolus schedule (no drug-related deaths occurred in either tumored or

non-tumored mice, Tables 3 and 6, respectively). The cross-resistance of Ara-AC with other antineoplastic agents was evaluated using sublines of P388 and LI210 with acquired resistance to various anticancer drugs (Table 4) [7]. An Ara-C resistant subline of P388 (P388/Ara-C) was cross-resistant to Ara-AC. P388 sublines with resistance to doxorubicin, vincristine, melphalan and methotrexate were sensitive to Ara-AC. An L1210 subline resistant to 5-fluorouracil was sensitive to Ara-AC. Other investigators have also observed that Ara-C resistant murine leukemia tumors are cross-resistant to A r a - A C [5, 13]. Ara-AC induced differentiation in the h u m a n promyelocytic cell line, HL-60. Following a 96 hour exposure to 10 ~M Ara-AC, differentiation was induced in 44% o f the cells as assessed by the ability of cells to reduce nitro blue tetrazolium (NBT) to the formazan derivative [5]. At this A r a - A C con-

321 uridine/cytid~ne kinase 5-Aza-Cyd

diphosphate kinase

GI'IP kinase 9 5-Aza-CRP

phospha~ase

x 5-Aza-CE~ osphacase

RNA polymerase 9 5-Aza-CTP

phosphatase

lribonucleotide reductase

[

deoxTcytidine kinase dOJd Ara-C Ara-AC~ phosphatase

dGyd/Cyd deaminase

~'-d--U

dCMP kinase

d~P 9 Ara-C~ I Ara-ACMP ~ / phosphatase

diphosphace ~" kinase dCTP alum % ~-CTP 9 Ara-C~ ~-ACTP Ara-ACIP " triphosphatase

DNA pol~erase

)EINA

d67iP deaminase

t~-a-II~

Fig. 2. Metabolic pathways of Ara-AC, Ara-C, and 5-Aza-Cyd.

centration, cell growth was markedly inhibited ( < 30o7oof control), and only 50o/o of the cells were viable as determined by trypan blue dye exclusion. In this system, the congener 5-aza-2 ' -deoxycytidine was 10-fold more potent on a molar basis than AraAC as a differenting agent [5]. For comparison, incubation of HL-60 cells for six days with 32 /zM 5-Aza-Cyd, a concentration which also inhibited cell growth to 31% of control, induced 36~ NBT + cells [14].

Mechanism of action

The ability of natural nucleosides to protect a n d / o r rescue cells from the cytotoxicity of antimetabolites has provided insights into their mechanisms of action. Deoxycytidine (dCyd) can protect cells from the cytotoxicity of Ara-C and Ara-AC, but is ineffective in combination with 5-Aza-Cyd [9, 15, 16]. In contrast, uridine (Urd) and cytidine (Cyd) can partially rescue cells from the toxicity of 5-AzaCyd, but these pyrimidines are not successful in protecting cells against Ara-C or Ara-AC [15]. These observations suggest that Ara-AC is metabolized by the enzymes that utilize dCyd as the natural substrate. Ara-AC enters cells via a nucleoside transport mechanism and subsequently undergoes metabolic conversion to the nucleotide level by dCyd kinase, an enzyme of the pyrimidine salvage pathway

(Fig. 2). The kinetics of phosphorylation of AraAC have been determined using a partially purified human cytoplasmic dCyd kinase [12]. The Km for the natural substrate was 3/~M, and the apparent Km (55/~M) and relative Vmax (310~ for Ara-AC were similar to that observed for Ara-C. Using a partially purified dCyd kinase from L l 2 l0 cytosols, the Km for Ara-AC was reported to be 70 #M compared with 27 #M for dCyd [9]. dCyd functions as a potent inhibitor of the phosphorylation of AraAC (Ki = 3.6/~M) [9]. The comparative anabolism of Ara-AC, Ara-C and 5-Aza-Cyd is shown in Fig. 2. Ara-C and AraAC share a common activation pathway via dCyd kinase. Ara-CMP and Ara-ACMP are anabolized to the triphosphate level by the sequential action of dCyd monophosphate kinase and nucleoside diphosphate kinase. Townsend reported that following a 3 hour exposure to 1 /zM [3H]-Ara-AC, 70~ of the intracellular metabolites in Molt-4 cells could be accounted for by Ara-ACTP, with the remainder primarily Ara-ACMP [12]. Degradative enzymes (phosphatases) exist which can cleave the phosphate group(s) from the arabinofuranosyl moiety. Another deactivation pathway for Ara-C is deamination [16]. Ara-C can be converted to the inactive metabolite, uridine arabinoside (Ara-U), by Cyd deaminase, dCyd rnonophosphate deaminase can inactivate Ara-CMP by conversion to AraUMP. Unlike Ara-C, Ara-AC is a poor substrate for deamination by Cyd deaminase [12].

322 The mechanism of cytotoxicity of Ara-C is thought to be mediated through the ability of AraCTP to inhibit DNA polymerase and also as a result of incorporation of the fraudulent nucleotide into DNA [16]. Ara-CTP must compete with the natural nucleotide dCTP for interaction with DNA polymerase. Similarly, Ara-ACTP has been shown to be incorporated into DNA in a dose-dependent manner [9, 11, 12]. Enzymatic digestion of the DNA indicated that Ara-AC is incorporated primarily in the internucleotide linkage as the original compound [12]. In contrast, 5-Aza-Cyd is phosphorylated to 5-Azacytidine monophosphate (5-AzaCMP) by the action of another pyrimidine salvage enzyme, Urd/Cyd kinase [17]. Upon further anabolism to the triphosphate level by nucleotide kinases, 5-Aza-CTP can function as a substrate of RNA polymerase, competing with the normal substrate, cytidine triphosphate (CTP). The incorporation of the fraudulent ribonucleotide into RNA can interfere with normal RNA processing and function [17]. The effect of Ara-AC on the synthesis of macromolecules has been studied by monitoring the incorporation of radiolabeled thymidine, Urd, and Lamino acids into DNA, RNA and protein, respectively [9, 12]. An 8 hour exposure to Ara-AC (1/,M) was shown to inhibit the incorporation of [~4-C]thymidine into DNA of Molt 4 cells to less than 20% of control, but did not appear to inhibit RNA or protein synthesis [12]. Vesely and Piskala reported that 3.5 hour exposure to 10/,M Ara-AC inhibited the incorporation of [14-C]-thymidine into the DNA of LI210 cells by 97%; in contrast, [~4-C]Urd incorporation into RNA was inhibited by only 26%, and [~4-C]-L-leucine incorporation into protein was unaffected [9]. Dalai et al. conducted flow cytometry studies in L1210 cells which indicated that Ara-AC slows cell cycle progression through the S phase, suggesting interference with DNA synthesis [5]. In addition to studies examining the effect of Ara-AC on DNA synthesis, its effect on methylation of DNA has also been investigated [11]. 5-AzaCyd has been reported to cause hypomethylation of the cytosine bases in newly replicated strands of D N A , and this has been associated with activation

of quiescent genes and alteration of the metastatic and tumorigenic phenotypes of treated cells [18, 19]. Glazer and Knode reported that whereas 5-Aza-Cyd produced a rapid and concentrationdependent inhibition of DNA methylation without causing a major suppression of DNA synthesis, Ara-AC and Ara-C did not preferentially reduce DNA methylation [I 1]. Instead, Ara-AC and AraC inhibited DNA methylation to an equal or lesser extent than DNA synthesis. These investigators also reported that a 24 hour exposure to 1/zM Ara-AC did not result in DNA strand breakage of either parental or newly synthesized DNA [11]. Cell lines that are resistant to Ara-C by virtue of a deletion of dCyd kinase are cross-resistant to AraAC, but not against 5-Aza-Cyd [5, 7, 13]. Ahluwalia et al. have studied the mechanisms of resistance of murine lymphoblasts and the colon 38 carcinoma [13]. P388 lymphoblasts with acquired resistance to Ara-AC were shown to have a diminished capacity to accumulate nucleotides from radiolabeled Ara-AC and dCyd. Since the initial velocity of facilitated diffusion of Ara-AC showed only minor differences between the sensitive parental strain and the resistant P388 line, it appeared that the failure to accumulate Ara-AC resulted from a failure to phosphorylate the nucleoside rather than from a transport defect. Enzymatic studies indicated that the specific activity of dCyd kinase in the P388 resistant strain was only 5 %0of that in the sensitive strain. The enzymatic profile of the naturally resistant murine colon 38 adenocarcinoma displayed a lower level of activity of the purine and pyrimidine salvage enzymes, including dCyd kinase, compared to the P388/S line. In addition, the specific activity of the degradative enzyme nucleoside triphosphatase was increased 4-fold in the colon 38 line as compared to the P388 sensitive line. In contrast, an HL-60 cell line with acquired resistance to 5-Aza-Cyd secondary to a deficiency of uridine/ cytidine kinase retained sensitivity to Ara-C and Ara-AC [20].

Preclinical toxicology Toxicology studies in mice were conducted at the

323 Table 5. Toxicity of Ara-AC in mice. The data represent a compilation of toxicity pilot studies of Ara-AC in CD2F~ mice performed at the Midwest Research Institute (iv d x 1 single and hourly dose, and iv d x 5 studies) and Hazleton Laboratories America (ip q 3 hr x 8) under the direction of the Toxicology Branch, DTP, NCI Dose ( m g / m 2)

Route/ schedule

0 2700 4800 7200 9000

ivqd ivqd iv q d iv q d iv q d

0 3600 3600 3600 3600 3600

ivqhr iv q hr iv q hr iv q hr iv q hr iv q hr

0 3OO 600 900 1050 1200 1350 1500

lp q lp q lp q lp q ~p q xp q lp q lp q

3 3 3 3 3 3 3 3

hr hr hr hr hr hr hr hr

x x x • x x x x

0 1350 2700 3600 4800

ivqd iv q d iv q d iv q d iv q d

x x x x x

5 5 5 5 5

x x x x x x x x x x x

1 1 1 1 1 1 1 2 3 4 5

Vehicle

Total dose ( m g / m 2)

# Animals

# Moribund or dead

10% 2% 2~ 10% 10~

0 2700 4800 7200 9000

6 2 10 4 4

0 0 0 0 4

0 3600 7200 10,800 14,400 18,000

20 10 14 13 19 20

0 0 0 0 0 0

10070 DMSO 10% DMSO 10070 DMSO 10% DMSO 10% DMSO 10% DMSO 10070 DMSO 10% DMSO

0 2400 4800 7200 8400 9600 10,800 12,000

12 12 12 12 12 12 12 12

2 (17070) 0 1 (8%) 3 (25070) 2 (17070) 5 (42~

10% 2~ 2~ 2% 2%

0 6750 13,500 18,000 24,000

6 4 10 10 9

0 0 0 1 (10070) 3 (33%)

2% 2% 2% 2% 2~ 2% 8 8 8 8 8 8 8 8

DMA DMA DMA DMA DMA DMA DMA DMA DMA DMA DMA

DMA DMA DMA DMA DMA

Midwest Research Institute and Hazleton Laboratories America, Inc. under the direction of the Toxicology Branch, Developmental Therapeutics Program, National Cancer Institute (Table 5) [21, 22]. A single iv bolus dose of < 4800 mg/m-' administered in 2% dimethylacetamide (DMA) did not produce significant signs of toxicity. A larger dose of Ara-AC could be administered if the concentration of DMA was increased. No lethality was observed in a dose range of 2700-7200 m g / m ~of AraAC administered in 10% DMA. Higher concentrations of the vehicle were associated with toxicity. Single dose studies using 10% DMSO as a vehicle indicated that an Ara-AC dose of 7200 m g / m 2 also was non-lethal. In addition, five sequential injections of 3600 mg/m-' of Ara-AC in 2% DMA given at hourly intervals (cumulative dose 18,000 mg/m-') was associated with minimal toxicity. The LDlo on

5 (42070) 8 (75070)

a five daily dose schedule was 3600 m g / m ~ in 2% D M A . A single day multiple dose study utilizing doses o f Ara-AC of 3 0 0 - 1 5 0 0 m g / m 2 in 10%0 D M S O ip every 3 hours times 8 doses was also conducted. Doses _ 900 m g / m 2 (total dose 7200 m g / m 2) were consistently accompanied by lethality. The toxicity o f Ara-AC administered as a sc CI was also explored. Mice given 22.5 mg/mZ/hr for 4 8 - 1 4 4 hours became moribund and were necropsied or died by day 8 (Table 6) [22]. Zaharko et aL reported that a sc infusion rate o f 20 m g / m 2 / h r for 24 hours in tumor-bearing mice was tolerated [10]. In contrast, prolonging the duration o f treatment to 48 and 72 hours was associated with 7% and 20% weight loss, respectively, and all animals receiving 20 m g / m V h r for _> 96 hours died o f toxicity. Studies conducted at the Southern Research Institute evaluating a 72 hour sc CI schedule in non-

324 Table 6. Toxicity of Ara-AC administered as a continuous subcutaneous infusion in mice. The data represent a summary of toxicity studies of Ara-AC in nontumored mice administered as a sc CI.

Dose (mg/m-')

Duration of infusion

Total dose (mg/m")

# Animals

# Moribund or dead

0a 22.5 mg/m2/hr 22.5 mg/m2/hr 22.5 mg/m-'/hr 22.5 mg/m2/hr

144 hr 48 hr 72 hr 96 hr 144 hr

0 1080 1620 2160 3240

12 3 12 12 12

0 3 (100%) 12 (100070) 12 (10007o) 12 (200%)

11.4 mg/m2/hrb 15.0 mg/m2/hr 20.0 mg/m-'/hr

72 hr 72 hr 72 hr

822 1080 1440

6 6 6

0 2 (33%) 3 (50%)

In the first studya, conductedat Hazleton Laboratories America, Inc., CD2F~ mice received a sc infusion of Ara-ACat a concentration of 250 mg/ml in 70~ DMSO in 0.9% saline. The animals in the vehicle control group appeared normal with no signs of toxicity. The second set of datab represents a nontumored control group of NCR-nu mice which received a 72 hour sc infusion of 3 dose levels of Ara-AC as part of the evaluation of schedule dependencyin a solid tumor model conducted at the Southern Research Institute (data of the antitumor activity is presented in Table 3). tumored mice indicated that dose rates _> 15 m g / m2/hr were associated with lethality, but all mice treated with 11 m g / m 2 / h r survived (Table 6). Clinical signs occurring during the treatment phase included hypoactivity, cloudy ocular discharge, ataxia, rough hair coat, bradypnea, hypothermia, tremors, gasping and hunched posture. Drug related hematologic changes in the mice included decreased WBC (affecting lymphocytes and granulocytes), decreased platelet count, and decreased reticulocyte count occurring at day 4. Reversible hematologic changes were seen at all dose levels. Histologic evidence o f drug-related toxicity bone marrow depletion, lymphoid organ depletion (thymus, lymph nodes, and splenic red pulp), and necrosis of the gastrointestinal epithelium with subacute inflammation o f the lamina propia and dilatation of the intestinal crypts. Maturation arrest and necrosis of the seminiferous tubules was observed in the testis. The toxicity of Ara-AC given iv as a single bolus and as a CI was studied in beagle dogs [23, 24]. Dogs tolerated a single dose of 4000 m g / m ' in 2~ D M A without toxicity [23]. Animals receiving 10,000 m g / m 2 had soft stool and moderate emesis the day following treatment, but myelosuppression was minimal. One of two animals receiving a single dose of 20,000 m g / m ~ died with clinicopathologic features characteristic of DMA-related toxicity.

The CI studies indicated that very low dose rates (l. 1 mg/m2/hr) could be tolerated for 5 to 7 days, but mild to moderate myelosuppression was noted (Table 7) [24]. As the dose rate was increased, however, the duration of therapy that could be tolerated decreased. For example, 108 mg/m-'/hr coud be tolerated for up to 24 hours, but longer durations were lethal. Clinical toxicity in the dogs included vomiting, decreased activity, soft stool, low food intake, pale mucous membranes, dehydration, ptyalism, muscle rigidity, weakness in the legs, tremors, ataxia, convulsions, irregular heart beat and phonation. At the lethal dose rate/exposure times, extreme, lethargy, bloody diarrhea and bloody emesis were seen. Drug-related changes in the blood counts included depression of the platelet, white blood cell, absolute neutrophil, lymphocyte, and the reticulocyte counts. No drugrelated changes were observed in the chemistry profile. Microscopic changes included bone marrow hypoplasia, lymphoid depletion in spleen, thymus, and lymph nodes, dilatation of the small intestinal crypts, and necrosis o f the gastrointestinal epithelium. In summary, the administration of a single iv dose of Ara-AC in mice was limited by volume and vehicle constraints. Single bolus doses _< 7200 m g / m 2 did not produce any lethality. The minimal lethal dose on the every 3 hours times 8 doses sched-

325

Table 7. Toxicity of Ara-AC administered as a continuous intravenous infusion in beagle dogs. This study was conducted at Hazleton Laboratories America, Inc. Ara-AC was mixed with DMSO and diluted to 2~ with 0.9~ saline. One infusion of Ara-AC was administered per dog. Dose (mg/kg/hr)

Dose (mg/m-'/hr)

Duration of CI (hours)

Total dose ( m g / m 2)

Number of animals

Day of death

vehicle control

0 0

48 168

0 0

1 2

---

0.054 0.054

1.1 1.1

120 168

130 182

2 3

---

0.21

4.2

120

504

2

10, 11

0.54 0.54 0.54

10.8 10.8 10.8

72 120 168

778 1,296 1,814

2 2 1

7, 9 11, 12 10

12 24 48 118

1,296 2,592 5,184 12,744

1 1 1 1

--8 6

3 6 12 24

1,000 2,000 4,000 8,000

1 1 1 1

---8

5.4 5.4 5.4 5.4

108 108 108 108

16.67 16.67 16.67 16.67

333.4 333.4 333.4 333.4

Table 8. Preclinical pharmacokinetics. The values obtained in the different pharmacokinetic studies have been converted to common units for ease of comparability. Values initially presented in terms of kg were converted to surface area by use of the following conversion factors: 3 kg/m-" for mice; 20 k g / m 2 for dogs; and 20 k g / m 2 for the rhesus monkeys (the monkeys weighed approximately 8-12 kg, and therefore a surface area to weight ratio (km) comparable to that for beagle dogs was used rather than the km = 12 which is usually employed for primates with a body weight of 3 kg). Species

Dose ( m g / m 2)

Route

Mean peak Cp or CPs~ #g/ml 0zM)

Volume of distribution ( L / m -~)

mouse 22

600 m g / m 2 600 m g / m 2 22.5 m g / m 2 /

iv ip sc

254 (1040) 78 (319) 1.56 (6.4)

m

mousO ~

600 m g / m ~ 20.4 m g / m V hr CI x 48 hr

iv sc

180 (737) 2.1 (8.6)

dog ~

8000 mg/m-" 8000 mg/m-"

iv po

690 (2829) 76 (31l)

108 m g / m " / h r • 48 hr

iv

10 (41)

iv iv iv

29 (119) 34 (139) 29 (119)

iv

130 (533)

28 + 16

iv

338 (1393)

16.4 + 0.53

iv

7.5 (31)

dog 24

333 m g / m V hr • 3 hr • 6 hr • 12 hr primate 2~

primate 24

4000 over 4000 over

mg/m 2 15 min mg/m 2 60 min

160 mg/m-'/hr • 12hr

Total body Terminal clearance half-life ( m l / m i n / m 2) (min)

AUC (/~g/ml/hr)

Comments

87

141 106

ip/iv = 75%

-

m

-

23 min -

-

44.4

76 min

--

--

178

69 min

763 171

po/iv = 22%

136 min

--

--

71 min 88 min 106 min

86.5 209 399

--

744 ___ 246

33.6 min

103.5

602 _+ 83

39 rain

113

CSF: plasma = 18070 CSF: plasma = 1307o

10.6 -

-

--

-

-

-

-

-

-

n

326 ule was 600 m g / m 2 / 3 h r • (MELDs). The MELDto on an iv bolus d • schedule was 3600 mg/m2/d. An LD10 was not defined for the CI schedules. A single iv dose of 10 g m / m 2 was tolerated by dogs, yet a total dose of 0.5 g m / m 2 was lethal if given for a 5 day iv CI. Therefore, the drugrelated toxicities did not appear to be related to peak plasma concentrations. Rather, toxicity was dependent on both dose and duration of drug exposure, with time as critical an element as dose.

studies in rhesus monkeys [25]. The C1TB appeared to be approximately 4 - 1 1 times faster than that reported in dogs and mice, respectively. The terminal tl/2 was 3 0 - 4 0 minutes. The drug had significant penetration into the cerebrospinal fluid. A Cps s in the target range of 1 - 1 0 /~g/ml was achieved with an iv CI o f 160 mg/m2/hr. The animals which received this dose rate for 12 hours did not experience any clinical toxicity nor evidence of myelosuppression (personal communication, Dr. Heideman, April, 1987).

Preclinical pharmacology The pharmacokinetic behavior of Ara-AC has been evaluated in mice, dogs, and primates [22, 24, 25] (Table 8). CD2F~ mice were given a 600 m g / m 2 dose of Ara-AC either iv or ip. The peak plasma concentration was approximately 250 /zg/ml; the alpha half-life (tl/2) was 0.5 min, and the beta tl/2 was 22.8 min [22]. The total body clearance (C1TB) was 28.9 m l / m i n / k g (or 87 ml/min/m2). The ip bioavailability was approximately 75 ~ of the iv dose. In contrast, Zaharko et al. reported a longer terminal tl/2 (76 min) [10]. A steady-state plasma concentration (CPss) in the 1 - 1 0 #g/ml range could be achieved with a sc CI of 20.4-22.5 mg/m2/hr [10]. Beagle dogs were given a 8000 m g / m 2 iv bolus dose of Ara-AC. The plasma concentration at time zero was calculated to be 925 _+ 490/~g/ml, and the initial measurable plasma concentration was 690 + 129/~g/ml. The terminal tl/2 was 69 min [24]. The CI-r~ was faster in dogs than in mice. The peak Cp of Ara-AC following po administration of 8000 m g / m 2 was 76 _+_ 27/~g/ml, and occurred one hour following dosing. The oral bioavailability (AUC i v / A U C po) was 22~ The pharmacokinetic properties o f Ara-AC administered by CI were also evaluated [23]. Based upon the preclinical data suggesting that an extracellular concentration o f 10 /~g/ml for _> 72 hours was optimally effective in reducing the viability of L1210 cells, a target plasma concentration of 1 - 1 0 ~g/ml was desired. The actual CPss measured during an iv CI of 108 mg/m2/hr was 8 - 1 0 t~g/ml. The actual CPss measured in dogs receiving a larger infusion rate (330 mg/m~/hr) was 30 #g/ml. Heideman et al. conducted pharmacokinetic

Plans for phase I clinical trials The preclinical in vitro and in vivo murine leukemia studies and human solid tumor clonogenic assays suggest that the more frequent the administration of Ara-AC, the better the cytotoxic effect. Therefore, initial clinical trials sponsored by the Division of Cancer Treatment, NCI will seek to define the maximally tolerated dose of Ara-AC on a 72 hour continuous infusion schedule. The toxicology studies demonstrate that toxicity depends on dose rate as well as total dose. The murine and canine toxicology studies o f Ara-AC administered as a CI do not permit a strict definition of minimum lethal dose. Instead, the recommended starting dose for CI schedules has been based on the apparent threshold in the infusion rate/time matrix. An additional factor which must be considered in the clinical treatment strategy is the potential interspecies differences in intracellular accumulation of the cytotoxic metabolites of Ara-AC. Information is available which indicate that the activity of dCyd kinase, the rate-limiting enzyme in the anabolism o f Ara-AC, differs between humans and the small and large animals used in preclinical toxicology studies. For example, H o compared the dCyd kinase activity in various human and mouse tissues. The activity in human intestinal mucosa and bone marrow exceeded that o f the comparable murine tissues by a factor of 2.5 and 20-fold, respectively [26]. Leiby and Grever measured the dCyd kinase activity in dog and human bone marrow, and reported that the activity was 5-fold greater in the human blood ceils [27]. Clinical experience with fludarabine monophosphate, which is also

327 anabolized by dCyd kinase, indicated that humans were much more sensitive to the toxicity of fludarabine than either mice or dogs [28, 29]. Therefore, the recommended starting dose for clinical trials with Ara-AC as a CI is based Upon tolerable doses for 72 hour CI in mice (10 mg/m2/hr) and dogs (1 mg/m2/hr), with an additional reduction to account for possible increased intracellular activation of Ara-AC by dCyd kinase: 0.2 m g / m g V h r for 72 hr. Preclinical studies indicated that Ara-AC was active against solid tumor xenografts when given on an intermittent schedule. An intermittent short infusion schedule of Ara-AC, if effective, would have the advantage of drug stability in Lactated Ringer's Injection, thus circumventing the need for a DMSO vehicle. In addition, drug administration in the outpatient setting would be possible, avoiding the need for an indwelling intravenous catheter and hospitalization. Therefore, a daily times five schedule will also be evaluated, with a starting dose of 72 mg/m2/d x 5, one-fiftieth of the mouse equivalent LD10. This latter schedule is expected to permit pharmacokinetic data to be collected early in the phase I clinical experience with Ara-AC. Additional schedules may be evaluated pending further preclinical data and according to the clinical experience with the aforementioned schedules. Phase I studies in pediatric leukemia and solid tumor patients are also planned. The toxicology studies suggest that the dose-limiting toxicity in human trials will be myelosuppression a n d / o r gastrointestinal toxicity. Human tissues are known to possess a relatively high activity of cytidine deaminase compared to that of dCyd kinase [16, 26, 30]. Both Ara-C and Ara-AC require dCyd kinase for activation, thus deletion of this enzyme would produce resistance to both agents. Unlike Ara-C, Ara-AC is a poor substrate for deactivation by Cyd deaminase [12]. It is unclear, however, to what extent resistance in humans to conventional or high dose Ara-AC is related to this deamination [16, 31, 32]. A key question will be whether Ara-AC offers an advantage over Ara-C, and whether it will have activity in leukemic patients with clinical resistance to Ara-C. Phase I evaluation in adult leukemia patients will begin

when adequate information in available from studies in patients with solid tumors. Phase II evaluation in leukemia will be a high priority. Because of its antitumor activity against murine and human solid tumors, phase II evaluation of AraAC should include broad testing in solid tumors. It will be of interest to learn whether the superior antineoplastic activity of Ara-AC in preclinical models compared to either Ara-C or 5-Aza-Cyd will be realized in the treatment of human solid tumors.

References 1. Beisler JA, Abbasi MM, Driscoll JS: The synthesis and antitumor activity of arabinosyl-5-azacytosine. Biochem Pharmacol 26:2469-2472, 1977 2. Beisler JA, Abbasi MM, Driscoll JS: Synthesis and antitumor activity of 5-azacytosine arabinoside. J Med Chem 22:1230-1234, 1979 3. Mojaverian P, Repta A J: Development of an intravenous formulation for the unstable investigational cytotoxic nucleosides 5-azacytosine arabinoside (NSC 281272) and 5-azacytidine (NSC 102816). J Pharm Pharmacol, 36:728733, 1984 4. Notari RE, DeYoung J I : Kinetics and mechanism of degradation of the anti-leukemic agent 5-azacytidine in aqueous solutions. J Pharm Sci 64:1148-1156, 1975 5. Dalal M, Plowman J, Breitman TR, Schuller M, del Camp AA, Vistica DT, Driscoll JS, Cooney DA, Johns DG: Arabinofuranosyl-5-azacytosine: antitumor and cytotoxic properties. Cancer Res 46:831-838, 1986 6. Driscoll JS, Johns DG, Plowman J: Comparison of the activity of arabinosyl-5-azacytosine, arabinosyl cytosine, and 5-azacytidine against intracerebrally implanted L1210 leukemia. Invest New Drugs 3:331-334, 1985 7. National Cancer Institute, Division of Cancer Treatment, Clinical Brochure, Fazarabine (Ara-AC), NSC 281272, January, 1987 8. Wallace RE, Lindh D, Durr FE: Arabinosyl-5-azacytosine (Ara-AC, fazarabine, NSC 281272) activity against human tumor xenografts. Proc Am Assoc Cancer Res (Abstr) 28:307, 1987 9. Vesely J, Piskala A: Mechanism of action of 1-beta-D-arabinofuranosyl-5-azacytosine and its effects in El210 mouse leukemia cells. Neoplasma 33:3-10, 1986 10. Zaharko DS, Covey JM: Arabinosyl-5-azacytosine plasma kinetics and therapeutic response (L1210) in vitro and in vivo in mice. Invest New Drugs 3:323-329, 1985 11. Glazer RI, Knode MC: 1-beta-D-Arabinosyl-5-azacytosine: cytocidal activity and effects on the synthesis and methylation of DNA in human colon carcinoma cells. Molec Pharmacol 26:381-387, 1984

328 12. Townsend A, Leclerc JM, Dutschman G, Cooney D, and Cheng Y-C: Metabolism of 1-beta-D-arabinofuranosyl-5azacytosine and incorporation into DNA of human T-lymphoblastic cells (Moh-4). Cancer Res 45:3522-3528, 1985 13. Ahluwalia GS, Cohen MB, Kang G-J, Arnold ST, McMahon JB, Dalai M, Wilson YA, Cooney DA, Balzarini J, Johns DG: Arabinosyl-5-azacytosine: mechanisms of native and acquired resistance. Cancer Res 46:4479-4485, 1986 14. Bodner A J, Ting RC, Gallo RC: Induction of differentiation of human promyelocytic leukemia cells (HL-60) by nucleosidcs and methotrexate. J Natl Cancer Inst 67:10251030, 1981 15. Dion RL, Jayaram HN, Cooney DA, Ahluwalia GS, Johns DG: Studies on the mode of action of 1-beta-D-arabinofuranosyl-5-azacytosine (NSC 281272). Proc Am Assoc Cancer Res (Abstr) 24:297, 1983 16. Chabner BA: Cytosine arabinoside, in Chabner BA (ed), Pharmacologic Principles of Cancer Treatment, WB Saunders Co., Philadelphia, 1982, pp 387-401 17. Chabner BA: Pyrimidine Antagonists, in Chabner BA (ed), Pharmacologic Principles of Cancer Treatment, WB Saunders Co, Philadelphia, 1982, pp 183-212 18. Razin A, Riggs AD: DNA methylation and gene function. Science 210:604-610, 1980 19. Trainer DL, Kline T, Mallon F, Grieg R, Poste G: Effect of 5-azacytidine on DNA methylation and the malignant properties of B16 melanoma cells. Cancer Res 45:61246130, 1985 20. Grant S, Bhalla K, Gleyzer M: Effect of uridine on response of 5-azacytidine-resistant human leukemic cells to inhibitors o f de novo pyrimidine synthesis. Cancer Res 44: 5505-5510, 1984 21. El-hawari AM, Unwin SE, Stoltz ML, Bronfman JE: Preclinical range-finding toxicity studies of 5-azacytosine (NSC 281272) in C D2F __. mice and Fischer 344 rats. Final Report, Midwest Research Institute, MRI Project No 8016-E, Prime Contract No NOI-CM-17365, 1984 22. Weltman RH, Boysen BG, Glaza SM, Nettum JA: Preclinical pharmacokinetics and toxicity studies in CD2F + mice with 5-azacytosine arabinoside (NSC-281272), vol I. Final Report, Hazleton Laboratories America, Inc, Report No HLA-6133-1-281272, Prime Contract No NO1-CM-17365, 1986

23. El-hawari AM, Stedham MA, Unwin SE, Bronfman JE,Koontz E J, Stoltz ML: Preclinical single defined dose toxicity study of 5-azacytosine arabinoside (NSC 281272) in beagle dogs. Final Report, Midwest Research Institute, MRI Project No 8016-E, Prime Contract No NO1-CM-17365, 1984 24. Weltman RH, Boysen, Dickie BC, Nettum JA: Preclinical Pharmacokinetics and toxicity studies in beagle dogs with 5-azacytosine arabinoside (NSC-281272), vol II. Final Report, Hazleton Laboratories, Inc, Report No HLA 6133-1-281272, Prime Contract No NO1-CM-17365, 1986 25. Heideman R, Balis F, Lester C, Poplack D: Preclinical pharmacology of arabinosyl-5-azacytosine. Proc Am Assoc Cancer Res (Abstr) 28:437, 1987 26. Ho DHW: Distribution of kinase and deaminase of l-betaD-arabinofuranosylcytosine in tissues of man and mouse. Cancer Res 33:2816-2820, 1973 27. Leiby JM, Grever MR: Deoxycytidine kinase activity in normal and malignant blood cells. Proc Am Soc Clin Oncol (Abstr) 5:52, 1986 28. National Cancer Institute, Division of Cancer Treatment. Clinical Brochure, fludarabine phosphate (FAMP), NSC 312887, 1982 29. National Cancer Institute, Division of Cancer Treatment. Annual Report on fludarabine phosphate (FAMP), NSC 312887, 1986 30. Coleman CN, Stoller RG, Drake JC, Chabner BA: Deoxycytidine kinase: properties of the enzyme from human leukemic granulocytes. Blood 46:791-803, 1975 31. Tattersall MHN, Ganehaguru K, Hoffbrand AV: Mechanisms of resistance of human acute leukemia cells to cytosine arabinoside. Br J Hemat 27:39-46, 1974 32. Chou T-C, Arlin Z, Clarkson BD, Philips FS: Metabolism of 1-beta-D-arabinofuranosykytosine in human leukemic cells. Cancer Res 37:3561-3570, 1977

Address f o r offprints: Jean L. Grem, Investigational Drug Branch, Cancer Therapy Evaluation Program, Division of Cancer Treatment, National Cancer Institute, Landow Bldg., Room 4C09, Bethesda, MD 20892, USA